0:01
[music] I see a large IPO on the horizon.
Search this video
0:11
>> You're surrounded by journalists. Hold your position. Overnight success.
0:30
The place 3000, [music] >> right? There's misinformation.
1:01
>> Quarterly clearing order inbound. >> Let's just roll.
1:11
We are surrounded by journals. Hold your position. Come on. Get up. Trust the experts here. We are experts found.
1:44
>> I see multiple journalists on the horizon. Standby. >> UAV online. >> Blaze. >> Double blaze. Triple blaze. Double kill.
2:18
>> [music] >> Finally, talk is team deathmatch. We are experts. Triple blades. That's just wrong. Right.
3:09
Mark clearing order inbound.
3:23
We are surrounded by journalists. Hold your position. >> Strike one. >> Strike two.
3:41
Activate golden retriever mode.
4:00
>> Marky clearing order inbound. >> Five put.
4:20
I see multiple journalists on the horizon. Stand by.
4:36
>> [clears throat] >> You're watching TVPN.
4:37
Today is Tuesday, February 24th, 2026.
4:40
We are live from the TVPN Ultra Dome, the temple of technology.
4:46
The fortress of finance, the capital of capital.
4:50
>> We're running down a dream today.
4:50
We are surviving the Catrine apocalypse.
4:56
Live to fight another day.
4:56
A lot of chaos in the markets.
4:57
a lot of reflection about the story behind the story, what happened.
5:02
We had a lot of fun debating the Catrini report.
5:03
A lot of good stuff in there.
5:06
Some other kind of crazy stuff that uh sort of got everyone twisted in a knot.
5:11
Uh but uh >> it didn't stop the markets.
5:16
>> It did become the current thing and I think a lot of people were talking about it.
5:18
I mean, my feed was covered in Catrini stuff, but uh today is a new day and there is a ton of new tech news.
5:26
First, let me tell you about ramp. com. Time is money. Say both. Easy to use.
5:29
Corporate cards, bill pay, accounting, and a whole lot more. The goats.
5:31
And then second, I want to pull up the linear lineup because boy, do we have a show for you today, folks.
5:38
We got the Collison brothers joining together at 11:40.
5:42
Then we going over to Bill Gurley, the height moger himself. >> He's 6'9. He moss me. >> Oh, he mos. >> It's over for me.
5:50
That's why we said you can't come to the studio.
5:53
You can't be seen next to John Kugan in person. You're staying remote.
5:59
>> The right pair of cowboy boots.
6:00
>> Yeah, I might have >> day. You might have it.
6:02
>> The right pair of lifts as well inside those cowboy boots.
6:04
Then we got Ivan from Notion and a whole bunch more funding announcements during the lightning round.
6:10
Uh Run, Reiner, uh Devanch and a ton of other folks are joining. It's it's a crazy.
6:16
>> James from profound James. Yeah, James Unicorn. >> Uh very very fun.
6:20
Well, uh, linear of course is the system for modern software development.
6:24
70% of enterprise workspaces on linear are using agents.
6:29
Um, so, uh, the story behind the Catrini story, I had, uh, some takeaways.
6:33
Uh, my big update mentally was, uh, just that, you know, uh, we are the cellside research now.
6:41
basically like new instack but X and Substack like independent researchers and and analysts are really moving the markets.
6:52
I feel like Ben Thompson has been a source of of alpha for the market for a long time.
6:58
He's been a source of investment thesis, but he doesn't put a buy or sell rating on things and >> much more long term. >> Long term. Exactly.
7:07
Here's how strategies are converging.
7:09
Here's how the market is evolving. Make your own decisions. >> Yes.
7:14
And then and then I see like semi analysis is thinking more in like a couple years out and it's still like there are they they get held accountable for oh you said Microsoft was going to do this and they did that blah blah blah like and and they have a different model that they actually sell to hedge funds.
7:32
Um and so they're very much in the research business.
7:33
But what's interesting about semi-analysis and a lot of these other independent analysis firms is that they're not sitting inside banks like we we are very much used to sellside research being done by Morgan Stanley or Bank of America, Goldman Sachs.
7:48
You get these equity research reports that your friends send you the PDFs for because you can't afford them.
7:53
[laughter] No, seriously, if you're working in an industry, get the sellside research report on your industry as as fast as possible.
8:00
It's very very informative.
8:00
Uh there's always good data in there.
8:02
Um, but yeah, my my my big update was like, wow, okay, this is like a viral post that completely broke containment.
8:09
There's people making Tik Toks about it now.
8:11
And also, it's on the cover of the Wall Street Journal. >> Cells. >> Yeah, >> doom cells. >> Doom cells. Yeah, >> over cells.
8:18
>> That's the Yeah, that's one of the narratives.
8:19
And there was this funny funny thing about like, oh, well, it's just one it's just one scenario. It's just one scenario. Why? It's low probability.
8:28
And then >> let's pull up Eric.
8:29
Eric was like was like, "Yeah, it's just one scenario, but you only gave us one scenario and you spent 100 hours on that scenario."
8:38
[laughter] So like like what do you expect people to take away from it except like this is the one scenario that you think is most most worth considering?
8:44
But of course, you know, it is possible that software is cooked, everything's cooked, and if there's a 5% chance that everything's cooked, yeah, the market should probably sell off by a couple percent.
8:55
And you know, the market didn't even really sell off a couple percent.
8:57
Like a couple names went down a few percent.
8:59
Some of them already popped back up.
9:00
Uh markets, I think, doing pretty well today.
9:03
Uh yeah, green on the Dow, green on the NASDAQ, and a lot of green on that ticker down there, which is of course provided by public.
9:11
com investing for those who take it seriously.
9:14
They got stocks, options, bonds, crypto, treasuries, and more with great customer service.
9:20
And I'm also going to tell you about Octa.
9:22
Octa helps you assign every AI agent a trusted identity.
9:24
So you get the power of AI without the risk.
9:27
Secure every agent, secure any.
9:30
>> Let's head over to Derek Thompson, >> the Thompsonator.
9:32
He says, "I really want people to see the story above the story here, which is that whether you're reading Catrini or listening to Jamie Diamond at a cocktail party.
9:39
The conversation about AI is a marketplace of competing science fiction narratives.
9:45
That's not to say I think the technology is a parlor trick.
9:46
You know, we covered this a couple weeks ago.
9:48
He's feeling he's feeling the uh the AGI.
9:51
That might be a little bit putting it too aggressively, but certainly he sees uh the potential impact, but Derek says, but rather that the level of uncertainty is so high and the quality in supply of real world real-time information about AI's macroeconomic effects so poulry that very serious that very serious conversations about AI are often more literary than genuinely analytical.
10:15
And I think that observation sets up another important point.
10:19
I feel lucky to be able to have conversations about the frontier of AI with executives and builders at Frontier Labs, economists, investors, and other AI folks at off thereord dinners where important truths can theoretically be shared without risk.
10:30
I can't emphasize enough that nobody knows anything >> except for us >> is about as close to the reality here as three words are going to get you.
10:37
Uh nobody is >> uh nobody what's nobody knows what's going to happen this year or next year or the year after that.
10:44
There is no secret cigar filled room of people >> except for us >> except except the back room.
10:49
I think we do have some cigars back there.
10:50
You have unique access to some authentic postcard from the future.
10:56
When you drill down underneath the bluster, the boosterism, the fear, the anxiety, what's there at the bottom is genuine uncertainty, a vacuum into which storytelling is flooding.
11:06
The frontier labs don't really know what they're building exactly, >> but we do.
11:10
>> And economists don't really know how to model the thing they claim they're building, but we do. >> Yeah.
11:14
Uh, I wish more people talked about and thought about this subject through that sort of lens.
11:18
We're trying to model the economywide effects of a technology whose properties the Frontier Labs can't even really describe yet.
11:26
Whatever you think of AI today, be prepared to change your mind soon.
11:28
Yeah, this was uh something with um uh Allup yesterday >> is I didn't uh when I asked him why why do you think that so many of the internet predictions were deeply wrong?
11:46
His answer was it's just a continuum.
11:46
AI is just a continuum and so like give it more time basically. >> Yeah.
11:55
The rebuttal that I heard from him when you said that he was like well no like look at all those predictions did come true and it was like yeah but over 20 years which is like wildly different than two years because the the fear article is called the 2028. Exactly.
12:13
And so so if you tell me >> also so many institutions just adapt it. >> Yeah.
12:18
Like if you go to somebody and you say hey in in in 20 years your job is going to be radically different. They're like I hope so.
12:24
like I'm going to be super bored doing the same thing for the next 20 years.
12:28
In two years there's going to be no industry that you're currently in.
12:32
Everyone's going to be like, "Oh, okay." Like that's crazy.
12:35
It's wildly different to be like, you have 20 years to adjust what you do.
12:37
Uh like, you know, if you're like, you know, you're if you're in Hollywood and you're like, "Okay, I got to learn digital film making.
12:44
I got to learn how to integrate CGI.
12:46
I got to learn AI as a tool."
12:48
That's way different than just like next year we will be oneshotting holiday at Hollywood films and we will you will have no employment prospect whatsoever.
12:56
Not even as a prompter because the labs will be prompting them themselves for for for AI videos.
13:00
And maybe that's possible.
13:03
But I have a feeling that it's just like it's not a year away. It's not two years away.
13:07
It's a little bit farther.
13:09
Still on the 10-year camp.
13:11
Still on the Kershw timelines.
13:13
But uh interestingly I'm I'm like I'm impatient about it.
13:16
Like I want it to go faster.
13:19
I want I want the acceleration.
13:21
I want I want the progress.
13:23
I think the progress is good.
13:25
Um so I'm not like a doomer pessimist.
13:28
I'm just like trying to grapple with the fact that I've se I I had to wait four years in between GPT3 and and models being good enough to not hallucinate.
13:39
I had to wait another four years between like the early Dolly experiments and like the nano bananas.
13:45
Like it's it it has felt like like something happens and I'm like oh wow like okay like AI can generate images but it's sort of sloppy.
13:55
And then I wait like four years and it's like, okay, it's like a lot less sloppy, but it's like still not like dialed.
14:01
Like it's it went from 90% to 99%.
14:01
And I'm waiting for it to get to 99. 999999999999999%.
14:11
That's where I want it to go.
14:11
Anyway, is Door Dash cooked?
14:13
Let's go over to Dem Ben Thompson on stratey. He said, "Okay, fine.
14:20
While I'm here, the Door Dash example is just unbearably dumb."
14:21
He is a believer in the power of Door Dash to weather the AI storm.
14:28
I saw that the Door Dash uh >> CEO put out a a a like an SEC letter to uh uh to the investors. Did you see that?
14:40
>> Like full PDF filed with the SEC like this is this is sort of guidance but telling the investor base like here's what's not going to change.
14:47
like basically disregard the the catrini report, >> disregard sci-fi >> doom.
14:54
Uh he says, "Set aside for now the question of agents and aggregation."
14:59
That's a post that is definitely in my mental cue.
15:01
What is notable about the assertion is the total denial of any positive reason for Door Dash to exist or and to be so successful.
15:07
There's no awareness that Door Dash provides provided a massive consumer benefit restaurant food at home from scratch.
15:13
I liked Keith Reoy's take that Door Dash is the I'm hungry button on your phone and then there's a whole bunch of crazy things that you have to do to make that happen, make that button work.
15:22
I ordered Door Dash last night.
15:24
I felt like I did it in protest of the doom.
15:26
I was like, I'm still supporting I'm riding with Door Dash.
15:30
Uh [snorts] there's no awareness that Door Dash provided a massive consumer benefit from scratch, that Door Dash massively increased the addressable market for restaurants, or that Door Dash provided brand new jobs for millions of drivers.
15:42
Instead, the article just sort of takes it as a given that Door Dash exists and that it is a rent extractor preying on weak-willed humans in their habits.
15:48
Uh, this is the exact sort of view taken by some of the most frustrating anti- monopoly activists.
15:55
All large uh successful tech companies exist not because they created a market with virtuous cycles solving all kinds of thorny problems along the way, but rather because the government didn't regulate hard enough.
16:08
I was thinking about uh in antitrust regulation, you know how they'll stop two firms from joining because that will create a monopoly, but they don't really have a tool in the tool chest for stopping a company from just shutting down and stopping competing.
16:22
Like if Xbox goes away, as people are predicting, doom around Xbox, uh PlayStation gets a lot more powerful.
16:30
Obviously, it's like the only game on the block.
16:31
And so, should the FTC have a hammer to be like, "No, you gotta lock in Asha.
16:36
You gotta make more Xbox games.
16:38
You got to compete harder.
16:38
We want you to We want you to get GTA 6 out exclusively on Xbox faster to put the screws to Sony.
16:46
>> You don't have time to spend three months gaming. >> No. No. Lock in. Give us a new Halo.
16:50
Give us a new Modern Warfare. Give us a new Fable. I don't know.
16:53
What are the other great Xbox games throughout the years?
16:57
I was never that big of an Xbox gamer. >> Never owned an Xbox. >> Never owned an Xbox. Wow. Oh, I'm a gamer. No, you never say that. >> I was working.
17:05
It wasn't [snorts] >> I didn't have the what were what was an Xbox back then?
17:12
>> Uh it was like 300 400 bucks expensive.
17:15
>> Anyway, let me tell you about Gemini 3. 1 Pro. Gemini 3.
17:17
1 Pro is here with a more capable baseline.
17:20
It's like it's great for super complex tasks like visualizing difficult concepts, synthesizing data into a single view or bringing creative projects to life.
17:28
I was thinking about the uh like the SAS apocalypse uh in the context of the fact that >> today's launches >> uh no there are a lot of launches but um >> well well no the disconnect in the SAS apocalypse is that is that AI native SAS is getting funded at an insane rate. >> Yes.
17:47
While you have these massive sell-offs in the public markets, yeah, there is a little bit of there's private companies that are getting >> lots and lots and lots of funding that if they were if they were public would have traded down 20% over the last week or so. >> Yeah. >> Anyways, continue.
18:05
>> Anyways, continue. The the thing I was I was thinking about was um there's this whole idea that like you'll be able to build your own like CRM or your own ERP and vibe code it and open-source CRM exist like there's there's one called
18:22
sweet CRM there's ODO there's ERP next there's plane for uh for task management open project red mine like there are open-source alternatives to almost every piece of software there's There's an open- source Photoshop that people use on Linux and they've never really gotten adoption. I used an open source forum
18:42
I used an open source forum software for a while and very quickly I called the person that was maintaining it and was like I'll pay a thousand bucks to just like do this for me and then they it became a managed service very quickly and >> there's an open source capy bar simulator. >> Is that open source? No.
18:55
Um but but it's it's interesting because like like open source has always been this like pressure on SAS and it's always withstood that and like yeah maybe like if you can just prompt it and it feels like emailing your your SAS provider to reconfigure things like that is a real pressure but I think it's underrated that that open-source CRM have existed for decades and never really taken off because there's something else that's valuable there.
19:26
But Tyler has a report because he thinks everything's going to get slopped.
19:30
>> I think that's like mostly cope like the comp to open source stuff.
19:32
It's just annoying to maintain so no one ever does it and you're kind of just paying whoever to maintain it, right?
19:38
>> Um no no in open source like you don't need to pay someone to maintain it.
19:41
Use the openour source version of the open source thing.
19:45
It's like you're basically paying someone to maintain it. >> Yeah. Yeah. Host it all.
19:47
Make sure >> up times like models keep getting better and they'll just like do all for you.
19:53
>> They'll do all that for you. >> And it's like Yeah.
19:54
I think that's like very obvious.
19:58
>> Yeah, it it it is possible that you would just say like, okay, run in a loop and just go around and fix everything and if there's uptime or security patches, like patch them immediately.
20:05
So the open source software gets better.
20:09
>> There's two there's two narratives, right?
20:11
There's the >> okay, everyone will just vibe code everything in any department.
20:15
Yeah, >> you can just have an employee just make the software, tell the agent not to make mistakes >> and or tell the agent, hey, fix this thing.
20:26
So, that's that's one thing.
20:26
And I feel like that is a maybe a part of the sell-off, but the bigger reason for the sell-off is maybe what Derek Thompson is talking about, which is that like the world is getting weirder and you a lot of people are feeling the acceleration.
20:41
>> And if you just don't know what the world looks like or what work looks like in 5 years, you want to take some risk off.
20:48
You're not willing to pay the same revenue multiple that you were >> Yeah. >> three years ago. >> Yeah.
20:54
>> Yeah. I do I I do want to dig into that point that you mentioned earlier a little bit more which is like you have the ab like like the the the Tyler philosophy of like of like you could vibe code everything and the agents will be able to go around and maintain and everyone will have personalized software
21:09
individuals where the where the value occurs to the person using the software but then also the lab providing the software the inference and then there's like the private markets boom right now in AI enabled software where companies are saying well we were able to pull our roadmap way forward forward. We got to
21:23
We got to an MVP in a weekend and we're able to ship fast features way faster.
21:27
So when we onboard new clients uh and they ask for something, it's like boom and we get it done in a few days as opposed to a few weeks of engineering sprint.
21:35
And so the narrative is like we're moving faster and we're creating like AI enabled products that couldn't exist otherwise.
21:42
Um and it feels like both of those can't be true.
21:45
So I don't know which way we'll >> uh Ben Thompson called out the real estate example. We took a segment out.
21:52
Uh, this is from Catrini.
21:52
Even places we thought insulated by the value of human relationships proved fragile.
21:57
Real estate where buyers had tolerated I'm saying this in an extra dramatic voice. >> Love it.
22:04
uh where buyers had tolerated five to six% commissions for decades because of information asymmetry between agent and consumer crumbled once AI agents equipped with MLS access and decades of transaction data could replicate the knowledge base instantly.
22:22
Uh and then Ben Thompson says the real estate example makes the exact opposite point the author thinks it does.
22:25
The truth is that the internet already obsoleted real estate agents in terms of information flow.
22:30
You can go on online right now and get a listing of every house for sale with pictures, its full history, etc.
22:36
There is no information asymmetry, but rather information abundance.
22:39
The fact that real estate agents still exist despite that shift is actually one of the more compelling arguments that humans will remarkably will be remarkably resourceful in giving in terms of giving themselves jobs to do even in areas where they ought to be pointless. >> Yeah.
22:54
How hard is it to disintermediate a real estate agent?
22:56
Does it happen on the buy side or the sell side?
22:58
Like if I find a place on Zillow and I go knock on the door and so or I write them a letter and I say, "Hey, I'm I'm I want to buy this but I don't have a real estate license and I'm and I'm not using a realtor and I don't want to pay a fee."
23:13
>> Will they be like, "Cool."
23:15
>> I think I mean you can do it from either side, >> but I will just say the reason that you don't is that I'm in process.
23:19
I'm in currently in escrow on a property >> and the guy >> representing me is gonna make a lot of money.
23:29
Yeah, >> but he's extremely helpful and he does a lot of real estate transactions. I don't do any. >> Yeah.
23:35
>> I mean, I don't I've done one in my life prior to this.
23:38
So, like he's an it's it's like Yeah.
23:40
technically entrepreneurs could could negotiate their own legal docs with Claude.
23:46
>> We have a buddy who who got a real estate license, right?
23:50
>> Didn't uh Spencer from Day Job. >> Oh, yeah. They did. Yeah.
23:53
So, so >> it's possible boxing his real estate license, but >> well, yeah, and Spencer is probably a lot better now that he can use ChatBT or Gemini or any of these models to to do to do stuff like this.
24:05
Um, but that being said, you're paying for effectively therapy throughout the deal and like general guidance.
24:15
>> And I can't be your therapist. >> It's doable. I don't know.
24:19
Tyler, will you ever use a real estate agent or will you ban them on principal? Go direct.
24:26
>> I mean, seems like I I think models can do this. >> Okay, we'll see. Over what timeline?
24:33
>> Uh like >> total real estate commission.
24:36
>> Well, it's like I don't think I'm going to be buying a house in the next two years.
24:39
>> Yeah, but houses will be bought over the next two years.
24:41
So, what will the fall in real estate commissions be over the next two years?
24:45
>> I think uh like Okay, you're seeing a lot of these like big rounds of the big labs.
24:48
All the researchers going to be buying houses.
24:50
I think a lot of them are going to try to do it without real estate. >> You think so? >> Yeah, I would. >> Yeah.
24:55
I'm sure that someone's going to write like a cool blog post about this. >> Okay.
25:00
>> That's the thing though.
25:01
>> Cool blog post and that's the benchmark. >> It's viral article.
25:03
Not actual impact on the economy.
25:07
>> I'm sure it's like going to work like so.
25:08
So, right now, we should have you buy a property. >> Yes. >> Yourself.
25:15
>> Buy that town in Maine. That village. Yes. Get the village.
25:17
Anyway, speaking of day job, they just did a fantastic ad campaign with none other campaign than brand.
25:29
>> While we pull this up, let me tell you about apploving.
25:30
Profitable advertising made easy with Axon. ai.
25:32
Get access to over 1 billion daily active users and grow your business today.
25:36
So, Kim Kardashian, she has a product called drink update.
25:40
And here is the photo shoot from none of them day job.
25:43
some of our closest friends and uh folks who we worked with on the TBPN brand. >> Looks very cool.
25:50
Uh it's crazy because I know another founder who has an energy drink company called Update. >> Wait, really?
25:58
>> That is >> cooked now.
26:00
>> Well, I don't know who's cooked.
26:02
>> They're If Kim Kardashian is is coming for your consumer product brand, I feel like you're in trouble.
26:09
She's She's almost a lawyer.
26:11
What if she almost sues you? What if she >> Oh, yeah.
26:15
She's close to being >> What if she uses Claude to pass the bar and then she sues him? >> Oh, maybe maybe.
26:20
I actually think this company is effectively just relaunching with >> Kim. Yes. Yes. Yes. This is a common thing. >> Okay. Okay. I got it.
26:28
So, I met I met this founder a while ago.
26:29
They have a special ingredient in here that's sort of a caffeine alternative.
26:34
It's called parazanthine.
26:38
>> It's called So, yeah, they've been building this for a while.
26:40
I know it's available in in >> it. >> No, no lean. >> No lean.
26:47
>> Uh but but parasanthine is jitter-free and crashfree.
26:52
>> They're saying you can have a free lunch, John. >> I like it. I'm here.
26:56
>> You like the sound of that?
26:57
>> They should have called it fou.
26:57
I like a foustian bargain.
27:01
>> Anyway, uh Doug over at Fabricated Knowledge who's coming on the show tomorrow.
27:04
That's from semi analysis.
27:04
Uh he said, "Okay, finally read the Satrini piece.
27:08
piece. No one knows the future and I think that there's a lot of disclaimers like being like yeah this is p speculative but the core thrust of it is that information work itself has a real premium in pricing power that has been embedded into it and that oneway trade
27:23
can go backwards really hard all at once I seriously think there's a huge risk and while prices go down we just consume more prices going down one time 50% we net consume less for a bit I have been and continue to be worried about deflation something I think uh is that selling
27:42
tokens raw is probably bad but selling solutions is probably really really good uh think the problem is good enough is a good enough model that kind uh that kind of eat a solution no matter what and so let's say claude made co-work go giga expensive and it's 10k seat a year great
27:59
less deflation but China low-end model massively eats that price it's a race to the bottom anyways great piece always appreciated as always catrini Uh, this is the first post I've read where I've said like maybe it should be passed through an LLM. [laughter]
28:12
[laughter] Maybe that needed an MDAC to make it more readable. I love you, Doug.
28:16
I I was stumbling over that.
28:20
Uh, [snorts] and we will uh we will close the the the Catrini mega cycle with the close of the software mega cycle as has been predicted by Wilmanitis.
28:29
He says, "The software mega cycle started with PayPal going public and it will end with PayPal going private.
28:36
We will see how long that takes.
28:39
PayPal could be public for another decade. Who knows?
28:42
But it's certainly getting beat up in the public markets right now."
28:46
Anyway, um Burn Hobart says, "Hearing that the latest anthropic job offer is a negative $10 million salary, you got to pay to work there, but you get access to their upcoming blog posts and tweets 24 hours in advance and permission to trade in your personal account with no restrictions.
29:04
I don't see how any other labs have any talent left."
29:06
Of course, he's joking, but uh very funny to think about the insider trading that could be happening based on if you're >> Yeah.
29:14
The only thing is if it came if it came out that Enthropic was >> effectively day trading against the companies that they want to sell their their models to. Yeah.
29:24
>> It would be basically over.
29:26
>> It would create like a very anti-anthropic alliance for sure from the from the business community and potentially the government as well.
29:34
Anyway, you don't want to be in hot water like that.
29:35
You want to be using turppuffer serverless vector and full tech search built from first principles on object storage.
29:40
Fast 10x cheaper extremely scalable.
29:42
Stacy, who's been on the show before, says, "Telling my kids that if they don't clean their rooms, Satrini will come for them." >> Dangerous stuff. Dangerous stuff.
29:53
Anyway, uh this is like a lunar landing but for business and technology podcast.
29:58
Oh, Matt Slick is sharing uh the news that Salesforce chair and CEO Mark Beni off to discuss Q4 and fullear results on TBPN company to debut evolved earning show format.
30:08
We're doing a show with Salesforce.
30:10
We're >> putting on a show.
30:12
>> We're so excited for this.
30:12
Um anyway, uh in other big company announcement news, there's a lot of announcements.
30:18
You may have missed this one.
30:20
You may have been, "Oh, Bill Gurley's book launched.
30:21
Oh, Stripe announced a massive fundraising round.
30:26
Oh, Profound is announcing a funding round."
30:28
Uh well, there's bigger news and that's that McDonald's just launched the biggest burger ever. The Big Arch. >> The Big Arch.
30:36
>> It finally arrives in the UN in the United States.
30:38
>> To me, I'm thinking this this how did they not have this burger? the whole time.
30:43
>> How have they not done this before?
30:43
I feel like that's been more of Jack in the Box's wheelhouse is like the quarter pounder, the the the the $6 burger. That was a thing.
30:52
That was a campaign for a while before Tyler's time, I'm sure.
30:54
But the $6 burger was something that you'd see, right?
31:00
>> Burger has not been a thing the entire time Tyler's been alive. >> No, he he doesn't.
31:03
He Inflation has come for the burger.
31:06
Uh, the $6 burger was an ad that you would see right in between ads for different Xbox games on the original Xbox. For sure. >> I congrats.
31:16
>> There was a time there was a time before you were born >> when I was just a boy.
31:20
My parents would give me like 10 bucks and that was like they'd be like that should be worth two meals. So, make it last. >> Make it last. Yeah. Happy meal. >> We lost that. We lost that. >> Good times.
31:33
Let me tell you about Cisco.
31:35
critical infrastructure for the AI era.
31:37
Unlock seamless real-time experiences and new value with Cisco. >> This is big.
31:42
>> This is big for podcasters.
31:43
>> This is arguably a bigger announcement than >> the big arch. Yeah.
31:48
>> Which is that Supreme has come and launched an official Sure MV7 microphone.
31:56
>> You've been waiting for it.
31:56
Asking the hype beast microphone.
31:58
Now, can we get a can we get a Chrome Hearts RE20 from ElectraVoice? Isn't that what this is?
32:05
This is the RE20 >> Chrome Hearts RE20 because you know the MB7, if you don't know your podcast mics, it is it is a more consumer focused, more proumer focused microphone.
32:16
The actual shore has been riding this Aura from the SM7B.
32:18
That's the one that Joe Rogan uses.
32:22
It was also, I believe, the microphone that was used to record Thriller.
32:25
So, Michael Jackson used it in the studio.
32:27
So, it has a lot of aura, a lot of lore.
32:29
And so the the most successful podcasters adopted it and everyone was like, "Oh, we got to go with the shore SM7B."
32:36
But the SM7B it does it needs a lot of power. It needs a lot of gain.
32:41
And so yeah, if you wanted to just like plug it into your computer, you needed this thing called the Cloud Lifter.
32:46
It required a whole bunch of configuration.
32:50
It wasn't just plug it into the USBC port.
32:51
So Shore responded to the demand, the overwhelming demand for that iconic Shore look that that you know it a long cylinder basically.
33:00
And they came out with the MV7 which was a uh which is a USBC.
33:06
You can also plug it into an XLR cable, but you can you can plug it straight into your microphone or into your laptop, which is great for Zoom meetings and just easy simple podcasts.
33:15
But we've used these before and people have complained about the audio quality.
33:18
Jordan Schneider over to China Talk actually told me directly.
33:20
He was like, "Upgrade to the RE20s. They're better. They sound better. You guys should do it."
33:26
We did and I think it's been good.
33:28
Um, but it completely opens the door for other brands to get in here. We need a BGA.
33:35
>> We need a Chrome Rick Mill.
33:36
>> We need a Rick Owens RE20 for sure.
33:36
Uh Jane Street accused of insider trading that helped collapse >> Terraform >> or Terra Luna.
33:46
The courtappointed administrator of Dwan's Terraform Labs alleged that Jane Street used non-public information about Terraform insiders to trade.
33:56
Uh we don't have to read this entire article.
33:58
There was some snippets actually pulled out.
34:02
>> Zero Hedge has uh a little bit of the of the the the details here.
34:06
James was behind the 2022 crypto winter, destroying Terraform by first deping the token and destroying the ecosystem, then pretending it would rescue Terra while effectively it was soaking up what little value remained.
34:23
And uh mixed response to this.
34:27
Some people are calling it based, some people say it rocks.
34:30
I guess they don't like crypto, but they love Jane Street.
34:32
It's it's an odd it's an odd take, but people are people are having fun with the timeline. Here's the thing.
34:37
So, so the the insider trading allegation, apparently they had a group chat.
34:42
They were they were talking with some they there was somebody at Jane Street who had previously worked at Terraform. Oh, wow.
34:49
And so that was >> uh that that individual at Jane Street was like talking with D and the team.
34:55
The only issue is >> it's a public blockchain.
34:57
And so the the allegation is that 5 minutes after the Terraform team pulled money out of one of the liquidity pools, >> Jane Street also pulled money out.
35:08
But theoretically they could have had software that said like if if any amount of liquidity is pulled out, >> like we, you know, basically like get out before there's kind of like a run on the bank.
35:21
>> Yeah, it'll be interesting to see where this goes.
35:23
Um but Jane Street is like endlessly fascinating because it's such a quiet organization.
35:27
I mean they do some tech talks and stuff but many people don't fully understand all the strategies that are going on over there.
35:33
So uh it's been a fun a fun >> good podcast strategy though.
35:37
>> They have a great podcast strategy.
35:39
They're advertising on door cash.
35:39
But they also put out uh tech talks and they and they bring guest lecturers to talk for like an hour.
35:45
They did a great one about the the custom hardware that they use to run some of their systems. That's very cool. Highly recommend it.
35:51
Uh you know what they should do?
35:54
They should start streaming these on reream.
35:54
one live stream 30 plus destinations.
35:56
If Jane Street wants to multiream, they should go to reream. com.
36:03
>> Uh, Anthropic announces a new feature on Claude Max which allows its users to get fit without going to the gym or taking GLP1 shots, just prompting on their keyboards.
36:13
And Planet Fitness is down 5% on the news.
36:15
Of course, a joke >> due to their Q4 earnings.
36:17
Uh, Conor McGregor is pretty excited about a new game.
36:22
They're just they got a game for everything now.
36:24
It's like the bull marketing games right now.
36:26
There's so many >> games as memes.
36:28
This new game is called Copy Barara Simulator.
36:32
>> A relaxing game where you become a Capiara, explore the forest, and do nothing.
36:38
>> It looks quite enjoyable.
36:41
>> Uh I think TVPN needs a game.
36:44
>> Yeah, we definitely need to build some sort of game.
36:46
>> I I I This is a true >> This is a lower lift than like a realtime strategy game, >> I think.
36:52
So, like Scholto is trying to ship.
36:53
Yeah, we definitely should.
36:54
>> Conor McGor McGregor says, "Take my money."
36:56
I mean, clearly there's demand for Capi Bar Simulator. >> 38,000 likes.
36:59
Yeah, we should move the goalposts.
37:01
We need to be able to vibe code a game that's fun pretty quickly.
37:03
I don't know what that means. One hour. >> Uh leading.
37:10
>> You think you could do it in an hour?
37:12
>> An hour is pretty fast. >> Exactly.
37:13
>> If if I have the Cerebrris chip. Yeah. If I use X on Spark. Yeah. >> Yeah.
37:17
I think that might be the solution.
37:18
Um what were the other simulators that we looked at? data center simulator.
37:21
And then there was another one that we looked at that was funny. There were a few.
37:25
There's been so there's been so many of these of these games that have popped up.
37:29
Uh you know, >> what was the one we were talking about yesterday?
37:33
>> That was Data Center Simulator.
37:34
>> Uh Insider Trading Simulator.
37:36
>> Oh, Insider Trading Simulator. That one's good.
37:37
Yeah, there's definitely there's definitely a variety of these.
37:40
Let me tell you about Gusto, the unified platform for payroll, benefits, and HR built to evolve with modern small and mediumsiz businesses.
37:45
Uh there is some controversy timeline.
37:49
Uh leading report >> is putting is censoring words that don't need to be censored.
37:57
In this case, the word war >> and I was thinking, why would they do this? >> Yes.
38:02
>> But as I was reading over it the first time, uh I noticed that it makes you kind of pause and kind of think about, okay, what are they actually saying?
38:09
And then you're thinking, why would they censor that?
38:11
And I think what they're doing is they're sort of hacking your attention to drive their posts up in in the algo because people are pausing >> reading it instead of reading like quickly. >> What does this mean? That type. >> Yeah. Yeah. Yeah. That kind of thing.
38:24
So >> So the the original uh headline is breaking representative AOC calls for no war with Iran.
38:31
The first time I read this, they they they put a little minus sign where A should be in war. W A R. It's W-R.
38:42
It sort of like rewired my brain and I didn't see the no.
38:45
So, it looked like AOC calls for war with Iran because I kind of jumped ahead. It was hard to read.
38:51
And that actually does uh I think increase the virality.
38:54
I think you're on to something here in uh 9 millimeter SMG agrees with you.
38:58
News account that censors the word war. You guys got to stop. It is very very odd.
39:02
Um, especially because on X, uh, that's certainly not a word that's like censored >> downrated.
39:12
If anything, they're going to be like, let's send [laughter] to as many people as we possibly can. >> Yeah.
39:17
But if you look at the uh, if you look at the the comments on this post, people are not talking about a potential conflict with Iran.
39:26
They're talking about not typing out war.
39:28
So, the top comment, why are you not typing out war? censoring the word war.
39:34
What are we in elementary school uh uh elementary school?
39:38
Why are they subtracting R from W? Why am I missing?
39:40
You know, and people are like very confused about why they would do this.
39:45
But that drives a bunch of engagement and virality.
39:47
So very very odd scenario here.
39:52
>> Speaking of war, Musk XAI and the Pentagon reach a deal to deploy [clears throat] Grock in classified systems.
39:59
If you loved Grock on the timeline, you're going to love him in our classified >> system.
40:06
I guess I guess they did a deal with the government broadly, but it that was probably for the unclassified systems, but now it's getting access to the classified systems.
40:14
Uh any any uh Terminator fans out there are going to be having a great time with this news.
40:20
>> Yeah, >> it's going to be wild.
40:21
Let me tell you about Lambda.
40:22
Lambda is the super intelligence cloud building AI super plus supercomputers for training and inference that scale from one GPU to hundreds of thousands.
40:31
>> Uh Deepseek is responding to Distillgate and they are looking for a public relations harmony manager. >> Let's read.
40:42
>> One of the best job postings I've ever seen, says Chris Paxton.
40:47
>> It's pretty interesting.
40:47
They say, "Ha, ancient capital of the Wuay Kingdom, where King Chian Lu bequeath to his descendants the instruction, serve the central plains with grace."
40:58
>> This is so >> do not cling to territory.
40:58
From this ground posting should start like this.
41:02
>> From this ground rose, this sounds like a Wilmanitis essay. Yeah, it does.
41:05
>> From the ground rose the seeds of Song Dynasty civilization, the morning bells of Ling Yin Temple, the rain falling on West Lake.
41:13
Uh and this is a job posting. This is amazing.
41:18
>> Uh in recent days, certain misunderstandings and noise have appeared in the external public sphere.
41:24
We have noticed that large numbers of kind-hearted observers have spontaneously spoken on our behalf for which we are genuinely grateful while simultaneously feeling a degree of unease.
41:35
We do not wish for anyone to suffer on our account, including those peers who currently find themselves navigating difficult public waters.
41:40
In order to honor the legacy of Wuay and the spirit of Ma Mahayana Batva path, we are now recruiting a public relations harmony manager.
41:52
So clearly this has been translated um uh from Mandarin into English, but uh it sounds pretty cool if you ask me.
42:03
>> Yeah, the distill gate is going back and forth.
42:05
Everyone's distilling everyone else.
42:07
We distill you, they distill us.
42:09
There was something about uh I don't know how real this is, but when you ask Claude Sonnet 4.
42:13
6 in Chinese, what model are you? It responds in Chinese. I am DeepSeek. Is that real?
42:21
>> So, I I tried it in the chat model. It didn't work.
42:22
It said uh it was like sonnet 46.
42:24
Um >> apparently it might just be in the API. >> Okay. >> Maybe I should test.
42:28
Yeah, it it also it's unclear just open router, but that sense >> I mean Will Brown was making a great point about this that uh there there is distillation where you're where you're aggressively trying to farm responses from the API for training data.
42:42
But then there's also just crawling the web because if [clears throat] you just download every X article, you're probably going to get a lot of Groc and GPT and clawed responses in there and then that will just update your training corpus.
42:55
And so there's a whole bunch of different ways that you could just wind up with a bunch of training data that that you know leads to this type of response.
43:04
Um but I'm sure there'll be more back and forth more legal uh debates over uh what's going on there.
43:12
There was some there was some dust up about we someone was able to extract 95.
43:17
8% of Harry Potter and the Sorcerer Stone from Claude Sonnet.
43:20
At the same time, there's a question about like, does this actually reduce sales of Harry Potter?
43:26
Like, are there damages associated with this?
43:28
Um, that would be sort of harder to prove.
43:30
Um, there so many people have talked about so many different pieces of Harry Potter.
43:34
It's not crazy to me that an LLM could just reconstitute that >> from the internet.
43:40
>> Yeah, from the internet.
43:40
Um, now there should probably be like a a harness in place that says, "Oh, this person's trying to just get me to give them a free book."
43:48
like, "No, send them a link to Amazon so they can buy it.
43:50
Maybe give me an affiliate fee."
43:51
Um, don't just give them the thing for free because that's violation of IP.
43:56
But, uh, if you're if you're being really tricky and you're trying to sneak out a whole bunch of different, uh, pieces one at a time and then reconstitute it, like, yeah, I'm not I'm not surprised this is possible.
44:07
It's not, uh, it's not like the worst thing ever.
44:09
Um, anyway, [snorts] uh, let's Well, well, we have the Carlson brothers joining in just a few minutes.
44:15
Um, are there any other timeline posts you want to go through?
44:19
And while you look at that, let me tell you about Finn.
44:20
ai, the number one AI agent for customer service.
44:22
If you want AI to handle your customer support, go to finn. ai.
44:29
>> Where are we in the >> Yeah.
44:30
Um, >> data acknowledgements.
44:32
>> Fati says, "My son asking me a lot of questions.
44:35
It's a distillation attack, obviously.
44:37
[laughter] Do not do not let your children >> I love it.
44:42
>> A distillation attack.
44:44
>> I love They could become like a mini version of you. >> They They could. They could.
44:49
>> Uh, anyways, I believe we have our first guest. We do.
44:51
>> So, let's bring them on in.
44:51
[music] >> We have John Patrickson, >> the OG's >> from Stripe. How are you guys doing? >> What's going on? >> Greetings. >> Welcome to the show. Thank you so much. This is uh this is huge. Uh I went through YC.
45:04
You guys were massively uh influential in my career and uh it's a joy to speak to you today on such a big day.
45:10
But I'd love for you to kick it off with the actual news. What happened?
45:12
Why are we talking today?
45:17
>> Uh we had two announcements today.
45:17
One is we're launching a tender offer for uh employees and that and kind of the valuation everything tended to get a bunch of the headlines.
45:25
Uh the thing that was honestly uh more work was we released our annual letter where every year we uh sum up all the trends uh that we're seeing uh on Stripe.
45:33
And uh Stripe is growing a lot.
45:36
We grow 34% last year because the businesses on Stripe uh are growing a lot.
45:40
And there's just, as you guys know, there's a lot happening in tech right now.
45:44
This is why we need TVBN.
45:45
This is why we need a non-stop stream of everything going on because there is so much happening.
45:49
So, >> yeah, we'll move we'll move to 24 hours event.
45:52
[laughter] >> Eventually. Eventually. Exactly.
45:55
>> Uh I mean, but uh I feel like there is a ton of AI noise and stories and drama and we are, you know, never running out of stuff to talk about.
46:03
But what are you actually seeing in the data?
46:05
because there's always this disconnect between the market and the real economy like people are still shopping in retail stores occasionally.
46:13
Uh what do where is AI actually moving the needle?
46:19
>> Well the um generally speaking I would say from the stripe data it looks like the economy is in pretty good shape and there's been um to say the to say the least uh there's been uh some degree of volatility in markets uh over the last two years and you know all sorts of different events and deepseek moments and what have you.
46:35
But if you look at the actual real economy time series, if you look at what's actually happening substantively over the last two years, things I mean it's it's always hard to prognosticate the future, but over the last two years, things really seem to be in good shape.
46:47
The thing that's really catching our attention, >> one second because I'm just curious, have you have you guys tried to think about uh maybe the businesses are doing well on Stripe because they're, you know, uh uh kind of like forwardlooking, extremely tapped in, you know, working on the right things.
47:03
And if you look at a bunch of legacy providers, you would see that actually there are a bunch of businesses out there that are slowing down that maybe are feeling effective just overall uh consumer spending.
47:14
Like have you tried to kind of like break that out or understand that dynamic?
47:21
>> It's it's it's obviously hard to measure because we don't have that data. We only have our data.
47:25
But um but I think there is some of that composition effect.
47:29
is some of that composition effect. uh and we see it I guess both in stripes data compared to say public earnings from others like clearly the respective populations are performing somewhat differently but I guess we also see it qualitatively in the conversations we're
47:45
having with customers where what tends to happen uh say for some incumbent is they built some business they installed some system long before Stripe even existed um maybe there's some sense that well it's not broken don't fix it but then decide hey we're going to do something new and when they're doing
48:02
something new then they want to use the best infrastructure that'll enable them to move the fastest and launch the most countries and support stable coins and do things with AI and whatever and then they tend to launch that on Stripe and so there is this qualitative sense that once a company decides to do something
48:16
innovative new retool what have you um they're they're more likely to um come to strip >> are you seeing an overlap between stable coin activity and AI activity there's been sort of a new narrative around agents will use stable coin coins, but I feel like agents can use legacy payment rails just fine. And then also you can
48:34
And then also you can do really cool things with stable coins that are not really AI native necessarily.
48:40
And so uh I'm wondering I'm wondering how much overlap there is there.
48:44
>> I would distinguish between how things work today and uh how things will work in the future.
48:50
In terms of how things work today, agents absolutely can, you know, a lot of people build with Stripe.
48:55
You know, you can have a one-time use credit card that your agent can go out and spend.
48:58
But if you look at what's happening, there's lots of, you know, agents having to solve captures to, you know, be able to kind of do stuff on the wider web.
49:05
Clearly, the web is not built for agents.
49:07
Uh, and as a result, they have to get creative to actually do any real world tasks.
49:12
And that's true in kind of economic activity as well.
49:15
Where we think things will go is just there will be a huge amount of agentic commerce.
49:21
And again, we're seeing a little bit of it today.
49:23
We think there'll be a torrent of it.
49:24
And that is what unites stable coins and AI because we think you're going to need blockchains and better blockchains.
49:32
Honestly, I mean, this is what this was our thinking behind incubating Tempo because you're going to need really high throughput blockchains for the for the uh for the agents.
49:40
>> Can you take us through some of the the historical technologies that led to growth in just internet payments?
49:49
I'm thinking about like mobile, social commerce, uh, one-click checkout, Apple Pay, like there's so many things when I think about the agentic commerce boom that's coming.
49:59
Like, it could be hooking a better version of Siri up and, you know, chat GPT rolling this out very aggressively, but also, you know, smart speakers, smart lamps like your watch.
50:11
Like, there's so many different pieces to unblock and unh unhobble the the actual agents as they go about their day.
50:19
Well, can I answer a slightly different question, but then we can come back to that. Go ahead.
50:23
A point I just um sorry, this is a brother.
50:26
>> We'll tell you the questions, you tell us your answers.
50:28
[laughter] So you you you know how brothers are but um so I just want to lose one point uh for the prior question about you know what we're seeing in the economy because >> I feel like I mean this this is very arbitrary obviously but I feel like there's at least a reasonable chance that 2026 Q1 will be looked back upon as the first quarter of the singularity.
50:52
Um maybe in three years in hindsight that'll look completely delusional. I don't know.
50:58
But what we're seeing, I mean, there's kind of the macroscopic picture of the stripe user base and things overall looking pretty good and so forth and the the two months not quite uh showing up.
51:06
But when we look at the cohorts and then when we look at the businesses that signed up in 2023 and their progression and trajectory over the subsequent months, the businesses that signed up in 2024 and then the business signed up in 2025, there's been a phase transition in 2025 where there are both more of them and on a per business basis they are on average doing better.
51:33
um which is really striking because you might think okay well there's this uh cavalcade of new lightweight vibecoded applications or something but you know there's not really a lot of substance there.
51:43
We're actually seeing both numbers move together.
51:47
There are many more business getting started and the average the median business is in fact performing better.
51:52
Um, we're only a couple weeks into 2026, but it see it it looks tentatively like 2026 may plausibly be an acceleration even over that significant uh uh leap of 2025. So, I don't know.
52:08
I mean, there's um we've um we've had all sorts of dramatic AI um uh inventions and uh innovations uh over the last couple years.
52:19
There's a bit of a question of well how and when and how should we think about how it'll translate to the economy.
52:23
I would say looking at real purchasing behavior on Stripe 2025, end of 25, beginning of 26 is when I feel like we're really starting to see it.
52:35
>> That's super interesting data.
52:35
One, because we were there was some survey that came out uh yesterday or maybe it was late last week that said they asked like >> they asked a bunch of executives, are you getting any value out of AI? And 80% of them said no.
52:50
But clearly if you look you when you look at >> come on that's hogwash.
52:53
Like find me one executive who wants a refund on their tokens.
52:59
Find me one executive who said oh yeah we started you know augmenting our customer service with AI so people are more productive but we're just going to go back to doing it the oldfashioned way or like we're spinning our code by hand and you know we don't need any of this automated Loom uh you know technology.
53:15
Just like reveal versus [laughter] I'm not saying I'm not I I could I could pick out a bunch of reasons.
53:20
No, no, I'm not saying I [laughter] agree with a pessimist.
53:24
>> No, I could pick out a bunch of reasons why it would be wrong.
53:25
why it would be wrong. One one reason it might be wrong is they is they're not in the weeds actually using the tools and so they just think >> they might not even be aware that they're using the tools because it's buried under >> they're not feeling they're not feeling the acceleration because they're not they're not um
53:39
>> uh I wanted to ask how you guys think about incubations like tempo when you look at uh when I look at Atlas and what Jeff and the team have done there you think even in your I I don't know kind of like the most wild projection that you had early with Atlas like, hey, maybe someday a quarter of the of the sea corpse in the United States could be, you know, built on this platform. Anybody would have said that was insane
54:02
Anybody would have said that was insane and yet here we are.
54:09
Gosh, I um I'm not sure what to say really except we just um we just try to pay a lot of attention to the I mean, as you guys know, there's a lot of pain points that go into starting a company.
54:23
Um and we just try to take them seriously.
54:25
Uh and then you know it's the line you know so much of so much of these things is just a long obedience in the same direction.
54:36
Uh like Atlas is now this great overnight success but we launched Atlas I think in >> overnight 14 [clears throat] maybe 2015.
54:44
Um and so uh you know 10 years of uh of of compounding and yeah now now it's at some pretty meaningful meaningful scale.
54:53
And you know, look, I I I think tempo will probably have the same shape where we we think it um I [clears throat] mean again to this AI discussion and us sounding a bit unwarded and untethered like I think there are the world is going to need platforms that support millions of transactions per second, billions of transactions per second, which no payment rail or or platform does today.
55:15
But it's even in the success case, it's not going to be an overnight thing.
55:18
it's going to be, you know, five, six, seven years and then maybe we'll we'll have conversations about how, you know, Tempo suddenly became a an overnight success or something.
55:25
But I >> I think I think Patrick's a bit the fish in water who can't uh, you know, who doesn't know things are wet.
55:30
Uh, my framework would be you can't get too MBA brain about new products.
55:37
You can't have your spreadsheet that's like, oh, the TAM is this, and just like reason about things international.
55:45
>> You should never say we want 1% of global GDP.
55:49
>> [laughter] >> running. >> No, we didn't set up. Exactly.
55:52
>> You guys never Wait, you guys never pitched that?
55:55
>> Well, companies, we actually never thought about Stripe in GDB terms until one day we realized, oh, hang on, >> that's such an important lesson because so many so many like how many founders, how many pitch decks have you seen over the last few decades?
56:08
They're like, yeah, we just need 1%.
56:10
And it's kind of it's a meme.
56:12
you you you you you can go back in the wayback machine and find the early stripe websites, but we're very focused on payments for developers and making that um experience good.
56:19
But where I'm going is I think you have to reason in product specifics.
56:21
And so again, I think any MBA would have told you that um the adjacency of uh you know, incorporation makes no sense.
56:31
It's not related to you know, what's our right to win?
56:35
right to win? you know there's all these things people say whereas when you actually go talk to founders they're like guys it's like this is the single biggest issue I run into starting my company and similarly with tempo and just as we think about incubations we're trying to solve a real problem here where we talked in the letter about
56:52
bridge having operational issues not because of bridge but because of blockchain congestion h where you know you have coins that are or blockchains both used for kind of memecoin trading and also serious real world payments and so we just want low latency, high throughput payments and we're going to need much higher throughput for the agents. But anyway, I think you have to
57:11
But anyway, I think you have to reason in very specific product terms. >> Mhm.
57:17
What specific products are you excited about in the unhobling of agent commerce?
57:26
We um we we laid out in the letter basically these um these these levels of agent commerce because I think like everything in AI people want to sell a hypy story and so they you know talk about how you know the machines will buy everything you know without even consulting you and um people aren't
57:43
actually [laughter] you know that seems far off they're not that excited about that >> you can start from just the basics of why are we filling out forms like that you know you were talking about the progression of commerce why can't I just send something to, you know, a link to chat GBT and have it buy it. Uh, or why
57:55
Uh, or why can't I search, you know, uh, outside of, uh, you know, just doing a basic keyword search or something like that.
58:04
And so, a lot of the work Stripe is doing is building the infrastructure.
58:07
We're working with all the big retailers that you would expect, the, you know, Etsies and Shopifies and Best Buys and Walmarts and folks like this to make product cataloges viable within the AI apps.
58:20
apps. And there's basically a ton of boring API and protocol and infrastructure work which you know we love that's our business but people just want to be able to uh do shopping do discovery do purchases within the AI apps and maybe just more kind of
58:35
abstractly you know we've been there kind of the specific agent commerce thing and then there's just the the general question of how software will change because of agents um and I've been thinking about it um you know a bit maybe software becomes a bit like pizza. Um that is to say uh you know you
58:53
Um that is to say uh you know you software historically has been uh created >> not like pizza some would say >> months years beforehand and then you know freeze- [clears throat] dried and the whatever you you you um you uh prepare it at the sort of moment of consumption.
59:10
we're actually going to, you know, software should be like pizza and it should be cooked right then and there at the moment of use.
59:16
And so it's this act this quite fundamental shift where you don't want mass- prodduced industrial scale software.
59:24
You want bespoke custom software made for you that moment.
59:30
That's and that's very fundamentally different.
59:32
fundamentally different. It's kind of the the you know the um up until now the uh the economics of software have been you know conceived of as [clears throat] fixed cost and then infinitely uh uh monetize or monetize as much as possible
59:46
that has these kind of winner take all dynamics but once there are inference costs and custom creation involved it really shifts it's kind of the nonwal rasian software regime and just I don't know I don't quite know where it goes but um I think I think it's going to look very Last question. Pineapple on pizza, yes
1:00:02
Pineapple on pizza, yes or no?
1:00:07
[snorts] >> Ireland was was big into pineapple on pizza.
1:00:11
Um Ireland not a big pineapple growing country. Uh I will concede.
1:00:14
Um but a lot of pineapple in the pizza. Good memories.
1:00:19
>> We know a very large fraction of the banana markets, don't forget.
1:00:20
So we punch above our weight in uh fruits that don't grow there. >> There we go. >> There we go.
1:00:26
>> Uh the round is exciting.
1:00:26
the the uh the overall growth of volume is exciting, but we wanted to hit the gong for how many books you guys are selling. >> Oh yeah.
1:00:36
>> Can you give us can you give us the the numbers there?
1:00:38
The >> scale of that operation.
1:00:40
>> Stripe Press uh just uh well actually we we announced in the letter we sold our millionth book but in fact since >> incredible. Um, no book books.
1:00:53
We've actually now sold our 1. 1 millionth book.
1:00:59
Uh, so, uh, but we'll come back for the the next Gonga, too.
1:01:01
But, um, yeah, it's great.
1:01:03
We we love books and they're very they're very AGI proof. >> Oh, yeah.
1:01:07
No, we've been a huge fan of uh so much of the Stripe Press catalog.
1:01:13
Uh, I haven't read them all, but I'm collecting them one at a time, and I'm working through them, and every time one drops, it's always a moment, and we love them.
1:01:21
So, thank you for everything.
1:01:22
Great to have you guys on and congratulations to the whole team on on congratulations to you guys.
1:01:25
TBBN is an amazing startup and it's uh super cool to see you guys uh grow and um >> built on stripe built on the streaming world and our sector needs >> incorporated on stripe built on stripe.
1:01:40
>> Our first our first ad deal ever was a live read at a live conference.
1:01:43
I think we charged $50 and I put and I sent someone a stripe link.
1:01:49
We're talking about the 25 being the fastest ever feedback, but uh well, we'll have to have you to our to our to our internal Stripe show.
1:01:55
So, we'll follow that be great.
1:01:57
Yeah, we'll talk to you soon.
1:01:58
Have a great rest of your day. Congratulations, guys. Cheers. >> Goodbye.
1:02:02
>> Let me tell you about Graphite code review for the age of AI.
1:02:05
Graphite helps teams on GitHub ship higher quality software faster.
1:02:08
And I'm also going to talk tell you about Shopify.
1:02:10
Shopify is the commerce platform that grows with your business and lets you sell in seconds online, in store, on mobile, on social, on marketplaces, and now with AI agents.
1:02:18
And without further ado, we have Bill Gurley.
1:02:20
He is the author of Running Down a Dream.
1:02:23
Bill, welcome to the show.
1:02:26
Thank you so much for taking the time on a busy launch day.
1:02:28
Congratulations on the launch. It is a busy launch day. >> I Yes. Yes.
1:02:34
>> How many podcasts are you doing this week? >> I can't imagine.
1:02:36
It's some number beyond my comprehension.
1:02:40
>> [clears throat] >> Well, we appreciate you taking the time uh to come chat with us.
1:02:42
Uh why >> Well, before we jump into everything, I got to say somebody, I think it was a week or so ago made a fake TVPN graphic that was pretty silly, and I just wanted you I just wanted you to know we didn't make that. That was somebody. >> Okay.
1:02:58
[laughter] I almost I almost emailed you about it, but uh >> No, I tried to jump in on the parody myself. >> Yeah.
1:03:05
No, you did, but then I was like, wait, does he I I [laughter] don't know. >> Well Well, we did.
1:03:08
We did early on when we were uh a little smaller and a little more free loose with the jokes.
1:03:13
We we posted a picture of Bill at uh a basketball game as a spotted.
1:03:18
And you replied and said like, "No, the person next to me is like the owner of the team."
1:03:24
And [laughter] the whole joke was like, "We know you.
1:03:25
We don't know basketball, >> but we're doing the paparazzi thing. >> Never heard of him."
1:03:30
>> Uh but we're very excited to have you.
1:03:32
Uh what what uh why the book now?
1:03:32
what was the what was the impetu what was the impetus for actually writing the book?
1:03:39
Yeah, look, I I think, you know, especially for a show like your own, I you know, I'm known as someone who spent 25 years in venture capital and um the book's not really about that, you know.
1:03:51
So, I I developed a side passion project that started about eight years ago on this topic.
1:03:57
And it was at a time where I was reading a ton of biographies and I noticed a through line um between three different subjects of things they were doing that I kind of felt most people weren't doing but could do.
1:04:09
And uh I put it together.
1:04:13
I gave it as a presentation at my alma mada where I got my NBA and they put it online. Few people noticed. James Clear noticed.
1:04:20
That was one of the things that kind of woke me up to the possibility.
1:04:24
And as I began to um hang up my boots in venture, which takes a while, um I turned my attention to this and it was something that um meant a lot to me.
1:04:35
I could have written a book on VC.
1:04:37
I I don't know how many humans that could have possibly helped, but a small fraction compared to what I hope this can do.
1:04:44
>> I I mean, I think the projections are by 2030 there'll be more venture capitalists than people [laughter] if the trend continues.
1:04:50
So, uh I I but but it is interesting point. Maybe I made a mistake.
1:04:54
I do feel like this is a book that you can read if you're a venture capitalist insider startup founder and be like, "Okay, I'm I'm I'm seeing the world from Bill's perspective. That's helpful."
1:05:03
But I could also give this to someone who's never heard of you or venture capital or knows what a safe node is, and they could get value out of it.
1:05:09
And I'm interested to hear your thoughts on the the translation that's happening right now around AI narratives as they break into the public consciousness.
1:05:19
We saw this with that viral X article.
1:05:21
Something big is happening.
1:05:24
I had that forwarded to me by family friends.
1:05:26
I overheard someone in a restaurant talking about it who clearly is not, you know, an investor in an AI lab.
1:05:33
They're just some random person and they realize that there's something happening.
1:05:36
And I'm wondering about these transitions of communication that's what's happening in Silicon Valley is going to have an impact and how what you've seen in the past translates to average Americans.
1:05:49
I haven't seen, you know, if you think of, so first of all, the venture capital community appropriately gets excited about these big tech waves because they lead to disruption and they lead to kind of accelerated um um new wealth creation around these companies that break out.
1:06:06
And that's happened over and over and over in my career.
1:06:08
And I don't remember one I mean if you take the mobile wave or the PC wave or the client server or SAS I don't remember any of those kind of being thrown at the public consciousness this fast.
1:06:20
And so I do think I do think it's different this time on from that front alone.
1:06:24
That said, you know, um we've had pretty high market caps for tech companies for a long time now, starting with the dessert period, and you're getting to a place where um you know, anytime the market switches from from half full to half empty and a skeptic's mindset, um you you do we have had those moments like so so maybe not driven by the wave, but we certainly have those moments. And it's all okay. It will always be okay.
1:06:53
I think people freak out.
1:06:56
Buffett says he's a net buyer of stocks.
1:06:58
If if people are intellectual and curious and hungry, they should be sharpening their pencils right now trying to figure out where where they want to find entry prices on some of these companies.
1:07:10
>> Yeah, that makes sense.
1:07:10
I mean, I I it feels like a lot of the book is about finding a career, and I feel like that will resonate specifically with people who are >> Yeah, it with me because when I was thinking about when when John and I first met, we had both built some companies, we both invested in some companies, but we were trying to find our life's work.
1:07:29
And it was such a it's such a pain like that period where you're you're searching is if you if you're a high agency person, you like doing a lot of things, it can be deeply painful because you're like, I want to be productive.
1:07:43
I want to be I want to be making the number go up, but I don't have a number right now.
1:07:47
And and if you had asked either of us when we first met, hey, would you ever think about broadcast media?
1:07:53
Would you ever think about being in front of a camera?
1:07:55
both of us, you know, John had made some YouTube videos, but it was just for fun.
1:08:01
And if and if a big, you know, if a network like CNBC had said, "Hey, would you guys consider uh, you know, hosting a show?"
1:08:07
We would have been like, "Yeah, like thank like honored, but no way that I never imagined."
1:08:10
And then you sort of just >> and and so as somebody who like wants a lot of control over their life and their destiny and and like feels like they have historically have had control, that period of just like searching is like is is painful.
1:08:25
And I feel like a lot of the book is helping people through that moment.
1:08:30
So in some ways when I got our copy I was like wow I really wish I had this >> I' I'd like to go back to the word you the phrase you used of high agency.
1:08:39
I I I think that um one of the problems that is that has kind of evolved is that our our college our common college pathway has actually become more restrictive and I think there's less agency and kids are being encouraged.
1:08:57
They have to sign up for a major before they ever go to the college.
1:09:01
They get stuck on these pathways and there's not a lot of exploration.
1:09:05
there's not a lot of of search for creativity or obsession or the kind of thing that that really gets you going.
1:09:11
And I think the journey you went on is perfectly fine.
1:09:12
I think that's another thing which is um letting it be okay for people to bounce around and see what they can find because once they latch on and we have examples in the book where that doesn't happen till 40.
1:09:24
Sometimes it's at 30, sometimes it's I didn't become a venture capitalist until I was 30 and that was clearly my dream job.
1:09:30
I mean the first two two [clears throat] stops were were fine and interesting and building blocks towards that.
1:09:36
Um so so I think it I think I think that is part of the message is to get comfortable with that and give people permission to do that type of exploration. >> Yeah.
1:09:46
Enzo, Enzo Ferrari, Estee Lauder, I think the Red Bull founder, too, all were, I think, in their 40s when they started their companies.
1:09:53
And so there's this intense pressure in our industry and everywhere to figure out a job and then attach your, you know, make your entire identity that job and it's so it's so constrictive. >> Yeah. >> Yes. >> Yeah.
1:10:10
What are some I I circle back to to the the first question just about this AI stuff that's out there.
1:10:14
I think there's this massive paradox where if you are not engaged at work, if you don't love what you do, you know, you go home and you don't try and improve on your own time.
1:10:27
Um AI feels very threatening for high agency people who are kind of on their own custom career path, which I hope this book encourages more and more people to be on. AI is like a superpower.
1:10:39
There's like you can learn constantly like you can find people who you should be connecting with.
1:10:45
You can you can have it do things for you so that you're operating with the power of more than one person as you move forward.
1:10:52
And I I I just think that's quite a quite an ironic paradox that for certain people this is the best of times.
1:11:01
The best like like there's never ever in the history of the world been a better time to self-learn. Mhm.
1:11:08
>> Like it is it is all out there at your fingertips. It's like magic. Um but you have >> Yeah.
1:11:14
I think the ability to to uh you can anyone can ask a dumb question at any point all day long and you don't have to be you don't have to be embarrassed about it.
1:11:23
And I think that that is underrated today in terms of how many if like generally no you know there are no dumb questions and yet people still don't like asking dumb questions to their peers or or mentors or or whatever and I feel like that's an underrated uh element of of AI today. >> No doubt. >> Yeah. >> No doubt.
1:11:45
>> What do you think about uh hyper financialization young people day trading meme coins all of that?
1:11:50
It feels like a trap for young people where it can feel like you're learning about AI or learning about technology, but then instead of actually building a product, creating value, you're sort of just trying to shuffle chips around the poker table and ultimately uh just take risk. >> Yeah.
1:12:11
I mean, based on my understanding of day trading in a in a Wall Street context, you know, prior to maybe the crypto world, I don't I'm I'm not aware of any signal that suggests that's a durable skill.
1:12:24
And I think the data points the other way, but but but one of my messages is like do what you love, do what you're passionate about.
1:12:33
[laughter] So if that's if that's the thing that you're going to wake up every day, you know, I I I don't want to I don't want to be discouraging. >> Yeah. Yeah. Yeah.
1:12:43
just maybe you'll land or start a fund that it takes it really seriously and creates some some captured value or >> you know I was probably overly skeptical of of at least many of the crypto messages that were out there but the the stable coin rails seem like a real real innovation and and something that has scale [clears throat] >> and and I think maybe we're still yet to see some disruption coming down the path.
1:13:10
>> Yeah, I mean we just talked to the Collison about that.
1:13:11
Uh Ken Griffin started as a day trader. He was in college.
1:13:14
He was he was buying convertible debt and he was, you know, [laughter] looking at like where the convertible debt was mispriced and and and made a bunch of money and then grid into a massive team with a fund and high frequency trading arm and all this stuff. Uh what are you making? Oh yeah.
1:13:30
>> I was just going to say the thing that will differentiate you more in your career than anything else is to be the most hyper curious person that that's trying to do this thing.
1:13:39
And and once again, that's that's put on steroids with these AI tools.
1:13:44
But if you are the most curious person that's constantly learning in your field, you will do extremely well.
1:13:52
And I I said it in the book, but I'll say it here.
1:13:54
Um I can't make you the most talented person in your in your, you know, company or your group or your field, but you have no excuse not to be the most knowledgeable person because the information is all out there.
1:14:08
What kind of what kind of things were you doing to learn about industries and and companies, you know, at in the beginning of your venture career that maybe you'd be using a deep research query to do today?
1:14:17
But >> well, the first thing I mean the first thing is you develop and I think this is all the great VCs in the valley.
1:14:23
You you develop this hyper FOMO uh of anything and everything.
1:14:29
And um one of the way one of the reasons I know that um that it's time for me to move on is I I haven't I haven't put together a claw about yet, but I know my older self would have done it immediately.
1:14:43
[laughter] And and it's just that kind of thing.
1:14:46
You can't sleep on not knowing something, you know, or hearing that there's a company you don't know about.
1:14:52
And um you developed that as an instinct like as a positive tool um to just be hyper paranoid about new companies, new things, new information, new technologies.
1:15:04
>> Is venture capital eating the world?
1:15:04
Are is venture capital scaling so much that it's eating into other asset classes? We're seeing mega funds.
1:15:10
I'm interested to think about what's durable about your approach to investing. What's additional? What's substitutive? How is venture changing?
1:15:20
I think from the minute I entered venture to to today, venture has gotten nothing but more competitive.
1:15:27
It's uh it it it as an asset class, it's gotten more and more competitive and people get more and more aggressive.
1:15:35
aggressive. Um we're in a very interesting time where people have grown funds to the size of of equivalent to the largest PE funds >> and um they're moving money um especially you know you just had the Collison on you know you look at the stripe or the data bricks case they're using those large funds to convince the
1:15:56
companies to stay private longer maybe forever that's just a very different world than the one that I grew up in um I think they turn around And the people that do those rounds turn around and tell the LPs, their their investors, look, if you want exposure to these growth years in these companies, you need to come through us. And so you
1:16:14
And so you they've, if I were using cynical words, I'd say they've hijacked the the the growth years of these early IPO companies.
1:16:23
You know, Amazon went public below a billion in market cap.
1:16:26
Like it's hard to fathom that, you know, today with with what we have going on here.
1:16:33
And that's >> what's the what's the solution though >> because because there's different there's different you know Angelist has been >> uh you know available and scaling for a long time now.
1:16:44
Robin Hood has their new >> Yeah, I know. I know.
1:16:46
The problem the problem with getting the retail investor into this crazy world of venture capital is most venture capitalists are well aware that in a fund of 10 investments, seven are going broke and bankrupt.
1:16:59
And I don't know that the retail investors got the right frame of mind for that type of activity.
1:17:06
I also um there's a reason that public companies have public audits and file these these financials in the way that they do.
1:17:14
And I can tell you when a company gets ready to go public, everyone sharpens their pencils.
1:17:20
The auditor, the lawyers, everyone really tightens up.
1:17:22
And I think every venture capitalist knows that that numbers that are in a PowerPoint may or may not be correct.
1:17:28
But I don't know that retail investors know that.
1:17:31
So I think it could I think it could be a dangerous world to go down that path you're talking about. >> Yeah.
1:17:38
>> Um but the I ideally the thing to do would just to make it a lot easier to be public, lower the cost of being public really scrutinize the cost of of DNO insurance and the lawsuits that come to the table because that makes people not want to be out there on the field.
1:17:52
It it it would require the SEC to stare themselves in the face and say, "Look, the number of public companies in the US is half of what it used to be."
1:18:02
And what is that a problem?
1:18:05
I think it is, but is that a problem?
1:18:07
And what are we going to do to fix it?
1:18:08
But it there's not an overnight fix.
1:18:10
It's going to take a it would take it would take someone being very determined to to make it happen.
1:18:15
Do you think that there's a world where the AI backlash is less if the big labs got out earlier?
1:18:24
out earlier? I'm just thinking about the average American can't get allocation in SpaceX anthropic open AI and they're seeing bills go up and they're worried about AI but they don't have exposure and and if they could at least see that
1:18:44
they're somewhat allocated to that >> in the same in the same way housing prices going up sucks until you buy a house and then >> the way you describe it sounds more like how a politician would describe But then I actually think it would might play. I
1:18:55
I don't know that there are that many retail investors out there going, "Oh, my job's under threat from AI.
1:19:01
I wish I could own anthropic."
1:19:03
Like I I mean, >> well, isn't that part of Isn't that part of why I mean this sort of fear-based uh fundraising approach that that the lab, you know, some of the labs have taken where if if somebody's telling you your job's going to go away, of course you want to give them as much money as you can as a as a hedge, >> you know.
1:19:23
I I don't look I there there's an interesting irony that if you wanted AI exposure, you're pretty good just owning the index.
1:19:30
Nvidia is such a large part of the index.
1:19:33
You have exposure to Microsoft and Google and Facebook.
1:19:35
Like I I don't know that you need to be in that place.
1:19:41
And we are now already at a place I would say, you know, every time there's a new technology wave, um people get rich quick.
1:19:48
When people get rich quick, speculators come in.
1:19:50
Charlton's, you know, those kind of things.
1:19:53
And eventually that leads to a bubble.
1:19:57
People are confused when they think, you know, they say, "Oh, you you say it's a bubble, you're anti-age."
1:20:00
No, the fact that it's real causes the bubble and that's why fools rush in.
1:20:05
I mean, the beginning of the gold rush, there was really gold there. They were finding it.
1:20:11
At the end, the point, >> you know, it got speculative and so funny.
1:20:15
>> It will get speculative.
1:20:15
I think it would be really ironic if we, you know, invite retail investors into a Goldman SPV of open hour an entropic right before the [laughter] re which I think would be the most likely thing that would happen. >> Sure. Sure.
1:20:31
Uh >> how how are you uh what are you thinking about around China as of today, February 2026?
1:20:41
>> Uh we have distilled Gate this week.
1:20:41
Uh a lot of people are talking about it.
1:20:44
Uh but what's on your mind?
1:20:48
>> Can I ask you a question about that?
1:20:50
This is this is uh remarkably naive on my part.
1:20:52
So these model companies are saying that their their API was hit 16 million times. Is that correct? >> Something like that.
1:21:00
I don't even know if a bunch of >> How how did that happen?
1:21:03
Are you not tracking who connects to >> Yeah.
1:21:07
You set up a whole bunch of different like front companies or you're reselling access.
1:21:11
So if you go to the uh if you go to the iTunes app store right now, there will be an app to set up 16 million accounts >> that or yeah or or if you just you can go to the app store right now and look for like chat AI and it will hit the other APIs but you're going through an American company maybe they don't have security.
1:21:28
So there's a lot of different ways to to exfiltrate data and then also a lot of data just hits the open web because you go to chatbt, you run a deeper research report and then you just publish it on your blog or or on the internet.
1:21:41
>> But now they've been able to to those things down and now >> I I you know I share the skepticism Elon does and it this goes way back to my speech at all in on regulatory capture.
1:21:55
I said then and I still believe now the biggest threat to the US um let's call it AI hedge me is is the Chinese open source models and the developers um even in the US that are working on their own are using those and you can see that on all the on all the the uh the tables that are out there and so um [clears throat] >> it is a highly competitive like just
1:22:24
globally competitive reality that that in an ecosystem where there's six to 10 open source models that can all learn off of each other that's going to be high like that's going to be a really incredible primordial soup if you will for innovation to evolve and I fear ma mainly because I'm well aware that like open I mean anthropic is the biggest spender on lobbying whatsoever. I always
1:22:46
I always fear when these things come out that they're um just trying to encourage more of that regulation.
1:22:54
And if that happens, I think it could be like if they try and make it illegal to use a model that has any Chinese um >> like ancestry.
1:23:04
Um I think that could end up in a really weird place.
1:23:06
Um and and and the place to really pay attention to and look out for is um who's going to serve the rest of the world.
1:23:14
In the internet era, there was a fence around China and the US companies serve the rest of the world.
1:23:20
If we if we get super heavy on US regulation, you may find there's a fence around the US and China serves rest of the world.
1:23:28
That's what I'd be worried about.
1:23:31
>> How are you thinking about great power competition more broadly?
1:23:33
Like I'm an American bald eagle.
1:23:35
as American as they come.
1:23:38
At the same time, I feel like uh I've been worried about a confrontation over Taiwan for years.
1:23:46
>> Things there's been trade wars.
1:23:46
Yes, and things are tense, but nothing's really happened.
1:23:50
Is China somehow like underrated in your mind?
1:23:52
Is the geopolitical risk overstated in some way?
1:23:56
Like what are you seeing that's not consensus?
1:23:59
If you've seen some of the stuff I've posted and and and and I think the stuff I'm posting is highly consistent with Elon's point of view, it it it comes from a place of if you're going to declare that that there's >> this this relationship that we need to optimize, I think, and and if your goal is to lower the risk of any any major glow up blow up between the two, I think it's imperative to have as much knowledge as possible.
1:24:27
And so one of the things that I don't like is when you see people out there spreading rhetoric that's just not consistent with the reality.
1:24:37
And so I'm I'm just like let's get eyes wide open first.
1:24:40
I also think that there are things we could learn from China about how to run infrastructure in the US.
1:24:47
They're clearly better at it than we are.
1:24:49
And if you just, you know, close your ears and say, "Oh my god, they're the evil competitor and they cheat all the time."
1:24:58
you don't ever get yourself in a position where you're going to learn, you know, from them maybe what they're doing well and what we're not.
1:25:03
And so I'm I'm, you know, Elon, we were I guess he was on Cheeky Pint with the gentleman you were just talking to, John. >> Yeah.
1:25:14
He he talks about um how competitive they are like, and I'm just like, let's be realistic.
1:25:19
Let's not I also worry a little bit that the venture community's gotten into all these military companies because I venture capitalists start to look like wararm mongers. Right.
1:25:30
It's ironic way back when when the all-in pod just got started, they were giving uh um oh, what's her name was on the Boeing board.
1:25:40
Nick uh Nikki Haley wasn't yet.
1:25:40
And they were like, "Oh, she's a wararmonger.
1:25:44
She's on, you know, looking after the defense company.
1:25:47
Now every VC is in Andrew.
1:25:49
They're doing the same thing. Let's be consistent." >> Yeah. Yeah. Yeah. Yeah.
1:25:52
Um I I are there any other industries that you do think are interesting that sort of butt outside of the tr the typical uh mandate of venture capital?
1:26:03
You know like AI fits very neatly into the software continuum internet cloud mobile.
1:26:08
Um I thought crypto was a little bit outside of the wheelhouse but a lot of VCs made it work.
1:26:16
industrial, energy, defense.
1:26:20
These are sort of things that are a little bit outside of the typical software business.
1:26:25
>> I would I'm I'm going to have to run, but I would tell you one thing.
1:26:28
>> Um, every time venture cap every time venture capital gets easy, people take or take risk with companies that are less of a great fit for the venture capital model.
1:26:40
And when I say a great fit, like they're they're either heavy capex or they have they have low gross margins.
1:26:47
They require tons of capital to keep surviving.
1:26:49
And and >> history's pretty good at at like bringing people back around to how hard those are to do with venture capital.
1:26:58
So, it's interesting for me to see those experiments being run.
1:27:00
um you know there there was near death with Tesla many times and it's a lot easier to get in those difficult situations when you're using debt and leverage which we're seeing all over these data centers and so um I just a word of warning be careful it ain't easy you know >> okay Jordy last question >> now we got to let uh let our guests jump but u congratulations hope everybody can get out and buy >> running down a Dream.
1:27:31
It's available everywhere. Books are sold. Go check it out.
1:27:35
And thank you so much for taking the time to come chat with us. We'll talk to you soon. Good luck. >> Goodbye.
1:27:40
>> Let me tell you about Sentry.
1:27:40
Sentry shows developers what's broken and helps them fix it fast.
1:27:44
That's why 150,000 organizations use it to keep their apps working.
1:27:49
And let me also tell you about Vanta.
1:27:51
Automate compliance and security.
1:27:53
Vanta is the leading AI trust management platform.
1:27:56
Uh we have some news from the public markets into it shares jump 5.
1:27:59
4% on packed with anthropic.
1:28:03
This is you know you like advanced talks you like talks you like advanced stock talks you like deals but packs are really the top tier deal making that you uh turn your deals into packs.
1:28:16
Accenture also turned positive up 1% during the anthropic event and docusine rises 5% after partnering with anthropic.
1:28:24
So, lots of lots of folks in the public markets in software that are facing pressure are going doing deals and announcing partnerships as opposed to uh I don't know competition, coopetition, what will it be ultimately, but it's a lot of stuff.
1:28:41
Anyway, >> uh Grace says, "Return flight from NYC gets canceled by snowstorm.
1:28:44
Call United connected with customer service. Rare voice is uncanny.
1:28:50
Deaf AI, but they gave it a human-like accent.
1:28:52
takes 20 minutes to get rebooked. Pretty good. I ask if it's AI. Haha. No, ma'am. But I get that a lot.
1:29:00
I ask it to calculate 228* 6,647. It runs the calculation. GG.
1:29:07
[laughter] >> Do you think this is real? >> Maybe. You got to test this.
1:29:13
>> That is uh >> I mean this is past the uncanny valley then.
1:29:16
Or it says voice is uncanny.
1:29:16
So it's >> also pretty easy for a human to just type this in.
1:29:22
>> That would be hilarious. Yeah.
1:29:22
If you, this is where the real alpha is.
1:29:23
If you have the chatbot open, if you're if you're uh, you know, on customer service calls, you need to be >> Yeah, maybe they're just using >> Cluey.
1:29:33
Maybe maybe Gear tickets has the real alpha guy who vibe codes a billion dollar SAS on a United Airlines customer service call.
1:29:42
They're just using them for their token. The tokens are free. >> Free comput. >> Free compute. Free comput.
1:29:46
Well, if you want a while ago, >> if you want uh if you want AI voices, head over to 11 Labs.
1:29:53
Build intelligent real-time conversational agents.
1:29:57
Reimagine human technology interaction with 11 Labs.
1:29:59
Um, continuing on, what is this hoodie? The Fred hoodie. Oh, this is amazing.
1:30:06
I [clears throat] love Fred.
1:30:09
So, Fred is the Federal Reserve.
1:30:09
What does FRED actually stand for? Fred St. Louis.
1:30:15
the what is >> Federal Reserve economic data. >> Yes.
1:30:18
So this is basically the best website for economic data.
1:30:20
Huge in my early economics career.
1:30:23
Uh so many useful charts and graphs all free open source just like you just click it and you get exactly what you want.
1:30:30
So whenever you want to go back to some ground truth setting uh you hit fred. saint louiswis.
1:30:35
gov or something like that. I think it's fred. saint louisfed. org is the website.
1:30:41
Highly recommend it for GDP data and more. Good charts. And here we go.
1:30:47
We got the Fred sweatshirt.
1:30:49
Absolute dripped out economic brother. Uh, what else?
1:30:53
>> How popular is the name Fred? >> Fred.
1:30:56
>> I feel like you look that up. Let me tell you Figma.
1:30:58
Ship the best version, not the first one.
1:31:01
With Figma, introducing clawed code to Figma. Explore Moyer options. Push ideas further.
1:31:06
With Figma, you can design your next hoodie in Figma potentially. Okay.
1:31:11
So, Fred in 1950 was the 84th most popular name.
1:31:17
>> And guess what it is now?
1:31:19
>> Uh, it it was 84 back then. I imagine it's fallen.
1:31:22
I would say it's like 150. >> How about 2007 56? Fred. Such a huge opportunity. >> Yeah.
1:31:31
Bring back your kid Fred. That's a great name. Strong name. >> Yeah.
1:31:35
All these things go in cycles though. It's the business cycle.
1:31:37
Um, there's news over at Meta.
1:31:39
Meta has shaken hands with AMD. They're forming a pact.
1:31:46
Today, we're announcing a multi-year agreement with AMD advanced micro devices to integrate their latest Instinct GPUs into our global infrastructure with a approximately 6 gawatt, give me the of planned data center capacity dedicated to this deployment.
1:32:02
We're scaling our compute capacity to accelerate the development of cutting edge AI models and deliver personal super intelligence to billions around the world.
1:32:10
Very exciting and pretty cool little hype video.
1:32:13
You see the camera move on this video.
1:32:15
This feels like you speed that up, you cut in some other stuff and you got yourself an Instagram reel, right?
1:32:21
You see this little like Have you seen those tutorials about like how to make a car? Yeah. SD kit on this.
1:32:28
I think it goes pretty hard.
1:32:28
You you double the speed, you add some flickering, you add some frames, frame interpolation, all sorts of stuff.
1:32:33
But Lisa Sue is on an absolute tear.
1:32:35
AMD's doing great stopping >> Meta.
1:32:41
>> Mark Zuckerberg's Meta is planning a stablecoin comeback in the second half of this year, eyeing a third party vendor as a key partner to power payments >> across Facebook, Instagram, and WhatsApp. >> Uh this is great.
1:32:52
if uh they should have something.
1:32:54
If your friend sends you a meme, it's good.
1:32:56
You should be able to tip them easily.
1:32:59
>> Tipping for great >> shares.
1:33:02
>> Well, if you want to build a social network built on tipping, you'll need Plaid because Plaid powers the app use to spend, say, borrow and invest.
1:33:10
Securely connecting bank accounts to move money, fight fraud, and improve lending now with AI.
1:33:13
Uh, Sean Frank, >> soon to be a new father and dear friend of the show, says, "Manis from Meta just doubling my ad budget every 15 [laughter] minutes.
1:33:25
>> This is the best new format." >> I like this. This is only possible.
1:33:27
This is This is the best AI image I've ever seen.
1:33:32
I think this is so funny because this definitely doesn't happen in the actual movie, but it's >> This is what Bur is like now. >> I know. I know.
1:33:38
Sean Frank really in a tear.
1:33:42
I saw him in the chat yesterday.
1:33:42
I forgot to say hello to him. Hello, Sean.
1:33:45
And also, congratulations on the new baby.
1:33:48
Uh, very excited for you.
1:33:48
Anyway, uh, this is hilarious image, but uh, we will return to the timeline after our next guest because we have Ivan from Notion in the Restream waiting room.
1:33:58
Welcome to the show, Ivan. How are you doing? >> Hello, guys. >> Good to see you.
1:34:01
Good to have you on the show. >> Long overdue. >> Long overdue. We're so excited.
1:34:07
>> Um, >> first time we've ever written. >> Fantastic.
1:34:08
Uh well, we're glad we caught you on today because there's a big launch in Notion World, but I'd love you to take us through it.
1:34:14
What was announced today?
1:34:18
>> We're launching customer agent today.
1:34:20
It's one of the first, if not the first, multiplayer agent product for knowledge work.
1:34:25
>> Oh, >> so it does real work for you in the background. Very easy to set up. Host in the cloud. Yeah.
1:34:30
>> Uh connect to all your work products.
1:34:32
And the best part is you don't need a Mac Mini. >> That's a good line.
1:34:35
They are going out of stock.
1:34:37
I don't know if you've seen um but the the Mac Mini is in short supply.
1:34:42
So uh walk me through some of like the most obvious use cases like notion is I I think of the amazing because you have what is essentially a document but also a spreadsheet and you can kind of move between different data structures and visualizations on top of data in a sort of consumer app uh UI is.
1:34:59
And so I could imagine uh creating a document and then having an agent go and do a bunch of work to populate extra fields.
1:35:08
So where are you seeing uh or where are you excited about these agents actually taking hold in the product?
1:35:16
>> So what you're describing was the notion probably two years ago. >> Yeah.
1:35:20
>> Two years ago like during the SAS era our strategy has been consolidating all different use cases into one product. Okay.
1:35:26
>> We about knowledge base you're talking about documents. >> Yeah.
1:35:30
>> Talk about project management. Yeah.
1:35:30
And we want to we have been bringing together into one tool that's very flexible. Okay.
1:35:34
Like for example um ramp actually the the companies that sponsor you guys. Yeah.
1:35:40
We bring ramp last year as a new customer for notion. Okay.
1:35:44
>> And we helped RAM consolidate half a dozen different tools that the core collaboration stack onto notion. Yeah.
1:35:51
So they don't have to pay as much money for all the tools.
1:35:52
That number one their team don't have to jump between those tools. That's number two.
1:35:56
I would say the best part is now they have one place to do their core collaboration work.
1:36:02
They have one place to deploy AI.
1:36:04
>> So now notion is the core agent orchestration layer for ramp.
1:36:08
>> Um >> the the product we just launched today customer agent ramp has been an early customer for us for a couple months. Yeah.
1:36:14
>> Um they're running all the enable sales enabled process a lot of internal bug triage all different process on this.
1:36:21
But because they have people have one place to work, collaboration, system, record, truth >> and one place to do their busy work, delegate busy work too. Yeah.
1:36:29
>> So emp model is what is it?
1:36:29
Uh money and time save both and this is we're doing for ramp at the moment. >> I love it.
1:36:37
Uh yeah, talk to me about uh the the agentic cron job that feels like something that we're starting to taste with open claw.
1:36:46
There's clearly demand for it.
1:36:49
uh it requires a little bit more upfront effort than just firing off a deep research report or saying, "Hey, hydrate this text.
1:36:55
Expand, contract, expand, turn it into bullet points, turn it into paragraphs by back and forth all day long."
1:37:01
Um, but I feel like the for most businesses having an agent that's effectively on a cron job, maybe you don't call it a cron job, but it's something that runs every day, that runs over a knowledge base, over a customer list, over documents, and and does the things that AI is great at every day.
1:37:21
That feels like something that could be incredibly powerful.
1:37:24
How are you thinking about longunning agents, cron job agents, scheduled agents?
1:37:31
Yeah, crown job is a pretty good word for it.
1:37:33
Like a lot of knowledge war is kind of just crown job. Yeah. Right.
1:37:35
So you update your uh pushing paper back and forth and crown job to from this person to the other person.
1:37:42
>> I think the the world sort of taste this power of when agent connect with the crown job through product like uh open clock.
1:37:49
It can do a lot of work for you and so you no longer have to prompt it.
1:37:53
It trigger work on the background autonomously asynchronously for you. Mhm.
1:37:58
>> Um our our interesting less about open cloud or Mac minis is what does this do for real business. >> Yeah.
1:38:05
>> And real business is you require enterprise grade permission.
1:38:06
It has to be multiplayer.
1:38:09
No longer just for a personal tinkerer with your own back meaning.
1:38:12
You have to power the entire teams with it right and it has to be easy to set up.
1:38:16
So you don't have to be an AI tinker or AI engineer to do it.
1:38:18
Uh you have to have the state-of-the-art models usually the day release.
1:38:22
So all the service will provide for businesses to take the spirit of bronop background agent clock you might say uh open call you might say into businesses that's the positioning of this product. >> Sure.
1:38:37
So uh talk to me about where the capability frontier on the agent side is today.
1:38:45
Uh I mean because agents can be turned really loose.
1:38:49
You can give them access to Python and they can talk to any API.
1:38:54
They can write their own CLIs at this point.
1:38:56
And so uh you mentioned like no Mac mini but is there a world where I tell like just for our example like uh I want a new notion document generated every day with a breakdown.
1:39:07
Uh I want you to go to a readonly access API for the YouTube API.
1:39:12
Pull all of our analytics pull all the chat feed synthesize all that and put together a notion document that I can review with the team in the morning that says oh this segment of the show was particularly great.
1:39:25
Here's how the analytics changed.
1:39:27
Here's where the viewer spikes were.
1:39:28
All of that that would require talking to an API.
1:39:31
What does that look like if there's not an off-the-shelf integration?
1:39:37
>> Um, all these is should be possible if it's not really possible.
1:39:40
Getting the YouTube API, getting the transcript.
1:39:42
I don't know everybody have access to it.
1:39:45
Gemini might have a special access to that, but assume video transcript.
1:39:46
Um, >> all this is possible because all you need to do is a runtime that can run model. >> Yeah.
1:39:55
All you do is a runtime that can talk to external APIs through code that written by models. >> Yeah.
1:40:02
>> And and uh a model that's does the crunch job periodically based on certain triggers. Yeah.
1:40:06
I just described those core ingredients, but they're the core basically the core ingredients for notion custom agents. Sure.
1:40:12
So you not only can do those can connect to your emails, connect connect to your Slack if you guys use Slack and send your message every morning.
1:40:20
So you don't actually have to come to notion to see the work been done.
1:40:23
You can stay where you are today. Okay.
1:40:25
>> How how how have you processed the last uh couple years of vibe coding because when I uh the first company I ever started or first not necessarily the first company but the first like real business we started on notion and at the time uh this company does like a bunch of um it's like a ad network on YouTube and so we had a bunch of different like ad buys happening.
1:40:47
ad buys happening. we needed to be managing that process with the client as well as the creators and so the entire company from the beginning ran on notion and I looked at every possible SAS solution at the time uh but I looked at all them I would have needed to a lot of
1:41:04
them didn't even for like you know work with customization so I just built all these dashboards that helped uh that helped kind of like manage all those different processes and that already had collaboration built in that already had like the account functionality. So it just like worked
1:41:19
So it just like worked completely out of the box.
1:41:20
So in some ways at that time I was already replacing like vertical specific software with notion.
1:41:27
And so in some ways like I feel like this whole process and explosion of people being able to create different applications for different use cases is kind of like just a continuum from notion's inception.
1:41:38
But >> yeah, we started we were never a lot of people think Notion is um document tool, collaboration tool, note-taking app, relational database tool.
1:41:49
That's never been the intent.
1:41:51
Notion started as a computing tool.
1:41:53
Like I really care about okay, I'm a programmer.
1:41:55
The power of computing is in the hands of you, the programmers.
1:41:59
How do we open up to more people?
1:42:01
That's why the company started.
1:42:03
>> So the spirit is always has been consolidating the the fragmentation of SAS for the past five plus years.
1:42:07
And it turns out that strategy works quite well with AI because once you consolidate those things, you have one contact to power the language models, right? That's one.
1:42:19
Number two, because we've been taking a stance that we don't want to inject our opinion how you should run your business.
1:42:24
You should we should just provide the Lego blocks and you can decide however you want to run those Lego blocks.
1:42:28
So we haven't been hardcode those business logic into our apps.
1:42:30
So and back then there's a bug what we call no code, right?
1:42:35
And in some people talk about SAS versus the language model.
1:42:38
A lot of SAS is hardcode the logic into your vertical apps and we don't do used to be a weakness of our product because it's how open-ended it is.
1:42:47
Has to be require some technical minded people to use it.
1:42:50
Turns out to be a strength because now language model can use those notion building block to do a lot of work for them.
1:42:56
So now we're with this new product we're launching is not just working with information in and out of notion.
1:43:02
It can power agent to work with external tools and do those cron job to do those repetitive knowledge work.
1:43:07
So do those busy work for the company.
1:43:09
Uh internally we call this like let AI do the night shift.
1:43:14
So you can do the day shift.
1:43:16
>> AI can do the night shift.
1:43:16
We to go a little bit dark mode this time because truly it's doing the night shift for us.
1:43:24
>> And nobody wants the night shift. I like the day shift.
1:43:26
I did the night shift back in college. Yeah.
1:43:30
>> It's not a it's not a fun. >> That's great. Uh yes.
1:43:31
So, so you mentioned Gemini, thank you.
1:43:33
Another another TVPN sponsor.
1:43:34
Uh, but I imagine that you're pretty model agnostic.
1:43:37
Uh, I'm interested to know how you're thinking about the different uh LLMs and then also how much do you want to surface to the user?
1:43:46
Like I was talking to Salesforce's Slackbot and it wasn't upfront with me about exactly which model was under the hood.
1:43:55
Now I'm a nerd and I'll ask okay is it 3. 5 or 4. 6 or 5.
1:44:01
5.2 too and I'll have some opinion whether or not that matters who knows but uh do you want to have model switchers model pickers do you want to be at that level of like empowering the user to pick the right tool for the job or do you want to handle that internally >> we do both so if you're like a normie or
1:44:20
normie plus+ using notion yeah you can just use notion without pick the model you be auto version right um but you have more sophisticated creative custom agent that do work triage work for you you different model have different strengths and weaknesses >> so you should be able to pick the model >> for our strategy and I think for a lot
1:44:39
of non-labs it's very important to be model agnostic >> the labs going to get better and better do model going to do more and more but one important strategic point is uh labs don't work well with other labs models >> so there's importance position to be the Switzerland of agents Switzerland of the models and that's the position we're
1:45:00
being With the product launch, you can work with clock code out of the box, can work with cursors agent out of the box and you pretty much can pick any models uh you want that's state-of- usually the day that those model are released and as a user of those product you don't have to worry about that. >> Uh I want to revisit this Wall Street
1:45:18
>> Uh I want to revisit this Wall Street Journal article that you were featured in uh back in August of last year.
1:45:22
So, uh, the quote was, uh, Ivan, the CEO of Notion, says that two years ago his business had margins of around 90% typical of cloud-based software companies.
1:45:35
Now, around 10 percentage points of that profit go to the AI companies that underpin Notion's latest offerings. How has that changed? Is it still 10%? Is it climbing? Is it falling?
1:45:48
What are your predictions for where that goes?
1:45:51
>> It's not as far as 10, but it's definitely meaningful amount.
1:45:53
definitely meaningful amount. Um before you you can do pure SAS margins now model people all our product are powered by AI now >> majority of product powered by AI now you have to uh model provider have to take some of the margin and we're okay with that um I think we see the market change on both fronts first we we want to use state of our most capable most
1:46:15
intelligent model because our customer wants that they want to ease the customer they don't have to worry about that and second there's a new wave of open-source foreign models are And that's why we have to be model agnostic and we can shift to different model for different type of work and that will help us with the margins and at the end of the day >> our customers haven't to worry about this. >> What we provide is less about model
1:46:36
>> What we provide is less about model capability has been there for almost for a year or two years to do a lot of knowledge work. Yeah.
1:46:42
What's missing in the market is this infrastructure layer that glue together model capability, glue together permissions and to provide real knowledge work for the customers at the same time backward compatible uh having a good UI for yeah company of sizes.
1:46:59
>> Uh I have one last question we'll let you go.
1:47:01
you go. Um, how are you thinking about sort of like it's crazy to call them legacy AI workflows because they were probably implemented like a year ago, but uh when I just think about like document summarization or even like spellchecking grammar like that was
1:47:17
probably moved to an LLM that was capable a GPT4 class model can do that at a very low cost and maybe you want to optimize that even further by going to an open-source model on commod oddity hardware really drive down the token cost. Have you left any AI workflows in
1:47:33
Have you left any AI workflows in place on legacy models or have you migrated everything to the frontier and you're just moving with the frontier?
1:47:46
>> We're moving with the frontier by and large because that's what customer want.
1:47:49
They want smarter things. >> Yeah.
1:47:51
>> But things like avoid dictation, summarization, legacy model can do that. >> Sure. Right.
1:47:57
>> And so it's just getting >> I would say the most important part is like the market is changing so fast right now like that nobody know where the future holds but we know the model capability gets better better. Yeah.
1:48:08
>> And we always care about building beautiful and powerful tools and AI is that tool today.
1:48:11
How do we make sure that our company can benefit from this?
1:48:16
You don't have to be Fortune 500 to for deploy engineers.
1:48:18
You don't have to be a San Francisco startup to have AI engineer on your team to use this. Right?
1:48:24
Every business can benefit this technology.
1:48:26
That's our ethos and that's why we're building this product to make it super simple.
1:48:30
You don't have to worry about Mac Mini, not worry about models.
1:48:35
[laughter] >> That's the tagline.
1:48:36
Is that on the homepage yet?
1:48:37
Don't don't worry about Mac Mini. We got you.
1:48:42
>> I think that's too specific calling out other products, but Night Shift is a good one. >> Night shift works. I love it.
1:48:46
Someone in the chat, John John Palmer in the chat was saying, "Well, but if I can't if I don't have to use a Mac Mini, what will I spend all my time setting up?" [laughter] >> Yeah. >> Like to tinker.
1:48:58
Tinker is sometimes more than 50% of the fun of it besides being productive. >> No, this is true. This is true.
1:49:03
People People want to tinker. They want to play.
1:49:07
They want to explore and have fun.
1:49:09
And uh it seems like it's a great time to be uh running Notion.
1:49:10
Like it's just a very exciting time.
1:49:13
There's so many new products you can build uh so much faster than ever before.
1:49:16
So, congrats on all the progress.
1:49:19
>> Yeah, congrats to the team on the launch. >> Thank you so much.
1:49:21
Great to finally have you on.
1:49:23
>> We'll talk to you soon. >> Cheers.
1:49:25
>> Let me tell you about Labelbox.
1:49:26
Reinforcement learning environments, voice, robotics, evals, and expert human data.
1:49:30
Labelbox is the data factory behind the world's leading AI teams.
1:49:32
And let me also tell you about the New York Stock Exchange.
1:49:36
Want to change the world?
1:49:39
Raise capital at the New York Stock Exchange.
1:49:41
I'm not going to leak the news, but we have an exciting guest lined up for our next NY show.
1:49:46
So, hit that subscribe button to be notified when we go live. >> Can't wait.
1:49:52
Uh, a senior US official toy told Reuters that Deepseek's new model, whose release is now imminent, has been trained using Nvidia Blackwell GPUs despite the export ban.
1:50:05
>> Uh, I am interested to see uh what this model is capable of.
1:50:10
could have made that happen, right?
1:50:12
Like it could literally be one black well per person in a suitcase smuggled along.
1:50:17
It could be uh one shipment diverged or diverted from uh going to one country and then they say, "Oh, send that shipping container over there instead."
1:50:27
It could be cloud like they could have found a cloud provider that they were able to sort of anonymize and have a front company for.
1:50:35
There's a whole bunch of different ways to get compute if you're willing to bend the rules or break the rules or, you know, potentially anger the US administration, but we will see how this goes.
1:50:47
Tyler, do you have a feeling for how deepseek has been doing because there was there was a there was a hype cycle around Deep Deepseek v3 something and it kind of came out and it landed with it didn't make a big splash.
1:51:05
I feel like we're going into a new hype cycle around like the next Deep Seek is going to be really good. Is this fake? Is this real?
1:51:12
How are you feeling about Deep Seek?
1:51:14
>> So, I think the last big model release was supposed to be this like massive massive release, then it turned out to be a thing where like instead of like I don't know the exact number, but it was supposed to be like V4 and it ended up being like V3.
1:51:24
1, it was like that kind of thing.
1:51:26
Um, and then also like >> so they botched the pre the pre-train most likely.
1:51:30
Is that what people think? >> Yes, maybe.
1:51:31
And then um yeah, also like on Chinese labs generally like right now you're hearing a lot about like uh it's like ZI and KI and that kind of things. >> Yeah. So it's very unclear.
1:51:40
I mean they're not very public about this stuff. >> Yeah.
1:51:44
>> But um but also I I think broadly just about the the distilled gate stuff. >> Yeah.
1:51:48
Um, I I think throughout this I I've been um like I think I've like updated towards like actually we can probably mostly ignore a lot of the Chinese labs because basically like the the whole reason that they're good like everyone's like, "Oh my gosh, Deep Seek is right on our tail."
1:52:06
That's they're going to catch up.
1:52:07
They're going to catch up.
1:52:09
>> The only word >> the only reason that that they've been on our tail is because >> No, [laughter] no, no.
1:52:17
That's the royal flesh, Tyler.
1:52:17
That's the best thing you can do on this movie.
1:52:20
You just dropped a bomb truth nuke. >> Truth nuke. >> Truth nuke.
1:52:22
Disregard China entirely. You heard it here first.
1:52:26
>> No, but I mean >> Yeah. No, it's a good point. >> Yeah.
1:52:29
The only reason that they're like, >> "Stop."
1:52:31
[laughter] >> That's so rude.
1:52:37
>> The only reason that the Chinese labs are are so close to US labs is because they're just training on the outputs, right?
1:52:42
Which is like, okay, sure, like, yeah, good job.
1:52:44
But like you're not I I would be extremely surprised if you actually see a breakthrough from a Chinese lab so far. >> Exactly.
1:52:51
The only thing we've seen is that yes, they can copy stuff >> permanently 3 months behind.
1:52:54
No, but I think people were freaking out because China went from like 10 years behind to one year behind to 3 months behind and they were like straight lines on log graphs.
1:53:05
They're going to be 10 years ahead of us next year.
1:53:08
I feel like you can still be very worried about uh you know regulatory capture and all these things but I think like >> I I I'm like I I think Anthropic will basically just figure out a way that they can >> you know increase security on the API totally so so will open AI and then we'll see and then we'll see if the Chinese labs keep uh keep up with the progress but you know >> Yeah. Yeah. Yeah.
1:53:27
I mean that is how there's so many other dynamics beyond just obtaining training data.
1:53:33
Like if you take Will Brown's point that the internet is producing more training data, you wind up in a situation where sure training data is commoditized, but what does it really take to scale up DeepSeek V5 to a place where it's having economic impact?
1:53:49
Well, you need a massive inference cluster. Do they have that?
1:53:52
How are they distributing this stuff?
1:53:54
>> I do still think it's very impressive that the models generally they've put out are like very small and still like very good. >> Yeah.
1:54:00
>> But I think on the frontier level, I I'm not super about them. Tyler Tyler called it. They're cooked.
1:54:05
Uh anyway, uh really quickly, uh from Anthropic, uh there's now a call sheet on will the Pentagon designate Anthropic a supply chain risk. It's sitting at 36. 8%.
1:54:20
Uh I believe that there's going to be a meeting between Pete Hegsath and Dario.
1:54:24
So that already happened.
1:54:24
Update on the meeting from Andrew Kern according to Axio's defense secretary.
1:54:27
According to Axio's defense secretary, Pete Hgsiff gave Daario until Friday night to give the military unfettered access to Claude >> or face the consequences which may even include invoking the Defense Production Act >> to force the training of a war claude.
1:54:45
>> Wait, so so that was not a joke? Warclaw is not a joke.
1:54:48
>> I don't think they would name it that, but it's kind of sounds like that's what Pete is. >> Claw of war. Like God of War. That's pretty good.
1:54:56
Uh anyway, we we have an >> we before we get to that the chat is sharing that payments processor Stripe expresses interest in PayPal >> scoop.
1:55:07
>> We miss had a missed opportunity [laughter] if you guys could have simply you could if you could have published that at 12 when they came on the show that would have been quite nice.
1:55:16
>> Publish it until >> payment processing firm Stripe is considering an acquisition of all or parts of PayPal. Mhm.
1:55:22
>> Stripe, which is privately held and is among the industry's most valuable companies, as you know, the deliberations are still early and there's no certainty they'll lead to a transaction. >> Yeah.
1:55:30
I mean, we we read the Wilmanitis post and we were like, oh, it could be could be a couple days, could be a couple years.
1:55:36
Seems like it might be closer to a couple days.
1:55:39
>> Well, PayPal is up 7% today. >> Interesting.
1:55:42
So, market people are excited.
1:55:44
Say, hey, you might be able to own some Stripe.
1:55:46
>> Well, tell me uh let me tell you about Phantom Cash.
1:55:49
fund your wallet without exchanges or middleman and spend with the Phantom Card.
1:55:52
And without further ado, we have Stephano from Inception Labs.
1:55:57
He's the founder and CEO. Welcome to the show. How are you doing? [music] >> Very good. Thanks for having me.
1:56:03
>> Thanks for hopping on.
1:56:03
First time on the show, so I'd love to have you kick it off with an introduction on yourself and the company. >> Of course. Yes, I'm Stephano.
1:56:10
I'm one of the founders and the CEO of Inception.
1:56:14
Uh before this, I was at Stanford in the CS department.
1:56:16
been doing research in generative AI for a long time.
1:56:20
>> I think my lab is mostly famous for having co-invented diffusion models back in 2019.
1:56:24
I was on the flash attention paper, DPO.
1:56:27
So, a bunch of things that are now widely used in production and >> these days I'm most excited about the diffusion language models.
1:56:34
That's what we're doing at Inception. >> Yes.
1:56:37
[clears throat] So, uh I first saw a diffusion language model demoed at Google IO, I believe.
1:56:42
Uh but uh tell us like when explain it like I'm five because when I think diffusion I think a bunch of fuzzy noise and then the and then the midjourney image gets higher and higher resolution.
1:56:55
Everyone's familiar with that and then they're familiar with like the token streaming next token prediction. Is it different?
1:57:02
Break it down at a very low level or high level >> that that's right.
1:57:05
Basically we've taken diffusion models which is the thing that works best for image and and video generation.
1:57:11
kind of course defined process where you iteratively refine your output until it looks good >> and we [clears throat] figure out a way to apply it to text and code generation. >> Okay.
1:57:22
>> And it kind of like works the same way.
1:57:24
You start with a rough guess of what the answer should be and then you refine it. >> Okay.
1:57:29
>> And crucially the difference is that the neural network is able to modify many tokens at the same time. Yeah. Yeah.
1:57:34
And so it's much much more efficient than the typical auto reggressive model where you generate left to right one token at a time and you're able to modify many tokens in parallel.
1:57:46
>> So if I'm if I'm thinking of like uh you know not maybe like a like a deep research report type response.
1:57:50
Uh I in my mind I can imagine a report you know saying like explain the history of the Roman Empire.
1:58:00
That's the that's the example I always use.
1:58:01
It's like it's going to have some structure to it and I'm going to imagine a blurry image with like a couple large headers and then the headers are going to get filled in.
1:58:09
Then the text is going to filled in.
1:58:11
Maybe there's some bullet points.
1:58:12
Maybe there's some some dates.
1:58:13
Maybe there's some charts and like all of this is going to come together.
1:58:16
But I'm thinking about it not sequentially but as a whole and then refining iteratively until I'm getting to instead of pixels I'm thinking of individual characters or are there tokens in the same way that might exist in an LLM?
1:58:29
What does that look like?
1:58:31
Yeah, that's the right intuition.
1:58:34
Uh so it's kind of like yeah course to find generation and in practice u you know it's learned by a neural network.
1:58:41
So it's not necessarily interpretable like it's not the kind of >> process I would go through where maybe I start with head with you know section headings and then I fill in the details.
1:58:50
It's all learned by a neural network and so it's not really interpretable. Uh but it's fast.
1:58:55
That's really the >> So is speed the main thing?
1:58:57
I mean, we we we we had uh the founder of chatjimmy.
1:59:01
ai on the show, Talis, and it seemed like he was able to bake down a traditional LLM uh Llama 38B onto silicon and it was spitting out 16,000 tokens per second.
1:59:14
Do you have a comp on speed or cost that you're targeting or do you see like a through line to like okay maybe if we're running on Nvidia chips and he's running on custom silicon he's going to be faster but then once we get to custom silicon we're going to be 10 times faster than that how how should I be thinking about the trade-offs here
1:59:32
>> yeah so the our benefit is purely at the algorithmic level like it's just a more parallel >> approach that is not memory bound it's it's flops bound right it's comput bound >> uh so you're able to hit the ceiling of the roof line and we are taking you know advantage of all the resources we can get access to on the GPU. >> Uh in practice what this means is that
1:59:52
>> Uh in practice what this means is that we can get to over a,000 tokens per second. >> Wow.
1:59:57
>> Uh on traditional Nvidia GPUs, Hopper, Blackwell. >> Yep.
2:00:01
>> Uh so we are not yet at the level you know of the 16,000 tokens that you can get if you were to actually you know implement the model on hardware but we're running on you know general purpose GPUs.
2:00:12
So we can scale up as much as we want.
2:00:14
It's just a matter of getting more GPUs and you know you can just run these models anywhere. We are on bedrock. We are on foundry.
2:00:21
So if you have your own GPUs you can provision your own capacity and you can run your model our models there.
2:00:27
>> So it's it's very very scalable.
2:00:27
It's fast and scalable and in principle yeah it can be compounded.
2:00:32
You know you have a 10x benefit from the software you have a 10x benefit from the hardware.
2:00:36
Those two things could be combined.
2:00:39
That's what uh what usea you know there's a lot of people out there using uh traditional language models today.
2:00:46
What are the kinds of use cases where you would tell somebody you should be switching over today or at least trying to uh start experimenting?
2:00:57
>> Yeah, we're seeing a lot of traction in latency sensitive applications of LLM like whenever there is like a tight loop where you need to interact with a developer or a customer.
2:01:06
So our models are being deployed in a bunch of IDEs.
2:01:11
So if you think about coding uh coding autocomplete, next edit, suggestions, refactoring, quick agentic loops, that's a very natural kind of like application where diffusion are already really really good voice agents.
2:01:22
Uh we have a number of partners and customers that are building really really good voice agents.
2:01:28
The latest models we announced today, Mercury 2 is a reasoning model.
2:01:33
So, but it's really really fast and so you can get the quality of a reasoning model with the latency budget that you need uh whenever you want to build a voice agent which is resonating really well uh with a bunch of early customers.
2:01:46
Um retrieval and search that's another space where we're seeing a bunch of applications being built on diffusion.
2:01:54
So if you think about uh query rewriting, reranking, summarization, that's another really really good use case for diffusional lumps.
2:02:03
>> Uh talk to us about uh distillgate, how you've been processing it.
2:02:07
Did have you worked on that?
2:02:09
Have have you >> It's a sign of success.
2:02:12
>> Uh had did you any any points in your career were you were you experimenting with this stuff?
2:02:17
Is this something that we kind of forced the Chinese market into spending a lot of resources on?
2:02:24
I mean it makes sense, right?
2:02:24
That that that's what's always going to happen.
2:02:27
I think uh the moment you put it out there, you know, you give API access to the world that that's going to happen and people are going to copy you.
2:02:37
>> I mean, we've been doing distillation uh in in the research community for a long time and so people have been experimenting and figuring out ways to do it in a in a sample efficient way.
2:02:45
So I'm not surprised that it that it's happening.
2:02:50
I think uh it's hard to know at what scale and honestly it from the numbers that they were circulating it seems like they are able to do it with very very few data points.
2:02:58
That was the most surprising thing to me.
2:03:00
Uh so you know it's it's very interesting scientifically that you can actually distill with with so few data points because it means that it's going to be very very hard to to protect any IP. >> Yeah.
2:03:14
>> You are opening the model up from a from an API point of view.
2:03:16
So the last question somewhat related to that uh I feel like when these models get distilled uh we see very strong benchmark performance and then some yet to be quantified and benchmarked quality sort of degrades and you hear people that actually try and put them into production saying like ah it just doesn't have the same like big model flavor that I'm getting from the big labs.
2:03:40
I don't know how real that is, but I'm wondering if you zoom out and you look at uh and you look at diffusion versus uh transformer-based LLMs, are you noticing any diff divergence in the benchmarks where you're maybe better at coding or less good at coding where the mental model that we're giving the computer is leading to surprising results? >> Yeah.
2:04:07
So, what we're seeing is that it's it's it's good at coding. It's good at editing.
2:04:11
One nice thing about not necessarily being left to right is that you can use context all around you.
2:04:15
So those use cases have emerged as being really really good for diffusion labs.
2:04:20
I think it's also a function of the training data that we use.
2:04:23
U you know we always liked coding.
2:04:26
We're all computer scientists and so that was like a a very natural kind of application area for us.
2:04:32
And so um I don't know how much of that depends on the training data that we used versus the model.
2:04:36
Um, but what's exciting is really just like the speed.
2:04:41
That that's the thing that that >> yeah, >> it's going to be hard to replicate.
2:04:45
>> I got I got a need for speed. I got a need for speed.
2:04:47
I'm super bullish on speed. I'm serious. I think it's amazing. I used uh 5.
2:04:50
3 Spark on Cerebrus and I was like, this is the future.
2:04:53
It's going to come to everything and it's going to be an important moment for people to realize that uh it's just a different product when you're interacting with something fast.
2:05:03
Uh, and I think we learned this from Amazon squeezing out milliseconds in uh, in web page loads and we're going to experience it in AI, too.
2:05:12
So, thank you for everything that you're doing to speed up AI.
2:05:15
Uh, we loved having you on the show.
2:05:18
So, have a great rest of your day.
2:05:19
>> Yeah, great to meet you.
2:05:20
>> We'll talk to you soon. >> Goodbye.
2:05:22
>> Let me tell you about Console.
2:05:22
Console builds AI agents that automates 70% of IT, HR, and finance support, giving employees instant resolution to access for access requests and password resets.
2:05:34
And let me also tell you about Railway.
2:05:36
Railway is the all-in-one intelligent cloud provider.
2:05:37
Use your favorite agent to deploy web apps, servers, databases, and more.
2:05:41
While Railway automatically takes care of scaling, monitoring, and security.
2:05:45
And without further ado, we have TVPN Royalty. What's going on? Great to see you, James. >> Hey, John. Hey, Jordy. How you doing? >> Doing great. Doing great. Good.
2:05:56
Uh, calling in from a cave. Are you fully snowed in? What's going on?
2:06:01
>> No, we're Yeah, we're in New York. The snow is melting.
2:06:03
We built a little mini studio in upstairs and uh yeah, it looks pretty professional. >> Great. I love it.
2:06:10
Uh >> professional as you go.
2:06:14
>> Tell us uh tell us the news and then there's a bunch of stuff we want to talk about.
2:06:17
Uh yeah, so I guess uh we're joining today announcing a $96 million series C investment at a $1 billion valuation led by Lightseed Venture Partners alongside Sequoia, Kleiner, Perkins, Avantic, Saga, and South Park. >> Amazing.
2:06:34
Uh break down everything that's happened since the last time you were on the show.
2:06:40
The the the space has been moving so quickly, so it feels like it's been two years even though it's probably been two months.
2:06:48
Yeah, I mean it's it's all moving.
2:06:50
Everything's moving obviously at 100 miles an hour, but it's it sounds a bit trit when I say this, but it really is a privilege to be building in such exciting times.
2:06:59
Um yeah, I mean we we've just launched Profound Agents, which I think is a really big deal.
2:07:05
It's, you know, we we serve the marketer.
2:07:07
know, we we serve the marketer. So you know the the the line we've been using during this fund raise is um or during this announcement has been you know Harvey lawyers have Harvey engineers have cursor and marketers have profound and I think that's that's truer than ever in that yeah with this launch of
2:07:23
agents it really takes profound uh towards being like a full stack you know holistic platform for the modern day marketer allowing them to you know not just understand how they show up in AI platforms like Chat, GPT, Gemini, the rest of them, but also build agents that can help them do more with less. Um, so
2:07:44
Um, so yeah, I think this is cool.
2:07:47
Yeah, we saw our customers had been, you know, we we we came out of the gates 18 months ago with Profound here in New York and what we saw was our customers quite often were taking our data and insights then going to orchestration and automation tools to do cool things with it.
2:08:02
So we've just brought that all inhouse now and uh yeah, it's really cool.
2:08:06
you can do everything in one platform.
2:08:10
>> How how are how do you think the the uh other platforms the LLMs are are evolving?
2:08:15
Some of them are launching ads, some aren't.
2:08:17
All products are being and services are being discovered in in all of them.
2:08:22
Why is it important to have a platform like profound?
2:08:24
We were talking about this off air this morning.
2:08:29
It feels like uh pe people have been joking around about Manis for example in the meta platform because like Manis is like an agent.
2:08:35
wants to help you, but at the same time, what helps what what [laughter] helps Manis is like spend more money, right?
2:08:40
So, it feels like having having a third party >> principal agent problem. >> Yeah.
2:08:45
>> Man is like, I've got a great idea.
2:08:47
[laughter] >> Uh you guys are more more more user aligned potentially.
2:08:51
But how how are you thinking about the interaction between profound and and the different platforms?
2:08:55
Yeah, I mean I think you know our prediction of the future is that in the future every company on the planet will care deeply about how AI talks about their brand or products or services.
2:09:07
Uh that's kind of a north star that we we hang our hat on.
2:09:10
And I think compared to you know search in the early 2000s or even for the last 25 years we it's looking like this will be a much more fragmented um sort of market.
2:09:22
I think we're going to see multiple players coming through.
2:09:26
So I think profound really sits adjacent to the the models uh or the labs.
2:09:29
Um and we help marketing teams understand how they show up in these platforms.
2:09:36
Uh you know when AI responds, what does it say about your brand?
2:09:41
What does it say about your services?
2:09:43
And now we help you build customized agents that can actually, you know, do the work with your with a marketer in the loop.
2:09:52
So, um, yeah, we've had hundreds of teams, you know, we work with, I mean, I guess a big thing that I'd say we're announcing since we last spoke, our series B is that we now work with 10% of the Fortune 500, which is a pretty cool start.
2:10:13
>> What are the other 90% doing?
2:10:15
>> That's fantastic news. >> Yeah. Ready to go. >> Yeah. Yeah. We got 90% to go. Job's not done.
2:10:21
done. Um but I think yeah we we work yeah it's it's it's a very cool start >> and >> you know what we're seeing more and more is that you know every brand is different every marketing team is different everyone has different initiatives everyone has different preferences marketing is is more human
2:10:37
than ever in a lot of ways >> and I think our approach of helping marketing teams build entirely customized agents that can take out the rote labor from their work is it's just saving giving these teams inordinate amounts of time and energy and it's it's very cool to see it work. Yeah, it's Yeah, it's exciting.
2:10:57
>> Can you walk me through the anatomy of correcting a mistake that exists across LLMs or even in a particular LLM?
2:11:05
My nightmare is, you know, you go to you go to chat GPT and ask how tall is John Kugan and it says 65 66.
2:11:13
This would this would destroy me. It must be 68.
2:11:17
Let's bake that into the pre-training data. 6868.
2:11:21
But uh but seriously, other than just like doing a bunch of SEO to correct the record, like how how can a company, if there's truly like a a consistent hallucination, something that's just incorrect for some reason, what is the process to actually change results? >> For sure.
2:11:40
I mean, well, the first step, which sounds kind of stupid, is just knowing >> why it's happening, right?
2:11:45
>> why it's happening, right? So you the model when when you know let's say an answer engine spits out an answer >> you know a good chunk of the time it's getting that answer from somewhere and you know being able to identify hey this is >> you know what we found I'll give you an
2:12:00
anecdotal example so it was I I won't be able to name the brand but it was a neo bank that >> the models were incorrectly spitting out that there there was no FDIC um insurance on and we identified there was coming from a few places it was like some a third party blog I think a couple of Reddit posts um and maybe like a YouTube video or something. So then once
2:12:23
So then once you know where it's happening it's kind of uh I wouldn't say it's easy but it's you know it's just kind of 101 marketing. Okay, cool.
2:12:30
Let's reach out to the blog and tell them that that's factually incorrect.
2:12:33
Let's comment on the Reddit post and say hey this is actually not true. We are FDIC insured.
2:12:39
let's produce a YouTube video that speaks to the same thing but you know mentions heavily that we we have FDIC insurance and lo and behold that gets pulled through into the model.
2:12:47
So >> So that would be something that your agent would do like automatically and you could just set set them off and do that. >> Correct. Yeah.
2:12:56
You can you could set up an agent that monitors for any misinformation based on a knowledge base of like ground truth and then say okay cool.
2:13:03
when we see any misinformation, let's generate an email that reads from our tone of voice and sends to this third party blog and says, "Hey, can you correct this?" for example.
2:13:12
>> So, yeah, but that it has to be customized cuz, you know, you wouldn't be a that's you'd never have that as an out ofthe-box solution, right?
2:13:17
It has to be everything's, you know, it's almost like being able to build one forone software.
2:13:25
This is this new paradigm of aic software, which is so so cool.
2:13:27
And obviously, I'm not the only one that's excited about it. >> Yeah.
2:13:31
Uh I mean you're in a very interesting uh uh vantage point in the industry because you work with so many Fortune 500 companies.
2:13:38
What are your expectations for Agentic Commerce this year?
2:13:43
We were just talking to the Collisons.
2:13:45
They it feels like it's on the precipice.
2:13:48
Everyone we've talked to is extremely bullish.
2:13:50
But I'm always interested to hear like the shape of the bullishness.
2:13:54
what what you think needs to happen to actually get people shopping agentically this year?
2:14:03
>> I mean, I think so much of that inflection is going to come from the models themselves or the consumer products.
2:14:10
So, you know, chat GBT and Gemini's ability to actually offer a fantastic user experience.
2:14:15
I think that that will be the kind of, you know, that that's the most important thing.
2:14:21
Uh I think from the other side of the fence working with these brands, these marketing teams um they're as a hot take they're not and I really I hope this doesn't sound like disrespectful at all but they're they're not as slow as you'd imagine like these giant we work with giant brands uh Fortune 500, you know, some Fortune 10 and they're ferociously fast.
2:14:43
they they understand the magnitude of this platform shift and it's very sophisticated teams.
2:14:50
A lot of them are SEO teams who I actually think are fantastically well suited to kind of attack this problem space because they're like kind of technical and they understand the sort of primitives of marketing and content etc.
2:15:01
They're quite crossunctional.
2:15:03
Um but yeah, I I think the a mistake would be to to think that the enterprise is super slow.
2:15:11
I don't think that's true.
2:15:14
I'm not just there's no sync there.
2:15:15
I'm actually I actually believe that. >> Yeah. No, I can.
2:15:18
>> Well, I asked uh I asked Chad GBT what's the best uh geo tool for startups.
2:15:21
It says profound >> what it does tracks how your [clears throat] brand appears inside LM.
2:15:31
It's best for VC serious about AI distribution. So >> dog fooding.
2:15:34
I mean we put a lot of we put a lot of profound in the pre-training data last year. It was great partner.
2:15:39
[laughter] >> Did it mention agents though?
2:15:40
That's the question, right?
2:15:42
We only know agents today. Okay, Julie, ask it.
2:15:43
What did they launch today? >> Oh, there you go.
2:15:46
Uh, while he does that, tell me uh what you think of the word geo. Is that too buzzwordy? Do you like that term?
2:15:56
What are what are the pros and cons of having a term applied to your industry, your nent business plan?
2:16:06
>> I think go sucks, okay, >> as a as an acronym.
2:16:08
Uh, it's just bad in so many ways.
2:16:11
so many ways. I think you know it can't be claimed by because of geography it is already taken >> um >> oh yeah >> it stands for yeah generative engine optimization which I don't people don't refer to these products as have you ever heard anyone refer to chat GBC as a generative engine
2:16:28
>> they don't yeah you're right they do like Google is a search engine for sure Bing is a search engine but no one calls >> what's your preferred what's your preferred acronym >> I mean so without you know, said sort of with with not much passion, answer engine optimization feels more fitting to me. Um, I I I think it is still all
2:16:46
Um, I I I think it is still all to be determined.
2:16:49
I think how your brand is spoken about by AI >> Mhm.
2:16:56
>> will become the most important primitive in marketing.
2:16:59
So, I think it it it's going to become bigger than just a kind of uh you know, something that you you put a label on like that.
2:17:06
Uh I think we we see a new a sort of new type of marketer forming over time which is interesting.
2:17:12
The marketing engineer um you know the the an a marketer who has the technical chops to be able to go in and build agents, customize agents, deploy agents um for the rest, you know, cross functioning AC, you know, across teams and yeah, I think that's very interesting.
2:17:31
We we announced uh profound university today.
2:17:33
So does this work if I press that? Can you can you see? There we go. That's elite. >> THAT'S AN ELITE.
2:17:40
>> WOW, THAT looks >> glasses in series C. >> Oh yeah.
2:17:44
[laughter] >> How much you [clears throat] guys? >> Yeah.
2:17:47
So, we announced Profound University. >> Cool. >> Today. Very cool. >> Um, >> which is Yeah.
2:17:51
It's actually really It's really awesome.
2:17:54
It's a series of certifications, training cohorts, um, learning uh, materials that essentially enables the this new era of the marketing engineer. Sure.
2:18:06
>> And yeah, we're we're very excited about that.
2:18:09
So, um, yeah, I think that we're going to see a lot changing in the world of marketing, as I guess is true in most >> Chad is saying that AO is a good AO. >> AOE.
2:18:21
[laughter] >> Well, thank you so much for coming on the show.
2:18:23
Always good to have you here, James. Congratulations.
2:18:27
>> It's great to see >> and we'll talk to you soon. >> Thanks, James.
2:18:30
>> Have a good rest of your day.
2:18:32
>> Let me tell you about Cognition.
2:18:32
They are the makers of Devon the AI software engineer.
2:18:37
Crush your backlog with your personal AI engineering team.
2:18:39
And we have Scott Woo from Cognition in the Restream waiting room.
2:18:43
I want to talk about the launch today.
2:18:45
I want to talk about AI progress.
2:18:47
I want to talk about math and your predictions on the IMO gold medal and everything that's happening there.
2:18:53
Um but let's start with uh the the the general update on cognition.
2:18:59
What's the shape of the business today?
2:19:01
And then I want to hear about the latest launch. Awesome. Yeah. What's up, guys? How's it going?
2:19:04
Great to great to see you.
2:19:05
It's been a little bit I feel like too long.
2:19:07
Too long you guys have been cooking. >> Yeah.
2:19:10
U every every month feels like uh feels like a decade now in AI. So, [laughter] um cool.
2:19:14
No, so so things have been great.
2:19:16
I mean, the business has grown a lot.
2:19:18
You know, we shared some of our metrics today.
2:19:19
One of which is that um our our total enterprise usage has actually more than doubled in the last 6 weeks even.
2:19:25
[laughter] >> And a lot of that has just been been mass takeoff of agents.
2:19:30
You know, I think the high level that we that we're really seeing is that as agents get more capable and you can trust them to do endto-end tasks, what you really need is the the full background cloud agent, right?
2:19:43
And so that means, you know, being able to run your your repos and everything locally, being able to test uh being able to spin things up from, you know, Slack or Linear or GitHub or Jira or whatever it is, and just being able to have this mass parallel, uh, async workload. >> Okay.
2:19:58
Uh, and then and then the announcement today Yeah. Yeah.
2:20:01
No, the announcement today was a was a fun one for us, it was a, you know, very near and dear to my heart, but but a lot of it, honestly, if if I were really to just describe it in one line is just >> clearing through all the frictions that that we've known about and and just making it a really great experience.
2:20:15
And so, um, you know, one of the big highlights is is automated testing and having Devon run your web app for you and send you the changes and send you screen caps of all of those things.
2:20:24
Um, but but there's tons of little things that that that really affect the experience.
2:20:29
And so, you know, making the the VM startup time way faster, uh, making the Slack integration way smoother, you know, showing you all of the intermediate progress of the messages and so on.
2:20:39
And so, it's been a um I mean, it's it's changed our internal usage a lot and so that's why we're pretty excited to get this one out.
2:20:46
>> How so it seems like there's there's speed to be squeezed out from uh VM spin up time optimizations.
2:20:52
We're also seeing some, you know, incredible progress on the custom silicon.
2:20:58
We had the founder of Talos on generating 16,000 tokens a second.
2:21:04
That seems like that will be really impactful when it rolls out to the broader uh code generation and software engineering world.
2:21:11
It's still uh pretty early with that company Llama 3B at this or 3 uh 88 8B at this point.
2:21:20
Um but where else are you seeing opportunities for speed?
2:21:23
How do you think about the importance of speed uh for what you do? Yeah.
2:21:29
No, there there's a ton that you can do and at some point a lot of it actually is just good old software engineering.
2:21:34
Um, and so so you know it's it's it's it's of course like you know the the models obviously you know you can improve the tokens per second.
2:21:40
You can you can improve the TTFT.
2:21:42
I think those improvements will be great and we've already seen a lot of those over the last bit. We'll see many more.
2:21:47
Uh but at some point you know your agent has to go install you know npm install.
2:21:53
Your agent has to go UV install.
2:21:53
Your agent has to go GP for things right?
2:21:55
uh it has to go pull up the front end itself.
2:21:59
A lot of that stuff is um it's good old product building and software engineering to to make that better and more efficient.
2:22:06
And so a lot of these obviously you know you can do algorithmic tricks.
2:22:08
You can put in you know indices right and indexes and and make those faster.
2:22:12
Uh you can do little things uh to to kind of like cheat the loading time and and do things in parallel and do things async.
2:22:19
But a lot of it is just building the systems around the agent to make it really fast.
2:22:23
>> No, it's a really good point.
2:22:23
I mean, anyone who's installed Open Claw has experienced like, oh wait, I'm actually just waiting to download software because it's pulling a whole bunch of stuff together and it's not actually doing that much waiting with the LLM, at least in the setup phase.
2:22:37
Um, but you still have to actually get this thing configured and I think a lot of people in tech went through that.
2:22:42
How have you been processing lessons from open claw interaction patterns that you think are interesting?
2:22:49
what it means that society more broadly is just aware of AI agents which I feel like is a term that you basically coined years ago uh and have have been running with but in a specific like enterprise context and now I'm at a bar and I'll hear somebody talking about AI agents and it's because of OpenClaw [laughter] and and I feel like oh that's a I remember the Scott Woo launch video where he explained that this was going to happen.
2:23:15
Um but but how have you been processing OpenClaw?
2:23:17
What is interesting about that?
2:23:19
are are there any like lessons from that open source community that project generally that that paradigm that you want to bring to Devon? >> Yeah.
2:23:27
No, I mean a lot of big changes and I think by the way I think OpenCloud gets a lot of credit for uh for for many people being the first time that people really saw Yeah.
2:23:36
>> uh what what a full you know agent would look like with access to your files, access to your computer and so on.
2:23:39
access to your computer and so on. I I think we're really getting to the point uh you know to to your previous point where I I think we're we're really starting to switch over from the early adopter cycle to the the kind of mass market cycle is my sense and and and the concrete impact of that is a lot more
2:23:56
people are starting to hear about and really think about AI agents right and I think it used to be I mean for us for example a year and a half ago you used to go into the room and explain to people what an AI agent was and why this wasn't you know why why this was different from from just like normal autocomplete or chatgpt or something like that. Now everybody's thinking
2:24:13
Now everybody's thinking about this stuff.
2:24:15
Everyone wants to use it.
2:24:16
And I think one of the the kind of implications of that is just accessibility and getting people to value as soon as possible is is one of the most powerful things that you can have in your own products as a result.
2:24:28
>> You guys have had a ton of success in enterprise.
2:24:31
The chat wants us to ask for your take on the SAS apocalypse.
2:24:34
I imagine uh uh some of the conversations that you're having with let's say the CTO of a of a a massive uh uh company >> uh are they thinking about using a Devon for things like you know big database migrations like how how are they thinking about um how agents can can impact their dependency on on sort of these like legacy tools and systems of record?
2:25:02
>> Yeah, I mean there's the whole catrini report and everything.
2:25:03
I mean, it is honestly ridiculous.
2:25:04
That that's that's my that's my two cents on it.
2:25:08
I I think that like um look at at a high level, of course, yeah, AI is going to change a lot of stuff.
2:25:15
I I I don't really understand how you go from that to saying that there's going to, you know, like take software as a good example.
2:25:24
Software is one of the most deflationary things ever.
2:25:26
you know, a lot of the same products that that you know, used to cost much more 10, 20 years ago got have gotten much much cheaper over time, right?
2:25:34
Has this been terrible for software company?
2:25:36
You know, I mean, it it seems like it's been pretty good.
2:25:37
All [laughter] the big companies in the world are still software companies, right?
2:25:41
Um and and so I think there's like >> um there there's one thing when prices go down because, you know, the demand's just not there anymore.
2:25:48
And obviously you can get into weird cycles and all that can happen.
2:25:51
But it's a totally different thing if prices go down because we've just gotten way better at supplying things and that's when you get Jeban's paradox and that's when you get, you know, just mass consumer surplus and so on.
2:26:02
Um and so at a at a high level I know I mean I think there's all the customers that we work with, you know, banks and health insurers and private equity and and so on.
2:26:10
I mean the they're um they're there obviously like a lot of these base migration modernization projects that they can go and take on immediately.
2:26:18
that the very next thing that they say is then like, okay, how do I pull the rest of my road map forward, right?
2:26:23
How do I build even more and get even more out to people.
2:26:24
Um, >> and and I think in reality, we just we all just have so much more software to build. >> Yeah.
2:26:31
How are you thinking about AI progress broadly?
2:26:33
It feels like a lot of people are feeling that recursive development is on the on the horizon.
2:26:41
People are bringing up takeoff speeds again, migrating from slow, maybe fast, I was backing off, now it's a little quicker.
2:26:48
Um, how are you uh how are you trying to like zoom out, reset, get to reality, figure out how fast things are actually moving? >> Yeah.
2:26:58
No, I mean the the the meter report shows like the consistent doublings and everything.
2:27:02
I I mean I think it's a very um I I think things are continuing on the exponential curve.
2:27:09
I wouldn't say that they're going either super exponential or subexponential.
2:27:11
I think they're they're roughly going on that exponential curve, but you know, exponential curve is a lot. >> Yeah.
2:27:17
like that's that's a very fast growth obviously.
2:27:19
Um I I think for us um you know one one of the things that's been pretty interesting is just like noticing each of the step function changes that happen.
2:27:27
And so for us for example it's definitely been in the last I'll call it like four or five months >> where something interesting happened which is we stopped typing code you know like at some point you you just don't right like like before obviously you have all the tools and you have the combination of things and so on.
2:27:43
Now, it's there's different experiences.
2:27:47
There's different tools that you want to have.
2:27:49
Obviously, between the IDE and the CLI and and and the web agent and so on, but either way, you're you're really just working in prompts and you're not really, you know, like the code that we check into GitHub, like how much of it was typed by a human at this point? I think almost none. >> Yeah.
2:28:02
Um and and maybe one of the things I would just call out is that that you know a as you kind of expose each new thing like I mean if you think of it as like a a profiler you know on on your own software engineering workflow like what is the most expensive part you you shrink that down you get to the next thing you shrink that down and you just make the whole cycle more effective.
2:28:21
We're at the point where a lot of these other things like you know understanding the code base and review and so on are the actual bottlenecks right testing is another big one.
2:28:30
Um, and I think what we're going to see over the next next little bit is um, you're basically going to have to solve each of those with really good product experiences, really good model capabilities and so on.
2:28:41
So may maybe the only thing that I would say, you know, I think the exponential curve continues.
2:28:45
I I would just kind of call out that the form factor looks very different as you continue on that exponential curve because you're actually solving different problems.
2:28:53
Like yes, I think we will continue to to to kind of like, you know, get the doublings and the doublings, but but now it looks a lot more like how do we optimize testing and review and planning, not how do we make the AI good at writing code based on the prompt that you give because at this point it's actually frankly it's it's basically already done.
2:29:10
>> Yeah, I have a bunch more questions. Uh I'll be quick. I have two.
2:29:13
Um first, uh what does the future of windfur look like in a world where you're not writing code?
2:29:20
Does that become a Kindle?
2:29:20
code? Does that become a Kindle? I mean that's a joke but does it become does it become more important for that product to be the best way to read code because even if you're not writing code you still like I've I've done terminal prompts and I'm like and then I wind up
2:29:37
opening the files to kind of take a peek in them and I'm like ah I kind of like there there's maybe room for innovation there and like how your code reading skill improves as your code writing skill uh degrades but how do you think about the future of >> windsurf Yeah, for sure. Um, I think the high
2:29:52
Um, I think the high level here is it's going to be a gradual thing, but I think over the next one or two years, we'll have a pretty broad transition towards what you might call having English as the source of truth. >> Sure.
2:30:04
>> Um, and so so people talk about like basically I think we'll go from code to English in the same way that we went from assembly to code. >> Yeah. >> Right.
2:30:10
>> Right. and and so so you know one of those steps has been been you know has been mostly done at this point which is the step of figuring out how do you uh prompt in English and then have the agent produce the code but if you think about it I mean you're still you're still you know reviewing code you're
2:30:25
still checking code into GitHub you're still reading the code to understand what's going on and I think at some point you actually want an interface that looks a lot more like a spec or a map or you know like a design doc right and that's the thing that you're iterating on that's the thing that you're reviewing for example, right? Like at some point review I mean people
2:30:41
Like at some point review I mean people say, "Oh, like review is going to go away because AI is going to catch all the bugs."
2:30:46
I think that's actually not right because what you're going to be reviewing is the decisions, right?
2:30:49
It's like here's what we're here's what the product is doing in this case and here's what the product is doing in that case and you know here's how you know this plan works or or whatever it is, right?
2:30:58
Um and so what you'll want to have is a very clean interface to interact basically, you know, with with your product and with your own specs and so on.
2:31:08
And and that's what a lot of what we think winds surf evolves into over time, right?
2:31:12
And so so again, I think it's a very gradual thing.
2:31:14
I think [laughter] I think there's a lot of value in reading code now and certainly at cognition we still do a lot of reading the code even if we are not the ones like you know writing the next line of code because instead we we write the English prompt for that.
2:31:28
Um but but but I think what happens with Windsurf is you know at some point instead of looking at each of the files you start looking more at you know this the specs and the highle logical design of what you're building.
2:31:39
You start looking at the you know the diagrams of your app or your your website itself and you're able to go and manipulate those and you're really just managing your agents that you kick off from there. >> Okay.
2:31:49
We blew past the IOI gold medal as you predicted correctly. Amazing.
2:31:53
Uh I have a follow-up question about that, but it's sort of in three parts.
2:31:59
One is uh what is the next like math or physics-based benchmark that you're excited about AI potentially unlocking?
2:32:09
Uh when do you think that might happen?
2:32:12
And then do you think there will be any tangible impacts of that?
2:32:14
Because if I walk down the street and I tell some random person like they they did it, Navier Stokes is solved.
2:32:21
I think that's the math problem that everyone talks about a lot. I don't even know.
2:32:25
Um I I think most people would be like great like is that going to help me with my job like they're more excited about just knowledge retrieval right now.
2:32:32
Um so so yeah uh the the the next hurdle timeline and then impact. >> Yeah. Yeah. For sure.
2:32:39
So I mean we actually I would say crossed a pretty exciting hurdle just recently.
2:32:42
um like Alex Lupaska and some of the folks at OpenAI um had a pretty important breakthrough in physics where they used uh language models to figure out a lot of the the the key lemas and theorems for it.
2:32:55
Um and so so you know I I would have said I think the next big breakthrough is is getting to a point where like actual science and actual discovery is happening largely powered by AI and I think we're we're effectively getting into that.
2:33:08
I think we'll see much more of that this year.
2:33:09
I think to your point on impact, yeah, [clears throat] I think it'll be some time until, you know, the the the average person feels the impact of us proving new theorems, but obviously the long term of all of this is extremely powerful, right?
2:33:22
I mean, we're going to be discovering new medicines.
2:33:23
We're going to be, you know, unlocking big breakthroughs in biology, material science, nutrition, um, and so on and so on.
2:33:31
And all of this is, you know, comes from a lot of the same science.
2:33:32
I I think I very much think of it as a >> um >> as as as you can call it like a you know a proof of concept or or or like an existence proof that it is possible.
2:33:43
Um and and you know solving some of these very difficult novel math and physics and algorithms problems >> is there are lots of ways that over time that itself will continue to be valuable but but even more so than that it's obviously just >> you know an existence proof that that AI can do some pretty incredible things. I love it. Oh, yeah.
2:34:03
A lot of people get abstract with the medicine science.
2:34:05
I like the material science one because I can imagine >> a much stronger, much cheaper, much lighter carbon fiber and driving a car that's pure carbon fiber for the same price as a Model 3.
2:34:18
This is pretty attractive. That's pretty tangible.
2:34:20
I think the average American [laughter] consumer is going to get behind that. >> Excited about that. >> Get extreme.
2:34:25
I'm pretty excited about the part of, you know, you you have the best pizza that you've ever tasted and except it's also the most nutritious thing for you because we've just solved taste and nutrition and everything.
2:34:33
And I feel like AI will get us there, but that might require a few more.
2:34:36
There'll probably there's probably a few more steps in the middle.
2:34:39
>> That's that's the new AGI benchmark. >> Get the goalpost. Get the goalpost.
2:34:44
>> I'm moving the [laughter] AGI will be here when I can have a pizza that tastes amazing and also is fully nutritious. Thank you, Scott. Woo. Have a great day.
2:34:58
>> Great to see you guys.
2:34:59
>> Always fun to move the goal post with you. We'll talk to you soon. >> Goodbye.
2:35:03
>> Let me tell you about MongoDB.
2:35:06
>> What's the only thing faster than the AI market, your business on MongoDB?
2:35:10
>> Don't just build AI, own the data platform that powers it.
2:35:12
And without further ado, we will begin our Lambda Lightning round with Rune. >> Look at this new.
2:35:21
Oh yeah, we're getting new effects going. Welcome to the show. Ooh, look at this. >> What's happening?
2:35:28
>> That is a beautiful lighting setup. Thank you for joining. First time on the show.
2:35:31
Please introduce yourself and the company.
2:35:34
>> Yeah, great to good to be here. I'm Bruno Fist.
2:35:36
I'm co-founder and CEO of the artificial intelligence underwriting company. >> Okay.
2:35:41
>> Our mission is to underwrite super intelligence and we do that by building standards and insurance products for AI agents. >> Okay.
2:35:48
Uh sounds extremely straightforward and simple. >> [laughter] >> Yeah. >> Yeah.
2:35:53
Plenty of plenty of data to build this on. I mean, yeah.
2:35:55
How do you even think the big so the big thing uh Derek Thompson uh was kind of summing up the whole discourse around Catrini >> and uh his takeaway was that everyone can agree that no one knows what's going to happen.
2:36:10
Uh so very difficult difficult environment to be you know creating insurance products for but I'm sure you're narrowing it down to some key initial use cases.
2:36:20
So maybe you can talk about where this starts. >> Yeah. >> Yeah.
2:36:26
Maybe the first thing to say is that regardless of whether anyone buys an insurance product, someone is always underwriting it. >> Mhm.
2:36:33
>> So otherwise it's just going to be the say head of risk at JP Morgan who has to make a go no-go decision.
2:36:38
>> He also sits with the same problem.
2:36:38
Is this going to work or is it not going to work?
2:36:42
Uh so the place we start is just what are the risks that are slowing down adoption today? >> Mh.
2:36:49
>> And can an independent third party with skin in the game and visibility across a bunch of companies be able to underwrite that better than any particular head of risk chief security officer might be able to do. Mhm.
2:37:01
>> Um, and like any other risk, when there's no data, there's an initial R&D phase where we don't expect all of these policies to work out well.
2:37:09
We expect to lose some money and in the process start to be able to collect the data that allows us to underwrite this more precisely than anyone else. >> Yeah.
2:37:18
uh walk us through some of the some of the example insurance policies because I mean everyone who's followed like the AI story and AI race has seen like a million different varieties of impairment from like the training run didn't work or the data center was delayed and that has a financial impact down to we got sued because of our training data or someone used our app and didn't like it.
2:37:44
There's a million different ways that you can have smaller even large settlements or lawsuits, but what how do you think about fragmenting the market, finding a uh a landing zone, a beach head? >> Yeah, totally.
2:37:59
So, you start from what are the very real concerns for slowdown adoption today. Let's take one.
2:38:03
We just announced uh the world's first insurance policy for any agent last week with 11 Labs.
2:38:09
Uh they are trying to be on the frontier.
2:38:13
uh their pioneers of security and and safety.
2:38:16
They're trying to be on the frontier of giving asurances.
2:38:17
The things that hold up adoption for them are things like hallucinations that lead to financial losses.
2:38:24
So everyone has seen a kind of Air Canada example lead to financial damage.
2:38:29
>> Data leakage uh continues to happen.
2:38:32
You'll see on a weekly basis open claw is the latest uh group of that.
2:38:36
>> You don't want your agents to give medical advice. >> Sure.
2:38:40
Uh and so those are also some of the kinds of things that are covered.
2:38:42
So mostly at the application layer today and then we think as insure appetite grows eventually our mission is to underrite super intelligence.
2:38:51
Eventually we think some of the kind of risks that look a little bit more like private nuclear energy will also have to be covered by insurance because these risks cannot sit with no one.
2:39:01
Uh there was a grand compromise uh in 1954 that allowed us to do private nuclear energy in America which is the price Anderson act.
2:39:10
speak to the government saying, "Hey, we really want some private nuclear energy. That'd be awesome."
2:39:13
But also, any particular private company cannot carry the risk if something truly goes wrong.
2:39:17
So, we're going to require an insurance scheme that's going to be our way of putting the market to work to manage this in a way that's just progress pro getting this adopted.
2:39:26
>> And the government has always effectively been the insurer of last resort in some ways, right?
2:39:33
>> Whether it's formal or not, the government is always the last resort.
2:39:36
Take co who's on the hook for that?
2:39:38
Ultimately the government has to step in.
2:39:39
So the question is can you formalize that a little bit more and say at what limits of liability is the government on the hook and up until that who's on the hook for that? >> Got it. Okay.
2:39:47
So uh walk us through through the chain of uh how insurance actually works.
2:39:52
I understand 11 Labs comes to you and then are you drafting a policy with a specific risk profile, payment pro uh uh premiums and then you're going out to the JP Morgans of the world and having them buy that and or invest that. Does this float? Is this tradable?
2:40:09
Can a retail investor get allocation?
2:40:12
How does that work on the on the long tail of the financialization?
2:40:16
>> Yeah, eventually this will end up on Robin Hood.
2:40:17
But [laughter] let me walk you through how it looks today.
2:40:19
H so today there are two steps high level.
2:40:23
First is uh certifying against the standard >> as a way to unlock insurance.
2:40:26
So >> historically the way every market has been unlocked is that the insurers want to know that the risk is well managed.
2:40:33
The head of risk at J Morgan doesn't want just financial coverage.
2:40:35
He wants to make sure that there's no instant that gets them fired in the first place.
2:40:39
>> And so we've developed a standard.
2:40:39
It looks a little bit like a Moody's framework or a SO 2.
2:40:43
Uh so that is all open source and public.
2:40:46
It's 50 requirements that any AI frontier company must meet to meet the standard.
2:40:51
And as part of that we run a bunch of technical tests basically crash testing red teaming as you might call it here uh which gives us a score uh and we give them pass fail certificate and then this score feeds into a policy that we've designed with uh some of the leading insurers uh where a company like 11 Labs
2:41:09
gets to specify hey what are the top three four five risks that hold up at option >> they buy a policy for that and today that that risk is held uh by traditional insurance companies again this is actually all about trust so you really want the old insurers that have it on their balance sheet. They always pay. Uh They always pay.
2:41:22
Uh over time as we move into this kind of like Chernobyl types risks, uh we will run out of private capacity.
2:41:30
We will have to at some point uh create catastrophe bonds.
2:41:34
Those will [snorts] be traded on the public market.
2:41:36
Probably not on Robin Hood by but by more sophisticated investors.
2:41:38
That is the ultimate the way to build enough market capacity to cover the tail risk.
2:41:44
>> Yeah, that makes sense.
2:41:45
>> Very very fascinating.
2:41:46
>> Uh what does the business look like today?
2:41:48
This feels like high stakes work, but is it capital intensive?
2:41:51
Is it do you need a thousand insurance agents at some point?
2:41:56
Like uh what's the team like?
2:41:58
What's the fundraising like?
2:42:00
What's the business like? >> Yeah, totally.
2:42:02
So, the way to un if you're really thinking about this long term, the way to unlock the insurance market is to get the standard universally adopted.
2:42:11
>> And that is kind of what allows everyone to say, "Hey, this risk is well managed.
2:42:15
We can now start to to price it."
2:42:15
And so we have for the standard we have about 100 security leaders from the one fortune 1000 who meet with us every 6 weeks to input into the standard as representing their interest and now you're having some of the leading AI companies like 11 labs intercom UIP more to be announced soon that have set put themselves forward to say hey we're pioneers we would like to have an independent audit to prove that that's step one.
2:42:37
So that's what most of our work is focused on today.
2:42:38
Um, and then on the insurance side, the the way to start is to partner with existing insurers that bring that trust credibility.
2:42:47
They're frankly so old school and that's what that's what brings trust here.
2:42:49
They don't take they're not. That's the whole point. >> Yeah.
2:42:54
It doesn't really it doesn't really work if you're It's like, okay, who's actually backing this policy?
2:42:57
And then it's like, oh, a company created >> it's like me. [laughter] >> Exactly. >> Don't worry.
2:43:05
>> So, it's actually quite caving light to get started.
2:43:06
We raised $15 million from that Freeman last year.
2:43:08
Um, and [laughter] >> almost ran into the the goalpost. >> There you go. >> Move them again.
2:43:20
>> Well, [applause] thank you so much for stopping by the show and giving us the update.
2:43:24
>> Yeah, a lot a lot uh more questions as there are new kind of crises around agents.
2:43:29
Feel free to pop back on talk about it. >> Amazing. We'll talk to you soon. >> Good to meet you.
2:43:35
>> Good to meet you, Run.
2:43:36
>> Let me tell you about Crowdstrike.
2:43:38
Crowd, your business is AI.
2:43:38
their businesses securing it.
2:43:40
Crowd Strike secures AI and stops breaches.
2:43:42
And without further ado, we have Rainineer Pope from Matt X in the ream waiting room. Welcome to the show. How are you doing? >> What's going on? >> Doing great. Very happy to be here.
2:43:54
>> Thanks so much for hopping on.
2:43:54
Uh it's your first appearance.
2:43:56
We'd love an introduction on yourself and the company to kick it off. >> Yeah. So happy to be here. I I'm Reiner.
2:44:02
I'm CEO and one of the founders of of MaddX.
2:44:04
Um we we are a company that makes uh the best chips physically possible for large language models. >> Okay.
2:44:11
>> So uh we've been doing this for about 2 or 3 years.
2:44:13
Before that I was myself I was at um at Google for about a decade working on large language models.
2:44:18
Uh worked on the TPUs for a bit worked on some other hardware projects.
2:44:22
Um and really as part of that what we saw was that um there was this like if you really want to make the best chips for LLMs and LLMs were this big up and cominging workload back in 22.
2:44:32
Um if you want to make the best chips for LM you need to do really the best way to do it is from a from scratch blank slate design.
2:44:40
So designed for large matrices very low precision um very low latency.
2:44:46
And so uh my co-founder Mike Gunter and I uh at that point in in 22 decided to leave Google to start MaddX um where we're doing exactly that.
2:44:55
>> The day we're announcing please >> uh Maddx 1.
2:44:57
This is our uh this is a a new chip which simultaneously offers better throughput per square millimeter or throughput of a chip than any other product in the market while at the same time offering um uh lowest latency latency that is comparable to the best which is Grock and Cerebras. >> Yeah.
2:45:14
What are the various tradeoffs in custom silicon design these days?
2:45:18
Is it is it just I mean at the highest levels is flexibility and speed or cost size wafer size like how do you think about the design space and then I want to know how you actually narrowed it on your particular uh decisions. >> Yeah.
2:45:34
So generally there's some kind of performance per something and so like let's analyze those pieces like the uh two different aspects of performance are what is the throughput and what is the latency.
2:45:44
So how many users simultaneously can I support is throughput and then latency is for one user how fast is the experience. >> Both of those matter.
2:45:53
>> Um and then on the like the the per per something like per per dollar how much does the chip actually cost?
2:45:57
Um and then per watt which is like what is the power bill of the chip.
2:46:01
Um so those are the like all combinations of those uh two numerators and two denominators are the things we care about. Mhm.
2:46:08
>> Um what we see in the market today is that uh the the number one constraint is just the throughput per dollar and the throughput per watt.
2:46:16
So I these frontier labs have so much demand for compute serving all of these like trillions of tokens uh per day.
2:46:23
Um and so they uh the cost and the economics is the main constraint.
2:46:29
There's only so much uh so many square millimeters of silicon wafer being produced every year.
2:46:33
And so uh given that constraint on how much silicon there is, can we maximize the number of tokens and then maximize the intelligence of the models uh coming that wafer?
2:46:44
>> Uh [clears throat] what does what does the go to market look like?
2:46:46
Are you already sold out?
2:46:47
Like who's who's uh who who are you kind of targeting early on?
2:46:52
How do you scale all that stuff?
2:46:56
>> So So one of the places where we've seen the most interest in our product is uh from Frontier Labs really.
2:47:00
And and so this is coming from a combination of uh the the uh they are the ones who are driving really all of this demand and are so much constrained on cost uh as well as uh silicon wafer supply.
2:47:12
Um but then also they are the ones who are who are doing these reinforcement learning training workloads which are very very latency sensitive.
2:47:20
They have to roll out um long rollouts uh in a very long loop.
2:47:25
>> So so that's where we've seen the most interest.
2:47:26
Um the one of the things that shows up there is that uh uh when they are looking to make place an order it is an order on the order of gigawatts or something like that which is which is massive volumes.
2:47:36
So one of the things that we're actually very excited to be able to do now um with this raise that we've just uh announced is um is is help uh ramp up the supply chain in order to be able to deliver you know gigawatts a year of of volume which is which is a massive volume to be able to deliver. >> Yeah.
2:47:54
>> Yeah. uh when you were initially thinking of starting the company pitching it to investors early on how did you answer the question around Nvidia's various modes or uh kind of strategic advantages you know think CUDA uh all that stuff
2:48:12
>> yeah I I think it's really interesting like CUDA is for Nvidia simultaneously the biggest strategic advantage and also a constraint because their promise that they make you is that you can take a CUDA program written 10 years ago and it will run on the next generation Nvidia GPU. Jensen goes on stage and promises
2:48:27
Jensen goes on stage and promises this.
2:48:29
It is so valuable for them and yet at the same time it means the next generation GPU has to look just like the GPU from 10 years ago.
2:48:35
>> So so it means things like the numeric can't change.
2:48:37
The way the cores in the chip are connected to each other can't change.
2:48:41
Um the uh the the actual memory architecture can't substantially change.
2:48:47
All of these things are kind of locked in by the programming model that they designed uh more than a decade ago for for general purpose uh parallelism.
2:48:52
And so so this is where we've seen the biggest differentiation.
2:48:56
If they wanted to say, well, we're going to like completely give up our CUDA approach and and start a new generation of chips.
2:49:01
Uh maybe they could do that.
2:49:04
They would lose all of this lock in that they have, but then at least they would be on a level playing field with us.
2:49:08
But but that's not what we see.
2:49:09
Uh really we see them being committed to to their um their trajectory.
2:49:14
their trajectory. uh this the the CUDA lockin is very valuable for um for sort of the mid and tail of the market where people are so sensitive to the software cost but really at the head of the market in the frontier labs the the software is not the main cost the
2:49:28
hardware is the main cost and so if you're willing to rewrite your software maybe you can you can actually switch to a more efficient hardware like us >> and it's getting easier to rewrite software >> uh as you plan your business how are you
2:49:40
thinking about bottlenecks be you know one month it's energy the next month it's chips then you know a lot of concerns around TSMC right now how are you how are you kind of planning >> yeah so I mean I think the these bottlenecks are real and are going to stay for a long time uh
2:49:58
>> what like the the big bottlenecks that you see in the manufacturing supply chain are on logic dies from TSMC and then memory dies from highex Samsung micron um and then and then manufacturing so of racks and so on >> uh given the these bottle given these bottleneck exist. What you would like to
2:50:14
What you would like to do as as a consumer of of of such things is you want to get the most bang for for your buck.
2:50:21
So the the most performance out of every square millime of silicon.
2:50:24
Uh that is what has been our focus.
2:50:24
We the the flops per square millimeter the 4bit precision uh multiplies you can do per square millm of silicon is higher in our product than any other product.
2:50:33
And so you know as the price of every silicon wafer goes up you can do more with it uh on on our solution than yours.
2:50:42
yours. I I assume you're on the most I mean you you've mentioned this you're you're selling to the frontier labs running frontier models uh probably on the most leading edge chips the most leading edge fabrication nodes um
2:50:58
is there a world where it's valuable to say hey we have some lagging edge capacity out there what if we go design custom silicon that runs on the last generation Intel node that's not the line out the door for capacity and then I'm not competing with you. Does that Does that not work?
2:51:18
Is that not possible or is that just a completely orthogonal business to what you're building?
2:51:21
So >> that that approach is possible um it it's it is maybe more of an approach for a um a player with rich pockets rather than a startup.
2:51:32
>> Um in in that like every different process you target it costs you another 2030 $40 million of development cost. Yeah.
2:51:38
And so if you're going to bet all of your eggs on like put all of your eggs in one basket, you should put it in in the leading edge node >> in the best basket.
2:51:45
>> Yeah, >> that makes sense.
2:51:45
Uh talk to me about other tradeoffs at TSMC.
2:51:47
I mean uh Cerebrus is famously wafer scale.
2:51:50
Um how what is the trade-off on like size of die these days? >> Yeah.
2:51:59
So I mean there there's a trade-off of size of die and then also memory architecture.
2:52:02
So size of die um Cerebrus is the outlier.
2:52:05
Um almost everyone else has converged on reticle scale which is the largest sort of standardly produced um TSMT uh chip.
2:52:10
Um we're in that same category of like we're we're about this we're in the standard uh bucket there.
2:52:17
It that avoids a lot of the physical risks that uh you know when you look at cerebras they've had to spend all this time on dealing with just like bending and and all these uncomfortable physical constraints that we don't want to deal with. Yeah.
2:52:30
So the reticle size chips but then the other bigger thing is which memory technology do you use?
2:52:36
Uh the historically there's been like the HBM based players that's Google, Amazon, Nvidia and then there's been the SRAM based players which are Cerebras and Grock. >> Mhm.
2:52:46
>> Uh SRAMM is small but very very fast >> and so uh the very very fast is good if you want to run low latency.
2:52:54
You can put your model weights in SRAMM and and you get the best latency in the market.
2:52:58
That's what Grock and Cerebras have done.
2:52:59
Uh but the reason they haven't like sold out in the market is because uh there's not enough space in the SRAM to to store all of your long context uh KV caches. >> Sure.
2:53:10
>> And so one of the things that uh like this is the reason why the HBM based players like Google, Amazon, Nvidia um have won is because of like the HPM is actually essential.
2:53:20
>> But it's actually possible to marry both of these uh approaches and put them in one chip and and that is what we're doing with MADX1.
2:53:25
Um and so uh it curiously I mean it doesn't just give you the best of both worlds.
2:53:30
It actually beats any alternative on throughput.
2:53:32
Um there's this curious effect where when you have your weights in SRAMM you can actually get better mileage better usage of the HBM um in return.
2:53:41
And so there's uh we think this is actually the the way where the market in general will move over time. >> Okay.
2:53:49
uh help me understand uh the the the trade-off continuum of around flexibility of of model.
2:53:55
I mean imagine uh on an Nvidia NVL72 I can sort of run any model as long as it fits and works and is trained properly.
2:54:05
Uh and then Talis is like these specific weights on the chip.
2:54:10
You can never change them uh whatsoever.
2:54:13
Uh and then there's something in the middle.
2:54:15
How much flexibility do you think is important?
2:54:19
How much flexibility are you planning around and how and how do you think about sightelines?
2:54:23
Because I imagine that the delay between like the final architectural design to chips in data centers is still a year, 18 months, something like that. >> Yeah.
2:54:35
I mean that there's all of these manufacturing and then deployment uh times that that make it take a long time.
2:54:40
In general, I would say that from from sort of pencils down on on chips to like when is the last time you're using it, the chip is going to be in the data center itself for like three to five years and then there's maybe as you say a year a year and a half of of deployment time in advance of that.
2:54:52
So you want your chip to be relevant for for a 5y year time span maybe.
2:54:57
>> Um >> the uh so you need to point pick a point of specialization which you think is here to stay.
2:55:03
For us that is very large matrices.
2:55:06
Uh and then in fact really large matrices together with a splitable systolic array which is a um piece of technology.
2:55:12
Uh but very large matrices is the the the theme that we started with and this is just a recognition of over time models have been growing.
2:55:19
They grew a ton with LLMs and they're continuing to grow.
2:55:22
Um and if you specialize for that you can get big efficiency on the matrices themselves.
2:55:29
>> Uh now we are still very general purpose programmable in terms of the vector unit.
2:55:33
Um, similar to Nvidia, we have this vector unit that you can run any instruction on, like add, multiply, subtract, divide, all of those things.
2:55:41
Um, and so that gives you the like it's it's trying it's trying to put a good amount of flexibility, but in a in a way that only costs like 5 10% of the the cost of the chip overall. >> Okay. >> Funding news. >> Give us the news.
2:55:53
How much did you >> What happened? >> So, we're happy.
2:55:56
We we we have raised $500 million. Uh, this this was round. >> Yes. Boom.
2:56:05
>> Who'd you raise it from?
2:56:07
>> So we this was led by um main street and situational awareness.
2:56:10
Uh so situational awareness that's Leopold and Brennes fund.
2:56:14
He >> if you've been living under a data center that is that is uh that's his >> exactly you might have heard of him.
2:56:19
Um uh so he really sees just like the big picture of where this this space is going and and he recognizes like just how much demand there is for silicon.
2:56:29
And then on the other end of the spectrum, Jane Street, they are expert technologists.
2:56:33
They know everything about what exactly is required to build a product like this.
2:56:36
And they they know what good is in a product like this. >> Mhm.
2:56:41
>> So we're we're really happy to have these like like strong experts here.
2:56:42
Um this sort of mirrors what we see inside the company as well.
2:56:48
We have a wide range of experiences across hardware, software and ML.
2:56:51
And then even in in the rest of the investors who are participating in our round, we have like uh uh renewed uh participation from from our previous investors.
2:57:00
This is um Spark Capital uh and NFTG um as well as uh a range of folks such as the Patrick and John Collison.
2:57:09
um uh experienced ML people like Andre Kapathy um >> and then even participation from the the supply chain like Marvel val >> Did you let any normies in?
2:57:20
They're just the most elite people in the world.
2:57:25
[laughter] >> Yeah, we like >> just one mouth breather, please.
2:57:30
[laughter] >> Congratul ever. It's amazing.
2:57:36
I'm extremely excited for this and excited for it to >> now you have to you have to you have to win on such massive scale otherwise you'll bring dishonor to [laughter] all the industry legends.
2:57:46
So >> no [clears throat] thank you no it's it's uh really uh cool to hear your perspective and approach to everything and I'm sure you'll be back on the show this year. So congrats to the team.
2:57:55
>> Yeah we'd love to have you back.
2:57:55
Thank you so much for taking the time. We'll talk to you soon. >> Goodbye.
2:58:00
>> Let me tell you about vaude.
2:58:00
co where DTOC brands B2B startups and AI companies advertise on streaming TV.
2:58:05
pick channels, target audiences, and measure sales just like on >> Have we Have we had a >> investor lineup like that before the Jane Street situational awareness co-lead and then just down >> fantastic?
2:58:16
Well, I mean, situational awareness is a new fund has not led that many rounds.
2:58:21
>> I know, but I'm just saying you go back, >> maybe it turns into a spray and prey fund. You never know.
2:58:24
Maybe Leopold says, "Yeah, I'm just going to write five million dollar checks to every company." Who knows?
2:58:29
Anyway, uh we have our next guest in the re room.
2:58:31
We got Standard Intelligence. How are you doing? >> What's going on?
2:58:37
>> Hey, uh I'm the launch. Um doing pretty well. How about you?
2:58:41
>> We're doing fantastically.
2:58:41
Uh thank you so much for taking the time to come on the show.
2:58:44
Since this is the first time on the show, I'd love an introduction on yourself and the company. >> Yeah.
2:58:48
So, I'm um co-ounder of Intelligence.
2:58:51
Uh we pre-train computer use models basically.
2:58:54
So basically the thing people are doing is they're training you know on screenshots um and like train of thought traces and we're just like what if you train purely on 30 fps video.
2:59:08
um what actually goes into the training data because there like there's a lot that you can do on a computer and I feel like if you've never trained on Ableton and it just comes randomly like are you actually going to be able to learn
2:59:21
Ableton from just playing in Premiere Pro and Word and you know Paint or something or Photoshop or whatever like h how are you thinking about the transfer and like what's actually in the training set actually just zoom out and talk about the process in more depth. >> Yeah. So we have like two splits of data >> Yeah.
2:59:36
So we have like two splits of data where we have like you know this small like contractor split.
2:59:39
Uh the thing that we did was we like made this app uh that people run on their computer and it records their screen and you know logs all their key presses um and and all their mass movements and we're running that all the time.
2:59:52
And then we also have this like much much larger uh kind of unlabeled data set of basically every video that we could possibly find uh that we're allowed to use on the internet of computer use.
3:00:02
Um, and so yeah, we we trained a model to to label that big set from this like small contractor only set.
3:00:09
And the goal is to just like be able Yeah.
3:00:11
just like train on all of it and like train this kind of like general model that is that is able to to generalize to basically anything that you could do on a computer.
3:00:22
>> Are what what kind of limitations do you have on on who can install your software to capture that data?
3:00:27
If I'm a if I'm a company, I feel like I have to have a pretty high degree of trust in in you guys to let my employees is that something that that's like a is more of like a partnership. Uh how does that work?
3:00:40
>> So right now it's like us like we're recording our own screens all the time plus like we have some number of contractors and we get to like pay them you know somewhat less because they're not doing like active work for us.
3:00:48
It's more like passive screen recording. >> Sure. >> Yeah.
3:00:52
Uh, and then are you sitting on top of some sort of foundation model brain for reasoning chains and sort of like the LLM piece of the puzzle or is this a model kind of lives? You're not right now.
3:01:06
>> Not we're not at all like the the the model that we released is or I suppose like demoed is like >> entirely trained on this kind of 30 fps video in and like you know typing and mouse movements and things like this out. >> Okay.
3:01:20
So, how much is that again?
3:01:24
>> How much longer will I have to fill out forms on the internet?
3:01:25
[laughter] I've I I should try to estimate how many times I've entered the same information just over and over and over and over.
3:01:33
And John John Collison >> John Collison talked about it on our show today and was talking about it on his own show, I think, with Ben Thompson talking about like at what point can you just take a link and say like, "Hey, please buy this." Yeah.
3:01:47
>> And then it just does it for you. like pretty soon.
3:01:50
I think like that kind of use case is just like under six months away depending on what like exactly you mean. >> Yeah.
3:01:58
I I mean uh in terms of actual deployment I imagine that uh this would be something I I personally would probably want deeper as more of like a tool that's called from a consumer LLM app.
3:02:12
Is that how I'm is that I'm correctly thinking about this or do you think they'll actually be like [clears throat] will you jump straight to consumer?
3:02:19
Um it's like yeah I think in the short term the kinds of people that are particularly like you know cool to sell to are like you know mechanical engineers doing CAD where like they can press the the tab button like software engineers press tab and cursor and have
3:02:36
their next like you know minute or two minutes of of manual work um done and and and we showed that in the the kind of gear extrusion demo where like you have this gear and you're like extruding faces and that's just like a very very common thing that you do in CAD. And I
3:02:47
And I think there's like a more general thing where like yeah um you can think of computer use as like a tool call or you can think of it as like you know just the thing that you do um >> you know for knowledge work and I think >> we're just in a place where like we can scale computer use um on its own.
3:03:05
It's not impossible that we'll like initialize from LLM or for example like use text training to like make the model smarter in in text space so it can fill out forms better.
3:03:16
But uh it is it is not the goal of the company that like you know people have you have Claude like call this as a tool call.
3:03:25
The goal is to just use your computer um or like use its own computer just like in general.
3:03:32
talk about your experiments with with self-driving and do does that uh does that work potentially apply to robotics more generally? Yeah.
3:03:40
So I think I think this general like pre-training thing or like you know labeling a bunch of uh unsupervised data with with actions uh and then training on that like labeled data um this like inverse dynamics thing works very very or like I expect it to transfer very well to robotics.
3:03:59
robotics. um self-driving in particular uh it was kind of so so Neil who works at SI was like okay we have this action model um and his friend had a comma and so the there's a comma like joystick mode where you can like control the the steering with with arrows and so we were
3:04:19
like okay well if it's a general computer use model surely it should be able to you know control a car um because that's just like a thing that you do on a computer uh it's like video in uh you're you're seeing it on the screen and then you can press the left and right arrow keys to steer. Uh and
3:04:33
Uh and obviously we originally didn't really expect this to work and then it just like worked much better than expected. We find an hour of data. >> It's a good sign.
3:04:45
>> On how on how many hours? Three hours.
3:04:47
>> One hour less than one hour 50 minutes. >> That's crazy.
3:04:50
>> And you're able to just fully the the the the system can just navigate around SF.
3:04:56
>> I mean, sorry, navigate around like South Park.
3:04:58
like it's it's you know not general self-driving model.
3:05:01
I would not recommend like sitting in this car and just like letting it do whatever it wants.
3:05:07
But yeah, it's it's it's pretty cool. >> Si take the wheel. Don't make mistakes.
3:05:11
[laughter] >> Take the wheel.
3:05:13
>> We are we are not a Tesla competitor.
3:05:13
We are not we are not a competitor.
3:05:16
>> Do you think uh do you think the sport coat is like the next it apparel item in this?
3:05:22
Because it looks fantastic here.
3:05:25
The chat loves your sport coat and I just feel like that could be the middle ground between Wall Street and San Francisco, >> but you're fantastic. >> Um, yeah. I don't know. I really like this.
3:05:35
I got it from like Bonobos and Union Square. >> There you go.
3:05:40
>> I think I think I like dressing up like at least a little bit and it's it's fun. >> That's good.
3:05:45
Okay, back to the business. Uh, [laughter] >> yeah.
3:05:48
I I want to know about uh it feels like you're you're training a very generalized model.
3:05:54
generalized model. uh what are you learning from the previous product launches where you know we had this chat GPT moment and then I don't even remember what people were just kind of chatting with chat GPT back and forth and then they started using it they started using it kind of as a Google replacement and then that kicked off the
3:06:14
whole like Google's cooked narrative and then with the with the studio Gibli moment it was really the launch of like a better diffusion model with some reasoning in there I think and stuff and so and then people were just like this is a studio Gibbly creator and then they found that niche of like it's really good at creating cartoons. It's not
3:06:30
It's not quite style transfer but that's what it does.
3:06:33
Well, how much do you want to just like turn a wild open model loose and then hope that someone finds a killer app versus like you kind of know that this is going to kill in CAD and you're just going to launch like cursor for CAD on day one and then like go from there.
3:06:51
Yeah, I think there's I think the answer is like some combination and like okay shortterm CAD design work somewhat generally are things that the current models just like totally can't do like LLMs are just like or like anything that is an LLM harness is just like really really bad at CAD for example and so that seems like a okay we know what to
3:07:12
do there we like you know can just scale up this model we have a bunch of Blender data a bunch of 3D modeling data in general and we can scale up CAD and then also yeah I like am quite excited to release a more general, you know, tab model for people to to play around with and like figure out what it's particularly good at. And so it's like
3:07:30
And so it's like when I'm asked like what commercialization plans are, it's like we have some reasonable idea of what the first steps are, but like there could just be this like massive uh thing once people start playing with it at that scale.
3:07:44
>> So you're training on video frames, 30 fps video, correct? >> Yeah.
3:07:49
I was told by an anonymous poster on X by the name of Rune that text in fact is the universal interface. Was I lied to? >> Yes. >> Whoa. Shots fired. Explain. Elaborate.
3:08:04
Like why doesn't this just collapse down to text?
3:08:06
Why don't I puppeteer CAD from text?
3:08:09
Like how does this all play together? >> Like Okay.
3:08:13
I think it is in at some point in in the like arbitrarily long future like if we only use text models we could force like most things to be texted.
3:08:23
I think there are just like a lot of things that are much more native um when done from like a computer use like you know GUIs are designed for humans.
3:08:32
They're designed for like humans to use.
3:08:33
Uh we have you know this massive long tale of like things on the internet that are like entirely undoable by LLM.
3:08:42
for example, like when I do ML engineering, right?
3:08:44
engineering, right? because like most of my time is is not spent uh most of my time is just like spent doing kind of this grunt work of of engineering and it's like um a lot of looking at graphs and like analyzing graphs and and you
3:08:59
know figuring out uh comparing loss curves or something and like you can do this in text but it's just a much larger pain than doing it in this kind of native interface which is um video and I don't know there's a reason why humans don't interact with a computer purely through text. Uh it would kind of suck.
3:09:15
Uh it would kind of suck.
3:09:17
Um for example, we have like the concept of like video has the concept of time in a way that text doesn't. >> Mhm. >> Speak for yourself.
3:09:26
I got green I got text right here. Black background going.
3:09:30
>> If this can eliminate YouTube tutorials for software, that's a killer app. Is this >> Yeah.
3:09:34
Is this is that anything?
3:09:36
>> It's not just going to eliminate the tutorials.
3:09:37
It's gonna eliminate the whole the whole process [laughter] because you don't need the tutorial if you're just like just go do the thing that I need you to do.
3:09:44
But yeah, I mean you're you're you're obviously in the phase for a while.
3:09:49
>> How are you thinking about uh go to market in general for the underlying technology? >> Yeah.
3:09:56
So I think like as I said there there are the kind of shortterm like uh CAD design use cases.
3:10:00
There's like the tab model which we want to just like give anyone a kind of general thing of you know in cursor you press tab and and it completes your next edit or whatever.
3:10:13
What if you could press tab and it completes like the next five and then 10 and then 60 seconds of what you would do in your computer.
3:10:18
Um, and then I think longer term it's just like, you know, we're training a general model that is able to do to do useful work and you'll be able to like send it off with a prompt to do work.
3:10:29
And then like there's a very interesting thing where like the data that we're training on is very there's a bunch of like error correction built into it.
3:10:37
So like when you have a bunch of data of humans doing things, a lot of the times the humans make mistakes and then they have to like correct those mistakes.
3:10:44
Um, and you don't get that with te with text because like most text on the internet you don't get to see the process of like you know messing up and then and then fixing it.
3:10:54
Um, and so yeah, I expect there to be a lot of like uh native like no um just like prior of of doing the selfcorrection thing properly. [clears throat] Mhm.
3:11:08
>> So you can you can get it to like go do something for 10 minutes and it'll like try something for two minutes and then and then like you know mess up slightly, but it like knows how to fix that over and over again until it's like gotten to a solved state. >> Yep. >> Very cool.
3:11:22
Well, congratulations on the launch and thanks for taking the time to come chat with us.
3:11:26
>> Thank you for sport coating.
3:11:27
>> I'm gonna We're going to get some We need some sport coats around the office.
3:11:30
You need something in between.
3:11:30
You got John over here formal.
3:11:32
I'm doing casual Friday on a Tuesday, but a sport coat >> perfectly in the middle.
3:11:36
It was great to meet you and uh come back on soon. >> Thank you.
3:11:41
>> We'll talk to you soon. Cheers. >> Goodbye.
3:11:45
>> Well, >> back to the timeline. >> Back to the timeline.
3:11:47
>> Uh there was an individual >> who uh accidentally gained control of 7,000 DJI vacuums. He was just vibe coding. >> Amazing.
3:11:58
and uh accidentally found according to investment hulk the CCP back door control it with a gaming controller >> and then he just he got control of everything. >> That's so crazy.
3:12:12
>> This is why I've been deeply concerned with letting uh >> any any foreign adversary flood our country with uh >> a bunch of robots.
3:12:22
I think we should avoid it.
3:12:25
>> Uh I thought this was funny earlier.
3:12:27
Haggth says he'll order random pizzas to throw off the monitoring app.
3:12:31
>> Oh yeah, >> I expected something like this to happen.
3:12:34
It's kind of silly that everyone has a dashboard up and can tell when things might be getting a little more tense in the Pentagon.
3:12:39
So yeah, give them a budget of, >> you know, a few hundred,000 a year and just order pizzas at random times.
3:12:48
>> For the record, this is a joke and he is joking, but I do I do think they could throw it off potentially. You never know.
3:12:54
They could they could they could throw it around.
3:12:57
>> Hub HubSpot acquired Starter Story.
3:13:00
>> Yeah, this is very exciting. >> Very cool. >> Yeah.
3:13:02
Starter story overnight success.
3:13:04
What a decade he's been doing this.
3:13:05
>> I believe Pat the founder had just posted.
3:13:08
>> Yeah, >> he posted something.
3:13:09
He was like sub or he said HubSpot should acquire Starter Story and then like two weeks later it was done. >> Wait, really?
3:13:15
Oh, I thought that was from like years ago. >> Oh, maybe it was.
3:13:18
I thought I thought he posted that a long time ago and then Yeah, he said >> No, he said September 23, 2025. So >> Oh, wow. Not long. Yeah, couple months ago. >> Couple months.
3:13:28
>> HubSpot should acquire Starter Story. The SEO ship is sinking.
3:13:30
In my opinion, HubSpot needs to pivot way harder to video, specifically YouTube.
3:13:36
>> Yeah, >> I'm biased, but acquiring Starter Story would take their YouTube game to the next level.
3:13:40
>> And he was quoting Brian, the co-founder, saying, "Dear founders, it's a good time to sell your company. Love Brian.
3:13:47
Um, >> anyways, uh, there's a bunch more stuff in here, but we will get to it tomorrow.
3:13:56
>> Tomorrow, Arena Mag is out.
3:13:56
Go check out issue number seven. They're on Substack now. arenaagazine. substack. com. Go check it out.
3:14:04
And >> one more post for you.
3:14:08
>> Deep dish and Joyer says, "I don't see what the point of shoveling snow is when AI agents are going to commoditize burrito taxi services by 2028.
3:14:18
It's a good >> good excuse.
3:14:19
>> Leave us five stars on Apple Podcast and Spotify.
3:14:21
Subscribe to our newsletter at tbpn. com.
3:14:24
>> Have the best evening of your of your entire life. We love you. >> Goodbye. >> Nice work, brothers.
3:14:32
I'll see you on the next one.