0:00
[music] >> I see a larger PR moving on the horizon.
Search this video
0:09
You're surrounded [music] by journalists. Hold your position. >> Overnight success. >> The mainstream media. Double click. Right. That's misinformation.
1:00
>> Walking clearing order in. >> That's just wrong.
1:09
>> You're surrounded by journalists. Hold your position. >> Comp, get up. Trust the experts. >> [music] >> Five code. Leave or enter. >> Founder Miller. >> Five code.
1:43
>> I see multiple journalists on the horizon. Stand by. >> Experts said what? >> UAV online. >> Please. Double blades. Triple blades.
2:09
>> [music] >> Double kill. Fire cut. Run. Go get up. Blades. Team death match. Blades are experts. Triple blades. Let's just [music] run. Right.
3:07
Marking clearing order in effect. Go >> [music] >> get up.
3:21
You're surrounded by [music] journalists. Maintain your position. Strike one.
3:35
>> [music] >> Strike two.
3:43
Activate golden retriever. Twist up.
3:57
>> [music] >> Monkey clearing order in bound. >> Fire cool.
4:19
>> Possible anomalies on the horizon. Stand by. >> Found her. >> You're watching TVBN.
4:36
Today is Thursday, July 9th, 2026.
4:39
[applause] We are live from TVBN ultra down, the temple of technology, the fortress of finance, the capital capital.
4:44
Let me tell you about ramp. com.
4:46
Time is money, save both.
4:47
Easiest corporate cards, bill pay accounting, and a whole lot more, all in one place. I need some soundboard. Here we go. Yes.
4:55
Today on TVBN, we're talking about model mayhem.
4:58
Everyone's launching new models.
5:01
Slow summer, but not for the AI race.
5:04
You got XAI unveiling Grok 4.
5:04
5, the first model that's built specifically for coding and AI agents, developed in collaboration [music] with Cursor.
5:12
Talked about it a little bit yesterday, but we have some more benchmarks, uh [music] some more uh discussion on the timeline about where this model fits in on the Pareto frontier.
5:20
[music] Also, why it might be outperforming so well on Cursor bench.
5:24
Uh lots of debates there.
5:26
Meta announced new Spark, a new agentic coding model with Mark Zuckerberg returning to X for the first time in basically a decade.
5:33
Three years ago, he posted one joke post about launching Threads, but uh he has not been an active user, but the AI vortex sucked him in, and he's got to post.
5:45
>> Oh, I think he's an active user, John. >> You think so?
5:48
>> He's just not an active poster.
5:50
>> He's just not an active poster.
5:50
>> an active contributor.
5:52
>> You're calling him a lurker.
5:52
>> I'm calling him a lurker.
5:53
>> You're calling him a lurker.
5:54
>> I'm calling him a lurker.
5:54
I think he's absolutely glued. >> You think so? >> I think so. >> You really think so? >> I think so.
6:00
>> I feel like I don't know, I'm so busy, so much other stuff going on.
6:02
I feel like he I feel like most people I know are >> people I know >> Okay. >> are not active on X. >> Yeah.
6:12
>> But they they are on X a lot. >> Sometimes.
6:16
But there are there's a different class of person who works >> quiz You can just quiz them.
6:20
>> Screenshots come to them uh via Slack or or via text message because they have a team that's monitoring the timeline and then is delivered.
6:28
This is the important This is the important stuff.
6:31
>> You're calling him Mark Lurkerberg.
6:32
>> Lurk lurk Uh but uh the other big news, OpenAI uh just released GPT-5. 6. Let's go. Let's go. >> I love her.
6:42
>> A new general-purpose model with expanded coding agent capabilities alongside GPT Live, which we talked about yesterday, a new real-time interactive voice experience.
6:49
Uh reactions are great to 5. 6.
6:51
Bunch of interesting details here.
6:54
Uh you had uh people have been uh identifying that uh while there is a frontier and there are just a few companies that are actually on the frontier, um the the the frontier is spiky and they have different flavors to them in different uh different reasons to pull different tools off the shelf.
7:15
Uh people are drawing analogies between Fable 5 being some, you know, recluse genius and and 5.
7:21
6 being a, you know, a collaborative coworker that you love chatting with or something like that.
7:29
Well, >> I said, "I don't know how else to describe it, but Fable 5 is like Kendrick on Good Kid, M. A. A. D City and 5.
7:35
6 Soul is like Chief Keef on Finally Rich."
7:38
>> Now it makes sense to me.
7:38
Thank you for bringing that up.
7:40
wanted >> it into, you know, 2010 hip-hop hip-hop of terminology.
7:45
>> Really, really clear there.
7:45
Thanks for clearing that up. Uh yeah.
7:49
>> the funny thing is that will be very explicit for like 100 people in the whole world. So.
7:56
>> Uh well >> This one's for you.
7:57
>> Uh the fun the the the the most interesting benchmark to me has always been Arc-AGI uh V3.
8:01
We've interviewed the team over there many times and had a lot of fun uh understanding what goes into that uh that benchmark. And uh 5.
8:08
6 Soul scored a massive 7.
8:12
78%, which is tiny >> [laughter] >> uh considering that uh the whole point of Arc-AGI is that a human should be able to get 100% on it and basically any human.
8:25
So, it is a true test of AGI in the sense of, you know, can you give this test to just actually anyone?
8:30
Not, you know, the the the the crazy math projects, the crazy hard programming projects, the hacking, all of that stuff is very economically valuable, of course, but there's a more interesting question where, you know, when there's less of a spiky frontier and there's just this question of what is something that anybody can do that AI can't?
8:51
Because we've been searching for those and the Arc-AGI team has done a fantastic job building out these puzzles that AI has historically struggled with.
8:59
Arc-AGI 1 uh the model sort of climbed, 2 became a little bit more complicated, and now 3 we're starting to see glimpses of progress, although 7.
9:08
76% isn't 99% or nowhere near saturation, but it's still a huge jump. Opus 4. 8 had 1. 5%, so uh GPT 5.
9:18
6 Soul is showing more generalization, more spatial reasoning, more uh puzzle-solving abilities. So, fun fun stuff.
9:26
Um I am trying to refresh my timeline, but uh the blog post is also very, very fun because it includes games.
9:34
I'm a big fan of the uh the GPT 5. 6 launch games.
9:37
I got immediately sucked into the uh to the uh the sailing mini game, which is uh like very high fidelity, but also delightful to actually play. >> Should we play it?
9:54
>> Yes, we should definitely play it. Yeah, Salt Wind.
9:56
You You You guys play it.
9:56
I want production team to see what they can get.
9:58
I think my time was 25 seconds.
10:00
>> this hosted on a on a site?
10:04
>> this is I mean this is hosted on the OpenAI blog, but I think the the idea is that you could vibe code this in the latest GPT 5.
10:11
6 in the app in ChatGPT and then deploy it and have someone Are you trimming the sails appropriately?
10:18
Because it looks like you're losing speed. You're losing wind. It's not working.
10:22
Yeah, I'm going to smoke you. I got 25 seconds. Wow. Amateur hour over here. Look at this. >> Did you lose? >> Yeah. Yeah.
10:29
Uh Well, you well, the whole the whole game, which you probably missed, is that there's a little bar there where you have to trim the sails to be in the sweet spot of the wind while you're turning.
10:38
So, as you turn See the bar?
10:40
There's a recommendation for where you put the sails.
10:43
You got to keep that line in See? It's moving over.
10:47
You got to You got to press the uh down. S Yeah, exactly.
10:50
Keep trimming those sails while you steer the ship.
10:55
This stuff is very very fun.
10:55
Um I am trying to open >> Uh one interesting data point from the live stream, which was just an hour ago, they said, "Already Soul has been transforming our research program as one example GPT 5.
11:08
6 Soul autonomously post-trained 5. 6 Luna." >> Yeah, that's fair.
11:15
>> A lot of people are uh having fun with that.
11:16
Dylan Field says a lot of people want to compare Fable versus 5. 6 Soul. This is a mistake.
11:24
They're apples and oranges.
11:25
Despite all the research achievements, we're still very very early in exploring the tech tree for model training. >> Okay, cool.
11:35
>> Sorry, I'm just getting set up again. Um what else is in here?
11:38
Uh Uh oh yes, I I I do think that uh didn't uh Dylan Abernathy write something about this?
11:46
What what what was the uh the the the the essay he wrote about uh uh interactive memes and this idea of like generative AI enabling these vibe-coded mini-games like we've been seeing a bunch of them with like the capybara simulator, the coconut simulator where it's something that's just a joke that's funny for like a few people, but and normally you would instantiate that in a in a tweet.
12:10
Or maybe if you were getting really crazy you'd do a Photoshop edit of a meme, but now you can go and create a full uh mini-game, something that runs in the browser.
12:20
And soon something that runs in Unreal Engine and can actually be distributed on the Steam store.
12:24
We've already seen that with like the data center simulators and all these funny uh simulator games that are going on Steam.
12:31
Uh all this uh all the all the advances in the coding models certainly speeds up the ability to actually deliver polished software.
12:39
I'm I'm I'm particularly excited for like >> Dylan's title was the future of entertainment is interactive. >> Yes, yes.
12:45
>> But uh but yeah, that that's part of what I honestly love about AI.
12:47
There's a lot of things you can make now that never would have made sense to make because they would have taken you 4 days and it was good for like a small laugh.
12:57
Now you can do it in 4 minutes. And uh it's just fun.
13:01
>> Yeah, I I I I think there's going to be there's if you have some sort of like small custom some sort of custom functionality in your business, uh it feels like there's >> Is this the David Senra simulator?
13:15
[laughter] >> Why is this David Senra?
13:17
>> Late nights in a Miami abandoned apartment complex in 2015 just recording podcasts and reading.
13:25
>> This is very creepy like uh horror backrooms liminal space game.
13:30
>> Stanley Tang co-founder and CPO over at DoorDash says, "I have an insane magic trick that so far none of the models can figure out including Mythos.
13:37
It's a bulletproof trick that I've shown to 100 plus people including magicians that couldn't figure it out.
13:43
It's not anywhere on the internet.
13:44
Only way to know it is through first principles reasoning."
13:48
Told everyone I'll believe in AGI when it can crack this trick. Well, GPT 5. 6 just did. >> How?
13:57
>> I want him to I want him to actually open like Okay, like give us now that now that a model cracked it, explain it.
14:03
>> of magic tricks are like slight of hand.
14:05
So, is he uploading a video or something? Like >> Well, yeah.
14:07
So, John Palmer says, "I have a hilarious joke that so far none of the models think is funny.
14:11
It's a bulletproof joke that I've told to 100 plus people including comedians and no one laughed.
14:16
It's not anywhere on the internet.
14:18
Only way to know it's funny is a first principle sense of humor."
14:22
Told everyone I'll believe in AGI when it tells me a joke. The joke is funny. Well, 5. 6 just did. >> Huge huge news. Huge news. >> Huge.
14:31
>> Um Yeah, people are [clears throat] going back and forth. Uh GPT 5.
14:34
6 is a Porsche Fables like warp drive.
14:36
I had a different experience.
14:38
If Fables is an F1 car, 5.
14:39
6 [snorts] Soul at Ultra is a Tesla Model X Plaid.
14:42
Does it find things that Fables misses during planning and coding? Yes, most of the time.
14:48
But for the hardest problems, does Fables routinely find things that Fables that 5. 6 doesn't?
14:51
Also yes, some of the time. Uh is 5.
14:53
6 way faster and affordable?
14:56
Yes, with an unlimited token budget.
14:58
What am I currently using 95 per 95 plus percent of the time? GPT 5. 6 from Siki Chen.
15:04
So, uh interesting take that the the parade of frontiers is alive and well and everyone's duking it out for their slice of the uh the AI opportunity.
15:14
Very interesting seeing uh how the how the market share is shifting while during a time of acceleration.
15:19
You have multiple companies that are growing revenues, even accelerating revenues, while market share is declining because the overall market is growing so fast that you that if you're only growing at 300% and someone else is growing at 400%, you're losing market share, but you have like one of the greatest businesses by modern metrics.
15:36
Very, very interesting dynamics in AI.
15:39
>> It's also funny because yesterday with Ben Thompson, you were like, "A some slow summer."
15:44
And then in in the >> [laughter] >> in the span of 24 hours, you get >> [cough] >> four or five >> [clears throat] >> Muse 1. 1.
15:51
>> Yeah, I mean, this I don't know.
15:54
This isn't This isn't as dramatic as the AI talent wars.
15:58
It's not as dramatic as >> rippling deal? >> Yeah. Yeah.
16:01
Uh >> [snorts] >> this is this is new technology.
16:03
Um and and there's only so much to uh >> [snorts] >> There's only so much of a take to be given around these things.
16:11
Although, uh AI 2040 launched today, the sequel to AI 2027.
16:16
That's something that's more of a thought-provoking piece that you can debate and interrogate and and talk through.
16:22
I'm sure we'll go through some of it because they pose uh a couple interesting um couple interesting ideas of where where AI might go and where they want it to go and how they want uh the industry to develop.
16:32
Uh sort of advocating for a slowdown generally, but uh it's an it's an interesting uh way they puzzle piece all the different geopolitical uh uh chips on the table around. Um What else?
16:45
Uh of course, people are joking about the lead is widening because um the the Anthropic and OpenAI version numbers over time. GPT-6 is predicted.
16:57
And uh it is it that that the the the model numbering We were talking about this this morning that uh the numbers they sort of don't mean anything anymore.
17:07
Do the model numbers mean anything in particular?
17:09
It used to be the the model number was the pre-train, and then the and then the version number was the post-train, but then that sort of got flipped around and now it's just like are you do you feel like you're competing at a four class or a five class?
17:24
So, I wouldn't be surprised if we saw like Muse Spark not not release Muse Spark 2, but Muse Spark 6 or 5 and jump straight.
17:34
I mean, the Samsung wound up doing this where they jumped to the year like sort of like the car manufacturers where you know, there's a five series BMW, but then there's also just the 2027 because that's the actual model year that's relevant. >> 2027 five series.
17:51
>> Yeah, which is sort of odd.
17:53
And we're sort of like duking it out between those.
17:55
Do you have >> Yeah, I mean, I think post reasoning models you just have like a different way to scale the models besides just pre-training.
18:01
So, it's hard to bake that all into one number that like is you know, evocative of both those like two ways.
18:07
>> Yeah, so the number is is becoming closer to the year in the in the second decade of the 21st century basically.
18:14
It's just like is this on the frontier in 2026?
18:16
You'll probably see a six by the end of the year in front of the models that are leading in in the year 2026. Something like that.
18:28
I'm very interested with Google's strategy because the the rumor is that 3.
18:31
5 Pro will be coming out this next week, I believe, but I it was it's very odd going into the Gemini app right now and seeing that there's 3.
18:44
5 flash, but then you have to go back to 3. 1 Pro. I think 3.
18:47
1 Pro is the most advanced model, but they default you to 3.
18:53
1 flashlight and I would expect them to jump just forward to four, but I think that they're going to do 3.
19:00
5 Pro, but it's been a little bit of a slower cycle there.
19:02
As silly I mean, all the obviously all these numbers don't really mean anything.
19:07
They're marketing terms, but they Uh, they I still think they do actually stick in people's mind and so there should be some strategy around them.
19:16
Um, but anyway, before we move on to our next story, let me tell you about the New York Stock Exchange.
19:19
Want to change the world?
19:20
Raise capital at the New York Stock Exchange.
19:23
Um, so uh, Mark Zuckerberg is on a press tour.
19:26
He's talking to the legacy media for the first time in a long time.
19:31
Uh, Andrew Bosworth, the CTO of Meta, also did an interview uh, with the uh, head of the Atlantic uh, dug into uh, some of the launches around the glasses and then also had a whole uh, discussion in that podcast around um, the goals of the keystroke logging thing and uh, uh, it was it was interesting.
19:54
We I mean it was framed as like you know, like a tough interview around uh, surveillance in the workplace and it certainly the headlines were very scary.
20:04
I don't know where I sit on it because I've I I kind of always assumed that everything you do at work is logged in the sense that like if you're on a work computer and every webpage you visit is uh, is going through the network and monitored for traffic and security purposes and all the code you write and all the emails you write and all the documents are stored in the shared document.
20:25
It's like it's it doesn't seem that crazy to go to keystrokes uh, because everything is already so monitored.
20:32
Um, but he was framing it as more of an experiment, something that they weren't sure was going to pan out.
20:38
Uh, something that they allowed everyone in everyone at Meta.
20:43
So there were there were certain sections of the workforce that were the by default opted out.
20:49
So anyone who was working on um, uh, like confidential or sensitive information was opted out of that program by default.
20:57
Uh, he said he himself, Andrew Bosworth, was opted out of that program because uh, he has a bunch of legal holes because they're getting sued all the time.
21:05
So uh, so they can't be recording everything, I guess, that he's doing because then that would be admissible in court.
21:10
And so, all of a sudden, the the lawyer who's suing him would be would say, "Okay, great.
21:16
In the email you said, you know, we we don't want to do this, but before you >> Let's see the Let's see your writing process. >> Exactly.
21:25
Yeah, let's see what you what sentence you typed and then deleted.
21:28
Like, what word did you use before minimal impact?
21:30
Did you say medium impact or whatever? You know. So, he was opted out.
21:35
And apparently, I I think all of the meta employees who were part of that program were able to were were were able to just turn it off indefinitely.
21:44
Like, you could toggle it on and off.
21:48
And and the idea was that they wanted to collect information on how work plays out over like a 12-to-18-month 18-month period.
21:57
And they couldn't get that from any sort of data labeler because they needed to have very high-skilled workers actually chopping wood on projects for a long long time to see how projects go from start to finish.
22:13
So, so basically, like, how do you compact the longest possible rollout?
22:17
Not just a like a single chain of code, but an actual series of meetings and decisions and tradeoffs and everything that goes into making a decision in a white-collar workplace.
22:30
Like, how do you actually reason through all of that?
22:32
It's it's hard to to distill that from just, "Oh, well, the code got written this way, so that's the right way to write the code."
22:38
That that might The code might have gotten written that way because a lawyer said, "Hey, oh, we have to do this."
22:43
And then the marketer said, "Oh, well, well, you know, we have an activation with this person, so we need to integrate it this way."
22:48
And then the business people came in and said, "Oh, well, like, the margins will be better if we write it this way."
22:52
And so, it's not entirely software engineering all the time when you're actually building real products.
22:59
So, interesting to see him sort of step into the you know a tough interview and and and and and sort of lay out his his side of the story.
23:08
But, Mark Zuckerberg is in Bloomberg today pledging aggressive pricing with Meta's first pay-to-use AI, which is a funny framing for just an API for a model.
23:19
But, that's the way Bloomberg put it.
23:23
In a crowded market for AI tools, Mark Zuckerberg wants to win on price.
23:25
Meta Platforms unveiled a version of its most advanced artificial intelligence model, Muse Spark 1.
23:31
1, that includes a new paid tier for developers, marking the first time Meta has charged businesses for access to its models and providing a new revenue stream.
23:40
It'll be among the most affordable options on the market, Zuckerberg said in an interview ahead of the release.
23:47
Quote, "Since this is not an open-source model, this is, I think, the first time that we're doing a real serious API."
23:54
Referring to the API used to access Meta's AI.
23:58
And the pricing is going to be very aggressive and attractive. Makes sense.
24:01
I mean, they own the data centers.
24:02
They're very efficient at building data centers.
24:04
They should be able to serve a model efficiently.
24:07
The new model's standout improvement is is is in its agentic capabilities, the Meta chief executive officer said.
24:13
Agents are a big theme of AI this year with the label applied to systems that can can complete multi-step tasks on behalf of the user.
24:19
Zuckerberg described Muse Spark 1.
24:22
1 as having quote, "State-of-the-art or very close to it agentic reasoning and tool use."
24:26
The model is also greatly improved when it comes to coding, and Meta employees are using it internally to build products and features for various apps.
24:35
>> Yeah, my big question is how how quickly do they move all of their internal workloads onto their own models?
24:41
So, they're buying they're buying they're getting access to models through Google, Anthropic, and OpenAI. >> Yeah.
24:49
>> I think that a lot of companies will look to Meta's own actions as a way to basically validate whether or not they should be using this model themselves, right?
25:00
Cuz it was just you know, within the last month that Google had said like, "Hey, we don't have capacity.
25:06
We don't have enough capacity for all of Meta's demand for our models." >> Yeah.
25:12
>> And so, um yeah, they can't they they can't get enough AI elsewhere, at least from some providers.
25:17
And so, how much of their workloads will they be able to run themselves is a big question.
25:23
>> There are a bunch of mobile cases.
25:25
First, I'm going to tell you about Figma. Agents, meet the canvas.
25:25
Your AI agents can now create and modify your Figma files with design system context.
25:32
So, uh yeah, Meta was one of the first companies to sort of um reportedly be token maxing and have a leaderboard and all of that.
25:39
Uh if you have your own model and your own data centers, uh the incentive to token max is much, much higher because you're just paying the electricity on the cards that you're already depreciating.
25:49
So, um you should sort of lean a little bit back into that.
25:54
Not that you want to be fully token maxing, but uh you do want your employees using the tools that you've built uh as efficiently and as effectively as possible.
26:03
It's just way cheaper to explore when you're not paying margin on a another closed-source model.
26:08
And you're you're you're not paying anything else, and you're actually improving the model.
26:13
Uh so, makes a lot of sense for them to roll this out broadly.
26:16
this out broadly. Uh the uh the interesting take that Ben Thompson had, which we didn't get to yesterday cuz we ended up spending the whole interview talking about Xbox, but um the the interesting dynamic is that when you are willing to sell API access, you're
26:30
willing to sell compute uh directly, and then you're also using your own tool internally, it creates this economic incentive internally that uh you you you have an incentive to always go with the most profitable, the most the most economically efficient outcome. That can be very good for business, uh
26:49
That can be very good for business, uh very good for the investments they made.
26:55
The the trick is that you can wind up in a little bit of a situation where your business team or your enterprise sales team goes and sells all your compute capacity or all your chips and then internally your team is frustrated that they're not making enough progress.
27:11
So there's a little bit of a dance there, but in general it it it's a forcing function on the internal use of their tools to say, "Hey, >> Wait, why why is someone willing to pay five times as much than what we're willing with the value that we're creating here?
27:25
Like we spent a billion dollars on energy consuming our own LLM and someone showed up and said, "Wait, we'd pay you five billion for that same compute power to run a different model and do a different task."
27:40
It's like why is their model not economically valuable internally?
27:43
That would be the question.
27:46
The flip side is that they do have low cost so they should be able to say, >> "Oh yeah, we actually did we we yeah, we we we inferred new Spark 1.
27:51
1 internally and we improved the ad model and boom, we made a bunch of money."
27:58
>> And and these are the same tradeoffs and decisions that every lab is having to make is how much how much compute do we allocate towards research, towards internal use, towards to the API subscriptions, to free plans, etc.
28:12
>> Yeah, there was that funny semi-analysis deep dive into Anthropic's forecast and in there, I mean some staggering numbers, really really optimistic, but the the flip side was who is it?
28:24
Ed Zitron was was taking shots at the fact that they had >> EBITIT >> EBITIT, earnings before >> training >> training, infer no, training infer training inference and >> [laughter] >> and everything.
28:39
No, earnings before training, interest, and taxes.
28:44
And what was odd about it was that Ed Zitron was was was saying it's like the new community community adjusted EBITDA and it is always odd when a new non-GAAP metric pops up.
28:53
In this case I think it makes a lot of sense because training runs do fit a depreciation profile.
28:59
It's a little bit different.
29:01
I don't know why you wouldn't just put it in uh in depreciation though.
29:05
Like just figure out how to account for training runs through a depreciation schedule and then maybe it's like a non-GAAP depreciation metric but it's still in there instead of trying to get everyone up to speed on a different a different like sounding phrase entirely.
29:20
Also you could just do EBITTRA instead of EBITDA.
29:26
Like earnings before interest taxes and training. TRA. >> Yeah.
29:30
>> And that might be a little roll off the tongue a little bit easier.
29:33
>> But I think some analysts likes to be a little cheeky. >> That's true. That's true.
29:36
But the But the other funny thing is that Ed Zitron is taking shots at this like oh like the the EBITIT is like so ridiculous but like in that forecast they had like net income of a billion dollars a quarter.
29:48
So it's like it's like okay well yeah you have this like funny metric that you could value the business on to get crazy evaluation but if the business is making money and and actually like generating net income you're not in a disastrous financial position.
30:04
So you know you you could debate about the valuation but you shouldn't the whole idea of like this oh it's like like it's very it's a very very different discussion than than WeWork which was not cash flow positive which was not net income profitable and and was using those terms to sort of skirt around the losses that were accruing from the business.
30:26
So >> Yeah I was looking back at Ben Thompson's earnings transcript or a script that he wrote for for Mark Zuckerberg.
30:34
He has a good segment on why AI matters.
30:40
>> Uh, Ben writes forgive the long preamble but this is necessary context for me to properly explain why AI is so important to Meta and why I'm making the right choice to invest so heavily in both talent and infrastructure."
30:51
Um, and he goes on and on and on, but he says, "What I've come to realize as I've embraced our status as an entertainment provider and ad purveyor is that our nature as a digital business notwithstanding, we are remarkably well placed to thrive in an AI era.
31:05
Remember what we learned about humans.
31:07
They are obsessed with other humans and they want to connect with them.
31:11
That obsession and desire only going to increase as we interact more and more with AI.
31:14
AI is going to make our properties more essential, not less.
31:18
Moreover, AI is a productivity tool, but productivity is not the end-all-be-all of the human experience.
31:23
I've talked over the last year about building superintelligence that helps you get things done, but that's a business story.
31:29
What we can do uniquely is give us give people the experiences they want from connection to entertainment to shopping when they are off the clock.
31:36
The fact that we are investing in AI, but not selling solutions to businesses is actually one of our business biggest advantages.
31:42
So, of course, this is um uh just a a sort of fan fiction for an earnings transcript.
31:48
Meta is in fact selling to uh businesses now.
31:52
Um, but who knows over time how big will the API business be relative to how much uh value they can unlock across their broader >> Yeah.
32:03
>> business uh with all of their infrastructure.
32:05
>> Well, let me tell you about console. com.
32:07
Console builds AI agents that automate 70% of IT, HR, and finance support, giving employees instant resolution for access requests and password resets.
32:13
Um, we have the perfect guest to talk about all of this with because Eric Seufert from Mobile Dev Memo is with us today in the TVP and Ultra Dome. Let's bring in Eric. How are you doing? Good to see you again. And congratulations. Is it Dr. Seufert now? No. Master's, right? >> Can you hear us? >> Can you hear us? Okay.
32:40
Let's bring him back in in a second because I want to hear his take about Meta's vertically integration specifically with regard to their image model because their image model has the potential to feed back in the ads product where they are getting a signal from what ads are performing, generating new images, and then using that as fine-tuning data and reinforcement learning data for the image model.
33:02
And it's a very different cycle than what you see in ChatGPT images where, you know, they're trying to be useful, educational, sometimes funny.
33:10
Same thing with Nano Banana, but Meta has a different job to be done, a different process, and a different different goal because that model, although it might wind up being something that people use just to post on Instagram, and people might just use it like any other image model, the real killer application of that is in the ads manager.
33:29
And we also have Sean Frank coming on later in person to tell the more business side of that story.
33:39
He's built Ridge into a nine-figure D2C brand.
33:42
Obviously has been deeply ingrained in the Meta ecosystem for probably over a decade now, and and can comment on everything that's happening in AI-generated advertising.
33:52
But I believe we have Eric Seufert lined up now [music] with audio.
33:57
We'll bring him in to the TBP.
33:59
And I'll throw it on Eric. How are you doing?
34:03
>> Hey guys, thanks for having me. It's good to be back.
34:04
Sorry about the the technical mishap. >> It happens. It happens.
34:08
>> It's great to have you here.
34:08
Question, did you did you rebrand? Mobile Dev Memo? >> What No, I didn't. >> To to Heraclitus?
34:15
To Heraclitus, or is this is this an error on our side? Okay. >> error on our side. It's Mobile Dev Memo. Heraclitus is the fund. Is that right? >> Yeah, the fund. Yeah. >> Yes, got it. Anyway, >> Great.
34:28
>> we were just talking about Meta.
34:28
They just launched Muse Spark 1. 1, their LLM.
34:33
they're selling that over API they're also selling some compute but I wanted to I want your take on the image model specifically and why that is an important uh technology for them why a vertical integration makes sense there why specifically is is their image model like it it feels like it has a different business case around it than a nano banana or a chat GPT images.
34:56
>> Yes so I think this is brilliant right like this actually gives them the narrative uh firepower that they need to um to sort of undermine the skepticism that investors feel about the AI investment right so like I was at this dinner I did these dinners a lot like these ideas dinners >> Mhm.
35:13
>> like a a research uh company will bring me in and talk to like a bunch of their hedge fund clients and like I was talking about you know why these investments that meta is making right now are bearing fruit right now like 33% advertising revenue growth last quarter on 55 billion dollars in revenue that's incredible like you look at Google search was 19% Amazon which is
35:32
much smaller it was 24% they're outgrowing everyone except for Apple 11 and Reddit and people don't believe it and I asked somebody okay what would it take to convince you that these investments are actually productive in this moment in time and they said 40% 40% [laughter] growth on 55 to 60 billion dollars in revenue where did that number come from this is pulled out of thin air >> Yeah. >> This is what they need to do they need
35:52
>> This is what they need to do they need to be able to point to something and say you see that ad that was created by our AI investments when you talked about gem gem's a >> foundation model >> Mhm.
36:01
>> Meta trained a foundation model for ranking that's important but you can't see it it's hard to convince first of all the research like you know a lot of the research analysts they're really smart people but they operate in this paradigm of like spreadsheet says this I get it I can understand why ranking investments would
36:17
actually be really beneficial for the company but I put a number in the spreadsheet and it spits out something that's the tool I have to work with and like these are really smart people I think I think successfully why these are good investments but they have nothing to sort of tie it to that's quantitative. I think when you can point
36:29
I think when you can point you can point to the ad which created and you say look, this ad was created with data that only we have.
36:35
It's the image model that we built, the foundation image model that we built that is trained trained not fine-tuned but trained on our own data.
36:42
Can be verified against that.
36:44
We've got our own custom e-values.
36:45
Everything about the training was built with data that only we have.
36:50
No other frontier lab can can can fine-tune their own models for this use case. Only we can do that.
36:56
And then I think if they can point to the output and they say that ad that you saw in your Facebook or Instagram feed was created only as a result of our ability to train on this data that only we have, then I think you can kind of make the case.
37:08
So I think bundling those two integrating those two things together is like the really smart move and I I kind of you know, I understand like they you know, they restarted the the AI efforts and this is the whole you know, this is the MSL rebrand, but my sense is like this might be more convincing than anything that they've been able to say with the ranking infrastructure and the transfer learning infrastructure.
37:27
I'm talking about Lattice, I'm talking about Gem.
37:30
So, we'll see, but my sense is like if you can actually point to some output and say look, this only exists as a result of our ability to train this model on this proprietary data that no one else has, my sense is you can make the argument more of robustly.
37:43
>> And if you ask advertisers what is the bottleneck to spending more on Meta, they will almost always say it's great creative.
37:53
And so there's there's a there's a path to sort of just like unlocking and removing that bottleneck so that the constraint just becomes how much revenue do you have and that will just become a proxy for how much you can spend with us.
38:07
>> Yeah, it's like what's your bid?
38:07
Well, your bid should be the value that you get back.
38:10
I mean, that's like the whole point of a second price auction is that you should build your bid your true value because you're going to get you're going to make money if you win.
38:15
But like the thing is like they built like a lot of their initiatives, right?
38:18
So the Gem foundation model for ranking, Andromeda is is is a whole system for doing retrieval.
38:24
And then Lattice, which is transfer learning.
38:25
But like the whole point of Andromeda was they were reacting to a lot more creative being deployed.
38:29
Now the creative being deployed though was creative with third-party tools.
38:32
Right, this creative being deployed is being built without the benefit of the actual performance data.
38:38
If you actually want this to work, you need to train it on the ROAS.
38:41
You care about the end result.
38:43
You care about what the person does when they go to your website or your app.
38:44
And that no one has that except for Meta at that volume.
38:49
Because they get it passed back to the capping the pixel. Right?
38:52
And so their ability to build this foundation model, I think really unlocks a lot of value.
38:57
And you know, I think you'll probably see that show up in you know, in the in the revenue growth.
39:01
Now do they hit 40% to satisfy these these needs of these hedge fund people? I don't know.
39:05
But like my sense is if you can point to the output and you can say, "Look, this addresses the core bottleneck, which is we need a lot of creative, but it needs to be built with the knowledge of what actually drives the outcomes and not just a bunch of variations that clog up the system." >> Mhm.
39:19
Um I want to push back on that idea that no other lab can do anything like this. What about DeepMind? What about YouTube?
39:26
Nano Banana VO3 video generation.
39:26
It does feel a little bit farther off.
39:30
And also it feels like the hedge fund analysts that you're talking to aren't asking the same questions of Google's investments in AI because they have such a incredible business with Google Cloud Platform and they're able to strike these massive compute deals so they have some off-take there.
39:49
But um how is how is the situation different in Google?
39:56
>> Cuz they don't have a history of tilting at windmills.
39:57
They don't have the they don't have an albatross around their neck, which is metaverse. That was a misadventure. It cost a lot of money.
40:03
It never resulted in anything meaningful. Right?
40:06
Now I would actually make the point that that whole rebrand was just a distraction. It was a smoke screen.
40:10
They had to get away from the whole Facebook files thing.
40:13
And that took everyone's eyes away from you know, that that scandal. >> Mhm. >> And was it worth it? I don't know.
40:18
If you achieve that, which they kind of did, they kind of did that with the meta rebrand.
40:23
Um maybe it was worth it, right?
40:25
Um you know, now it's trending movies and now >> even if you ignore like the actual investments they made in in metaverse and like horizons and things like that, like meta platforms is is the metaverse.
40:36
Like it is the place that people exist online.
40:38
So like the name makes sense even if you ignore that and it makes sense for the strategic reason, like he said, to get away from the Facebook files.
40:44
So like they could have just rebranded and never done the metaverse and they'd be in a much better position, right?
40:50
They'd have that >> Yeah.
40:52
>> uh sort of sort of >> But you can't really do the rebrand unless you tell the story or else everyone accuses you of what you just described. >> Right. Yeah, yeah.
40:59
I mean you're going to put your money where your mouth is. >> Yeah.
41:01
Um but uh on on video generation in particular, do you feel like we are further out uh just further away from that?
41:10
I was I was uh demoing the latest uh VO model and it's good, but it's still clockable as AI generated.
41:16
They're still I had some cars spinning around and you know, it's three cars and then it's four cars and then it's two cars and they're sort of melding into one another.
41:26
It's incredible video model, definitely state of the art, but uh I don't know that that that's ready to be deployed in YouTube across uh a ton of video ad impressions.
41:35
So do you have a timeline for that or do you have some thoughts on uh when that will actually be important to the business?
41:42
Because you have to imagine that Instagram video ads perform better than just image ads and so they'll try and do this, but what what what how are you seeing the AI video advertising model evolve?
41:56
>> Here you got to look at ByteDance.
41:56
What ByteDance is doing is incredible on this front, right?
41:59
So they put out this paper a month ago.
42:01
I did a summary on Twitter and and LinkedIn, but um what they're doing with TikTok Shop is these real like, you know, basically photorealistic 3D avatars that are selling stuff, right? Like so infomercials.
42:12
And they've you they built custom models to build those ads. And those are all ads.
42:16
A lot like I mean not all of it, but like you if you go on TikTok and you're looking at the TikTok shop stuff, a lot of it is AI generated.
42:22
And so, what they they in this paper that I summarized like they invested in, you know, in in essentially fine-tuning this model to make sure that there was no collision with the hands.
42:33
What they were finding was that like so when you get in that uncanny valley um situation where people can tell it's AI, then they turn off.
42:38
Like there's a lot of research that's been done on this.
42:39
If people know that it's AI, they penalize the ad.
42:42
But when they don't know it's AI, the AI ads outperform the human creative ads.
42:45
It's It's really fascinating.
42:46
But so what they found was like when you saw the collision between someone holding something in their hand and the object, then people the the the click-through rates dropped, the conversion rates dropped.
42:53
But so what they did was they fixed that.
42:54
So they built this whole like visual um interpretability model that just focused on that with it with with an expert like in the model.
43:02
And so it just addressed the hands.
43:04
And so, like if you but that's use case specific.
43:06
You needed a general purpose model that's going to build photorealistic video, we're probably pretty far off of that for like all ads of all types, but I think with something like, you know, okay, well, we need human photorealistic kind of like infomercial style, I think you could get to that point now.
43:20
And maybe we were there now.
43:21
And maybe ByteDance is there now.
43:23
But I think it takes a lot of investment, right?
43:25
I mean they had something like if I remember correctly from the paper, it's been a while since I saw it like 12,000 hours of live human product interaction.
43:31
So I mean it takes a lot of data to do that, right?
43:33
And so, you know, it's just it's whatever you want to invest in.
43:36
I think if YouTube wants to do that for a general purpose photorealistic video ad tool, we're probably pretty far off. >> Mhm.
43:44
>> Uh >> Uh how do you think that the market will react if Muse 1.
43:47
1 is like a very much like a base hit on the API, where they they get some comp you know, big companies move over some workloads, but it's not, you know, this runaway hit.
44:01
And And I say that because so many models that have been good, not great, have a little demand just because there's a lot of demand for AI, but but they don't sort of like have these sort of breakout uh revenue charts.
44:17
>> Well, you know, they clearly are going to be very aggressive on the pricing.
44:18
I mean, they talked about that today, right?
44:20
And so, my sense is like what you're going to start seeing is that people don't need to operate at the frontier, and you have a lot of use cases that work just fine with and and and basically they're just good enough, right? With some legacy model.
44:30
And so, it just comes down to then, okay, well, that's commodity.
44:33
And so, are these priced like a commodity?
44:34
So, like like if you if you think about like and and also like I think we're going to see a lot more um uh like people relenting from needing to be on the on the frontier when they've built stuff using a model that then gets upgraded.
44:47
And the whole idea there is like, well, am I going to upgrade this tool because I have to adapt it to the new model, right?
44:52
Like if I'm if I if I, you know, cuz essentially like the model name it like if you if you use like Vertex AI, right?
44:57
You just got a model name as a variable.
45:00
Like you're sending this, you know, system prompt to to, you know, Google, but like it's just a variable.
45:05
You can swap that out in 30 seconds.
45:06
But the the fact of the matter is you're sampling from a new distribution if you do that.
45:09
And it's going to change the output.
45:10
It's going to qualitatively change the output.
45:11
And it might change it in quantitative ways that like with retention and engagement.
45:16
And so, the thing is like if okay, well, now we're talking about this big cycle.
45:19
It's a new product development cycle because I have to adapt this product that I built to this new model and the output that it provides, right?
45:24
And maybe it was working perfectly.
45:25
It was working exactly as I expected it to before.
45:29
Now I've got to invest a bunch of hours, a bunch of engineering time in adapting it to this new model.
45:32
So, even if it's just a swap of a variable name, it's still a whole lot of testing, QA, determining like how that impacts long-term retention, going to do AB test.
45:39
So, my sense is like you're going to see a lot more people just saying, "No, this works fine. This is perfect."
45:44
And and getting more getting like more robust output, let's say the token price is exactly the same.
45:47
Getting more robust output wouldn't benefit me.
45:49
Why am I going to invest the resources into adapting to the new model? >> Mhm.
45:54
>> Yeah, but isn't that So, so, if they're going after workloads that are running on old models that are working fine.
46:03
And it's a lot of work to switch over.
46:06
>> It's not a lot of work to switch over.
46:07
>> Not it's it's fast to switch, but then again, you have to go through this process of like Q A'ing and running it through your own benchmarks and all this stuff.
46:14
Like the question is like how I I I just don't know like I I'm I'm thinking about a scenario where like a year from now we're sitting here and like meta has a $3 AI API business and the analysts that you're talking to are like that like to me they're like that doesn't get you to 40% year-over-year revenue growth on the on the business overall.
46:39
So it like doesn't solve at least what those analysts in particular we're talking about.
46:42
And like zero to three billion on like a new business line would be crazy and is like you know, only been done a handful of times throughout history over the last few years, right?
46:53
So um I'm just saying like we're we're there's such big numbers now that there's possibility where you have like this incredible breakout revenue growth, but it doesn't actually move the needle enough that that the market still says like hey, uh we're we're not super confident about about about like the next capex cycle, right?
47:14
This 2027 numbers that are coming out.
47:17
>> Well, that's and that's the problem with this whole business line in the first place.
47:20
Like I think it's a mistake.
47:20
I think it's a capitulation.
47:21
I think you're going to get much more value out of that compute if you apply it to your own core business, which is advertising.
47:29
My sense is you get better growth, but like the problem is the investors don't buy that right now.
47:32
Like they've got a narrative issue.
47:33
It's not a it's not a productivity or a competency issue. It's a narrative issue.
47:37
And like the problem is like, you know, and you know, you you you cited um Ben's brilliant essay from yesterday or the day before about like what Zuck should say. He should say that.
47:46
Like he Ben is totally right. He should say that.
47:48
He should come out and say, "Look, we like Senator, we run ads. Senator, we run ads."
47:51
When he said that, I was at F8.
47:54
It was like the next month.
47:56
Every Facebook employee is wearing a shirt that said, "Senator, we run ads."
47:59
They know what business they are in.
48:01
Zuck seems to be confused about it.
48:02
Like, I don't understand what he's talking >> Well, here's Here's a big question that'll be interesting.
48:06
>> [laughter] >> This year they're going to spend, I don't know how much they're going to spend on external models.
48:09
Like, I would say I would I would expect them to spend like maybe like 10 billion dollars, right?
48:14
Like, something in the range of 10 billion dollars from from Google, Anthropic, and OpenAI.
48:19
If next year they can say, "We're not spending money on any external models," >> Mhm.
48:26
>> then then that could help them with the narrative issue of saying like, "Hey, like we're we're invest we're basically getting instead of having to give this money to other businesses, we're just using our own infrastructure.
48:36
It's a lot more efficient. We can like token max.
48:37
We can use way more tokens.
48:38
We can do way more workloads.
48:42
Um and so that's potentially it's potentially setting But but the question is, can they actually move all the workloads that are on these other models to their own models?
48:53
>> Well, so that's where you actually do need the frontier, right?
48:54
Like, coding tasks, like you actually benefit from having the frontier.
48:57
But if you're talking about customer support stuff, right?
48:59
Like, that doesn't need a frontier model.
49:00
That could use like a three or four, you know, sort of like uh release back model.
49:04
And it'll be it'll be producing reliable results that you know work.
49:09
Like, you've measured you've tested, and you don't need to upgrade like the customer support or like the chatbot, you know, for customer support uh integration to the to the the bleeding edge model every single time.
49:18
But like, coding, yeah, if you're actually using these models to build models, you probably want the best of the best, right?
49:24
>> And like, what Meta's doing is really actually at the frontier with like integrating um agents into the coding workflow.
49:28
If you look at like their system they built called Confucius, it's like a self-learning agent like that actually helps them deploy better.
49:34
They've got a whole pipeline for uh data science and machine learning tasks that helps them decide like, "Okay, which which which Where should we even apply this?
49:41
Where should we even do testing?" Right?
49:44
Because that actually takes a lot of time.
49:45
Like, you're just figuring out what kind of experiments, what kind of test you want to run.
49:47
They built a whole pipeline around that that's all driven by agents, right?
49:50
So, my sense is like, there's where you want the the the top of the line, and maybe their own models don't perform best there.
49:55
Um but like also, I don't know how how excited investors are going to get when you say that we cut expenses.
50:00
I think they really need to see the revenue growth at the top line.
50:02
And so, my sense is like you get that you get more of that by just making the ads platform better, and they've done that.
50:08
All they need to do is say, "Look, we can If they could forecast out that growth and say, 'Look, we're really dedicated to this.
50:12
This is what we're pointing everything at.'"
50:13
My sense is you could get investors excited over time.
50:17
You keep printing 33% or whatever every quarter, like you're going to get investors excited after some time.
50:21
If you start saying, "We're going to compete with Call C.
50:22
We're going to build an AI pendant hardware."
50:27
Like they're not going to get excited.
50:28
They're going to think you don't know what to do, and they're going to think you've got all this compute capacity, you don't know how to use it.
50:33
>> Help uh help us understand the the prediction markets play.
50:35
Is that I think the the answer that we landed on was it is just in Meta's nature to copy the new hot thing.
50:43
So, regardless of what it is, we're just going to do it, right?
50:47
And you can see the history of like these shots on goal with like every new hot thing in consumer, they just build a version of it or they try to buy it.
50:52
So, I think that's the most simple explanation.
50:55
Um John was trying to explain it as like maybe there's some way to do it with like your either there there is no dollars, uh and it's just for like social status, you know, who who can be you know, an oracle.
51:07
Uh I think that that was we got more news that maybe showed that that wasn't the case.
51:12
wasn't the case. Um but then when you look at again, you look at the market, like you look at the total market for like gambling, and again, even if they got a meaningful amount of that market, they would not
51:25
really move the needle in the way they need to on the core business, and they would invite all these new regulators that are now saying like, "You're not only trying to harvest, you know, my teenagers' attention and making them, you know, sad about their life. You're
51:35
You're also getting them to like it just feels like it opens up this huge can of worms for no reason."
51:42
>> I it's just a totally misdirected move.
51:44
Is Is really the space you want to go into right now?
51:46
That is so politically fraught.
51:48
That is such a political hot potato.
51:50
Why do you even want to touch that?
51:51
I would steer clear of that.
51:51
I would say, "Look, Facebook apps, that's time well spent.
51:55
You connect with friends.
51:56
All this gambling, you're going to get addicted to that. Don't spend time there. Spend time on Instagram. It's more wholesome.
52:00
Why are you going to touch that at all?
52:02
Are you just going to open the door to more scrutiny? That's insane.
52:05
I really have no idea why they're even talking about that.
52:07
It doesn't make any sense.
52:09
But they've got like, you know, look, they said, "We're going to We're going to publish a lot more apps."
52:12
They've got the new pocket app.
52:14
I mean, it seems kind of interesting.
52:15
Maybe some of the stuff sticks.
52:16
I think they could just try a bunch of stuff.
52:17
But why would you touch the most politically toxic area right now?
52:21
Like when you could just be touching anything else? >> Yeah.
52:24
Uh take us through the Prosperous Society.
52:26
I wanted That's what originally why I wanted to bring you on. I want the thesis.
52:29
I want to dig into it uh because I I I found the It's a 3-hour, four-part podcast series.
52:34
You've written a lot about it, but uh introduce it for those who haven't been following along.
52:40
>> Yeah, Prosperous Society I started out I wanted to write um an economic bull case for AI.
52:44
Yeah, we've heard all of the you know, bear cases.
52:46
We've heard all of the you know, doom narratives around the around the economy.
52:49
It's going to just basically displace all white-collar work.
52:52
You know, you're going to get DoorDash created with a five-coding session.
52:55
And so, all these companies going to go out of business.
52:57
And I wanted to make the case that this is probably not going to This is probably not going to happen.
52:59
And actually, there's a lot of reasons to be optimistic, right?
53:01
And I think, you know, in writing that, it ended up becoming you know, we live in a in a sort of like very pivotal moment, I think.
53:07
You know, if you look at the elections um you know, in the sort of the the house primaries in New York, you look at what's happening with the the you know, the New York City uh mayoral election.
53:14
And you look at what's happening in the LA um mayoral election.
53:17
Like it it there's there's this this this sort of moment where these sort of like um impulses against capitalism have become a lot more popular, right?
53:24
And like there's reasons for that.
53:25
And you know, I you know, I'm I'm not an expert on those reasons.
53:27
So, I won't delve into them.
53:29
But I think, you know, ultimately, it's a mistake to go down that path.
53:30
And the thing is like my sense is a lot of the AI or the anti-AI narratives are actually have nothing to do with AI, right?
53:38
That's just seen as an avatar or like a bogeyman for capitalism.
53:40
And so, what I wanted to do was sort of like anchor this economic defense of AI, this economic bull case of AI in, you know, the sort of like liberal tradition of of of the Western world.
53:50
And and and in in doing so, like you could say like, you know, you could anchor it to these these sort of like these these great thinkers, you know, these sort of like Enlightenment thinkers and and these sort of the economic giants that have built, you know, that built this sort of intellectual framework that our that our Western civilization is is based upon.
54:06
And people could say, "Well, look, well, look, you've you've misinterpreted them."
54:09
And so, okay, well, maybe, but but if if that's not the case, then they you're going to make them drop the mask.
54:14
You're going to make them drop the mask and say, "No, that's not my problem with AI.
54:17
I just don't want to live in a liberal economic society based on these Western thinkers."
54:21
And I think if you actually kind of force that to be articulated out loud, you make a lot of progress against the anti-AI narrative.
54:28
But the whole point of the Prosperous Society is, my sense is, you know, a lot of these AI investments, they're going to push the economic constraints away from production and towards just distribution. Right?
54:39
They're going to make distribution the binding constraint.
54:41
And so, cuz you just have this this this flourishing of of content creation.
54:45
And and and and so, when that becomes the actual problem with distribution and these AI investments go into things like ads platforms, digital advertising, you know, rec sis, recommended recommendation systems, then actually commerce, the economy becomes much more efficient.
55:01
And it's a really good thing.
55:03
And you just generate a lot of value by pushing that binding constraint to the distribution layer.
55:08
And you get then as a result, you get a lot more heterogeneous product development because you can actually reach those people economically, right?
55:14
So, like, what is the constraint now?
55:16
It's like, "Well, can I reach a big audience?" Right?
55:17
Can I reach a big enough audience to support a business because well, I've just got this kind of like blunt tool.
55:21
But if these AI investments go to making recommendation systems better, digital ad systems better, reaching these these these these pockets of people that have these very specific interests that were totally unserved before, then you enable a lot more commerce, right?
55:37
And and then and then you know, you support this like flourishing of people making this wide diverse variety of goods.
55:43
You get rid of this idea of like the Pareto principle.
55:47
We have to serve, you know, the the 20% that supply 80% of the commerce.
55:48
Well, no, now you can serve everybody and they can pay what they're willing to pay for these things.
55:53
You reduce consumer surplus uh sorry, you reduce um consumer surplus.
55:57
You just you create this flourishing of of of everyone getting exactly what they want, right?
56:01
And that's the prosperous society.
56:02
And so the way I frame it is kind of a reaction or like call call it a conversation with John Kenneth Galbraith.
56:09
He wrote uh The Affluent Society, right?
56:10
But this was written in the post-war economy.
56:12
It serves kind of as the Degrowth Handbook.
56:16
And in my sense is like, you know, John Kenneth Galbraith is a brilliant man.
56:19
I'm not saying he's wrong or he was wrong when he wrote the book, but I'm saying it just doesn't apply anymore, right?
56:23
He was talking he had this idea of the dependence effect.
56:24
Advertising actually is a way to whip up demand so we can maximize production because that was what he called the conventional wisdom at the time.
56:32
You should be maximizing uh production.
56:33
But the reality is in the time he wrote this book, 1958, you had people moving to the suburbs, getting big houses, they you know, the GI Bill helped them buy these homes.
56:41
You had the suburb the the idea of the suburbs was was being deployed.
56:45
And so, you know, people needed washing machines. They needed cars.
56:47
They needed refrigerators for the first time.
56:48
And so, you had these big companies that made these mass market goods.
56:51
They advertised in mass market media.
56:53
And they and and the and John Kenneth Galbraith's idea was like, well, that's just creating demand.
56:57
The the there is no actual inherent demand for these things.
57:01
It's creating it through advertising.
57:03
And my point is the opposite.
57:04
You know, we don't have this homogenized society anymore and we don't have people that need these homogenized goods anymore and we have a lot more particular specific media now.
57:12
And we can reach people and advertise to them the things that they have demand for for products that that weren't economically viable prior to these systems, these distribution systems.
57:23
And that is the prosperous society.
57:25
It's being able to reach people to meet the demands that they have with the products that they couldn't access before with app ads they wouldn't otherwise see apps in these systems.
57:34
And so I think it's it's you know I think it's very it's my sense is like you can make a very credible bull case that that's what AI delivers to us and it's not about wiping out white collar labor because the reality is like that's going to create more jobs and we're seeing that now like there's no there's no justification for that skepticism. It just doesn't exist.
57:52
We're seeing an increase in hiring maybe not at the entry level and you can discuss if there should be some intervention there but my sense is like AI actually if you look at the data and there's a Financial Times article about this the other day.
58:04
If you look at the data it doesn't support that bear case and so that bear case should be absolutely eliminated as something that even enters the conversation.
58:12
>> Yeah, no I completely agree.
58:12
That was an amazing speech.
58:16
I don't know if I have anything to add.
58:18
Yeah, no I I I love this idea.
58:18
We've been talking about it a bunch just more customization the long tail of commerce getting even longer and it feels unfathomable there's already like you know specific shirts that are just for you designed and targeted to you on Facebook.
58:31
We've seen that but like it can in fact get more personalized.
58:36
>> Yeah, I mean we've we've seen this with like content in these platforms right?
58:40
They're very good at they're very good at like find they're they they they are very good at serving you.
58:43
They can serve you a video that has 50 views from a new channel on YouTube and you'll be like that's interesting. I will watch this right?
58:51
And it and it and it's a niche that is like so small it never could have existed in the era of you know radio and television and you know seeing that seeing that trend accelerate.
59:00
It reminds me I was um I I was wanting you know the you know these like kids RC like ride on cars.
59:07
I got like incredibly frustrated with these cuz I've tried a bunch of the different brands.
59:13
I've tried spending like $800 on them and you know $400 and all of them just suck like the kids even when you have the you're driving the kid on the controller the kid can still hit the gas and just like run into stuff and you know it's just absolute chaos.
59:27
I'm like, what is the version of this that is like, you know, uh you know, the the top-of-the-line version of this? Cuz I want to get it.
59:33
I I use these things a lot.
59:34
And I searched around, couldn't find it anywhere, and then John just like >> Within 24 hours, I got served exactly what he was looking for.
59:43
>> [laughter] >> I don't even know how how it got served to me, but uh it was fantastic.
59:47
Uh thank you so much for taking the time to come chat with us. >> Electric.
59:50
Always Always a great time.
59:52
>> My favorite My favorite conversations.
59:55
And when you're on, it makes me feel like we're on we're on we're actually, you know, on SportsCenter. >> Yeah.
59:59
>> [laughter] >> Because most people that are talking about this stuff are not high-energy, and you're just like full-on SportsCenter. It's amazing. I love it.
1:00:07
>> Well, congratulations on the progress and uh and the Prosper Society. Go listen to it.
1:00:11
It's a 3-hour, four-part series, uh and sign up for Mobile Dev Memo if you haven't already. Of course, you should.
1:00:16
But thank you so much for coming on the show, Eric. We'll talk to you soon. Take care. Have a good one.
1:00:20
Let me tell you about public.
1:00:21
com, investing for those who take it seriously.
1:00:23
They got stocks, options, bonds, crypto, treasuries, and more with great customer service.
1:00:26
And our next guest is Bert from 1X.
1:00:31
He's the founder and CEO. How you doing?
1:00:34
>> What is that behind you?
1:00:36
>> Welcome back to the show. Thank you so much. That is incredible.
1:00:39
>> hand, and then it's like Neo's hand. >> Yes.
1:00:41
Introduce uh the launch today. What happened? Tell us about it.
1:00:45
>> I mean, we've been cooking on this for quite a while.
1:00:47
So, super excited to finally show this to the world.
1:00:49
And also, it's just so exciting because we're so close to shipping now.
1:00:52
So, everyone's going to pick this apart anyway.
1:00:56
So, we don't need to be careful anymore, and we can just like open it up, and to me, it's this beautiful machine that becomes almost more art than engineering, right? At this point.
1:01:04
And uh Yeah, we're we're excited to show people what we've been cooking, and uh excited to put it in people's hands.
1:01:13
And I think also uh it has a special place in my heart because we've been working on this problem for more than a decade, and uh one thing that Neo and One-X has really been pushing is how do you use highly miniaturized high power motors and tendons to create these machines that mimic humans and the hands is the kind of culmination of all that work, right?
1:01:35
It's It's where all of the complexity comes together to to meet the world and hopefully we've created something here that can really remove that final barrier for how intelligent our models can become, right?
1:01:46
Like so much of human intelligence comes from our ability to to probe the world for truth and to really figure out how the world works through our hands.
1:01:56
>> Okay, take us through the full journey of developing hands for Neo.
1:02:03
Uh a lot of people in tech love to you know follow an Elon style playbook.
1:02:06
Just make the most simple version of something.
1:02:11
Simplify, simplify, simplify.
1:02:13
This looks incredibly beautiful, but also incredibly complex.
1:02:15
And so I want to understand like how you how you got here basically like version by version.
1:02:24
>> So it is a very complicated hand, but in my opinion it is the simplest version of it that exists that is good enough to do what needs to happen.
1:02:35
So first of all, One-X is fully vertically integrated.
1:02:37
So we do absolutely everything in house and in our factory here in California where I'm sitting right now.
1:02:44
We do everything from designing the production processes to producing our own motors, our own tendons, the entire system, right?
1:02:50
Sensors, electronics, everything in house.
1:02:53
And that allows us to iterate very very fast.
1:02:55
Because I do really believe in like you have to have this first principles approach, right?
1:02:58
So start with like what what am I trying to solve?
1:03:00
Uh we're quite lucky that we have humans to look at with respect to how do you solve this problem?
1:03:06
Nature did a pretty good job.
1:03:08
>> So when you say you're you're you're looking at humans, uh what type of training data are you using? What's useful?
1:03:14
What are you discarding from just There's a lot of hand videos on the internet, I'm sure, people doing all sorts of things.
1:03:20
You can do teleoperation.
1:03:22
You can have people wear gloves and do motion capture. You can do simulation. >> Yeah, x-ray.
1:03:27
I mean, there's so many different ways.
1:03:29
Like, are you using everything?
1:03:30
Is there one one path that you found specifically valuable?
1:03:37
>> So, I think actually it starts from like starting with first principles, right?
1:03:40
So, what makes us always been about we want to sound robust and safe.
1:03:43
So, they can live and learn among people.
1:03:45
But also because safety is what allows us to learn.
1:03:49
So, we probe the world for truth.
1:03:52
And of course, if our fingers break while doing that or we break whatever we're trying to touch, it doesn't work.
1:03:56
So, you need to design these beautiful kind of like compliant soft systems that force can flow both ways because you're both seeing with your hands and acting with your hands. >> Mhm.
1:04:07
>> And that's really what this is all about.
1:04:08
So, like, how do you create that?
1:04:10
And you kind of have to look at how nature works.
1:04:12
Like, our muscle not really moves fast.
1:04:14
There's no gears, no nothing like this.
1:04:16
So, we've designed this from these first principles.
1:04:18
But we go way deeper than that.
1:04:22
I think what we haven't talked enough about yet and we'll share more about this later is like how incredibly seriously we take closing the gap towards the human.
1:04:31
And it's not to look like a human.
1:04:33
It's because we want the work system to work like a human.
1:04:36
So, even like if you look at Niels' hand behind me here, we worked so deeply on how do you make these fingers non-linearly just like be compliant exactly like a human finger.
1:04:47
Because if you get all these details right, you can take all of the video that's out there on the internet.
1:04:52
You can train huge robots based on this and it just works on a robot.
1:04:54
And that's what the One X robot lab is about.
1:04:56
So, to enable that general intelligence for robotics, you need to design the robot so that it interacts with the world exactly like a human.
1:05:04
And then you want to do that with the least amount of complexity possible.
1:05:06
And that's essentially what we have here.
1:05:08
But of Of the complexity of that is pretty high because you're you're mimicking a human hand. >> Yeah.
1:05:13
>> talk about grip strength.
1:05:13
We got grip strength testers here in the studio.
1:05:15
Uh is this an important benchmark?
1:05:18
Clicking these together, how strong is the hand currently? Where do you want to go?
1:05:24
Because it feels like there's a trade-off there where if the hand like the stronger you make the hand the heavier the more uh the more dangerous it could potentially be.
1:05:32
At the same time, there's certain tasks that you expect a certain level of grip strength. >> Also, yeah yeah.
1:05:38
Yeah, it's be interesting to understand how often tasks come up in your daily life where you need like insane grip strength, right?
1:05:47
>> pretty rare, I think, but certainly in you know, industrial capacity or even around the home picking things up, moving a chair, you need to be able to grab it without dropping it.
1:05:55
It's a safety issue at at the end of the day.
1:05:59
>> Yeah, I think it it actually appears quite often, but you don't think too much about it because you don't do it for a long period.
1:06:03
Like yeah, you're not grasping that hard, but then something start slipping and you tighten your grip or like you're actually using quite a bit of force.
1:06:09
So, the hand is roughly the same. Yeah, that's a good one.
1:06:13
The hand is roughly the same strength as an average human. >> Really?
1:06:18
>> So, and that's it so you I mean it needs to be able to do the full capabilities of a robot, right?
1:06:24
So, the robot can deadlift 150 lb.
1:06:26
So, the hands need to hold the bar of 150 lb.
1:06:28
Not because deadlifting is useful in everyday life, but because it's a good metric for like how capable we are. >> Yeah.
1:06:35
>> So, we really worked hard to make that kind of power to weight ratio also about the same as a human. >> Mhm.
1:06:40
>> So, if you look at the general hands in the market right now, this thing is roughly three times as high force as the other hands.
1:06:48
And that is really also something that's going to enable a lot of new applications, right?
1:06:51
Because in the end your AI will be as smart as the diversity of the experiences that you have lived and experienced.
1:06:58
Like diversity of data is directly correlated with the intelligence of your model.
1:07:02
And if you are as a third as strong as a human in your hands, there's a lot of tasks you just can't do. >> Yeah. Uh I have one more.
1:07:12
Um it feels like you have jumped to the frontier of hands specifically.
1:07:17
Uh is as I saw people joking, can I just buy the hand?
1:07:22
Obviously, they're making probably rude jokes, but um is there a world where you partner with other robotics companies to sell a piece of your hardware, maybe just the hand to someone else that already has a wheeled robot, but it needs a hand?
1:07:35
Is there a world where you're selling parts of your technology, or do you want to be vertically integrated from end to end, the full experience?
1:07:46
>> I I think there there is a world like this.
1:07:48
Uh I do think it's very important though that like we want to we're we're about to also launch Neo as a platform, where we we're going to invite everyone in to build on this.
1:07:55
And having like a homogeneous platform that everyone is building on is so incredibly powerful because that doesn't exist today, and that really allows you to do benchmarks across systems like you have in the rest of the AI community.
1:08:06
Um but that being said, it's not a hill we're going to die on.
1:08:10
Like if the if the collaborations are the right types of collaborations, we just want to make sure we can scale our manufacturing and get as many out there as possible, and build the ecosystem, and really give robotics all the love it deserves, right? >> Yeah, totally.
1:08:24
>> to accelerate the path. >> Yeah. Amazing.
1:08:27
Uh timeline around uh shipping, what's what's the what's the update there?
1:08:30
I know a bunch of people that are that are in line that have that have ordered, so uh everyone's very >> So uh yeah, I'll be kind to my team and not give you a specific date, but we have promised that we are going to ship this year, and we will ship this year.
1:08:43
So we're going to keep that promise, and it's going to be incredibly exciting.
1:08:47
And like I said, like the reason we can be so open, right?
1:08:49
So you can just read into that.
1:08:50
Like I said, the reason we can be so open is that this is about to ship.
1:08:53
So people will pick it apart anyway.
1:08:54
And I do think uh this is going to be so big, right?
1:08:58
Like as AI now becomes physical, it's really hard to understand what kind of impact that will have. >> Mhm.
1:09:04
>> Um we're getting so much interest from, let's say, uh wet labs that want to have their AI for science actually design, manufacture, and run their experiments.
1:09:15
Um there's like hospitality, uh elderly care.
1:09:18
You have the home that we're already working towards.
1:09:20
Like there's this enormous surface area.
1:09:21
And uh it's going to happen a lot sooner than people think.
1:09:27
And I think uh right now it's just about the really growing the pie and making sure that everyone has uh platforms that they can work on to to solve these hard problems. >> Yeah. Amazing.
1:09:39
How will how you I mean the last question I have is like like th- this this feels like a uh a technology that even after you solve development, design, and the the the AI that powers all of this, like it is much more gated by the real world and thus we would see like a slower takeoff like what we've seen with Waymo.
1:09:58
It's it's you know, everywhere in San Francisco, but as you go around the world, you don't realize that cars can drive themselves.
1:10:05
Um whereas, you know, chat. openai.
1:10:08
com was available in every country and it was just like the Turing test is passed for everyone at the exact same time.
1:10:14
And that's that feels impossible in robotics in the physical world, but do you have a different view of it or am I roughly correct with that prediction?
1:10:26
>> The ramp is going to be slower, but the total uptake is going to be way way way higher. >> Sure. >> Right?
1:10:30
So like if you think about and I'm I'm very bullish on this.
1:10:32
Like I think it's just a couple of two to three years away, but even if it's a decade away, like robots will build robots. >> Yeah.
1:10:41
>> And we're already working on this in the factory.
1:10:42
But they won't just build the robots.
1:10:43
They'll build the data centers, the chip fab, the energy infrastructure, getting into mining and refining. >> Yeah.
1:10:48
>> And this full automation of the physical substrate that enables everything including intelligence. >> Mhm.
1:10:54
>> That can only happen with robotics. >> Yeah.
1:10:56
>> And that's going to look like this, right? >> Yeah.
1:10:58
>> So, you you need you need to kind of like enter that curve.
1:11:00
And I think the optic ramp is going to be slower in the beginning, but way way way higher as you kind of like hit the vertical on the curve. >> Yeah.
1:11:08
>> And I think this is also where it gets extremely interesting, right?
1:11:10
I'm back to like how we're going to solve some of the remaining problems in science, how we're going to create an actual true abundance of labor across society. >> Yeah.
1:11:22
>> Um this is only possible if you automate the physical substrate.
1:11:24
So, it's going to take slightly longer, but it's also worth it because the impact is going to be so tremendous.
1:11:30
>> Yeah, and it's it's still should be an exponential curve because once you get to the point where uh you know, five robots can make one more robot in a month, then you wind up compounding and the exponential just grows and grows and grows. Fascinating. Very exciting times.
1:11:45
Congratulations and thank you so much for coming on the show.
1:11:48
>> excited you guys shared this.
1:11:50
You guys continue to have the the best aesthetics and robotics by by 100X.
1:11:57
>> Yeah, it makes me feel more much more C-3PO than uh Terminator, which I think is the right the right direction to go.
1:12:03
>> The hand's a little Terminator. >> 100%.
1:12:05
>> The hand's a little Terminator, but you know >> Well, one should not underestimate how important it's going to be to do this together with people in the sense of like adoption needs to uh come through making everyone used to this technology, right? >> Totally.
1:12:19
>> Uh we we we want to make sure everyone understands how helpful this can be and really make sure that we don't don't hit any barriers where like this becomes something that people don't want because it's such a great opportunity and we want to make sure we can accelerate the path.
1:12:32
>> Yeah, VR uh you know, VR was useful in in certain pockets, but it was it was awkward and it was never adopted and it was always seen as like this this like uh yeah, very like uh niche technology and I I think the uh the aesthetics are underrated, so congratulations on nailing them.
1:12:48
Thank you so much for coming on the show. >> Cheers. Great >> stuff. >> Have a great day. >> Awesome. Thank you guys. >> Have a good bye.
1:12:54
>> Let me tell you about Codex.
1:12:54
Codex is a powerful workspace for getting work done with AI agents.
1:12:58
Whether you're writing code, analyzing data, creating content, or automating business workflows, Codex helps you move projects forward from start to finish.
1:13:05
And we have Teemo joining in just 40 minutes or 30 minutes to give us the update on 5.
1:13:10
6 and what's going on in Codex, of course.
1:13:12
But first, we have Josh Lindgren from CAA, who's the head of podcast development here to give us an update on all things >> Suited up. >> Podcasting.
1:13:21
Suited up, looking good, Josh.
1:13:22
[music] How are you doing? Welcome to the show. >> I'm doing well.
1:13:24
You guys are always looking good as well. >> Yes.
1:13:27
Uh >> Uh great to have you on the show.
1:13:28
Great to hang in France just a couple weeks ago. >> That's right.
1:13:32
>> Um give uh an introduction on yourself.
1:13:35
How and when you got into podcast, and then we'll talk about where we are now. >> Yeah.
1:13:41
>> Yeah, it's been a wild ride. Uh 12 years for me.
1:13:44
I started representing podcasts 12 years ago.
1:13:45
I was a music agent at a boutique music agency in Seattle, booking tours for indie rock bands, and I used to listen to podcasts all day while I was routing tours, and I had the idea that maybe podcasts could have agents.
1:13:58
Started cold emailing podcasters, and uh was surprised to discover that there was a business there for me. >> Mhm.
1:14:06
>> Uh and very surprised in that time to discover that there wasn't really much of a business infrastructure yet. >> Yeah.
1:14:11
>> Uh so there was a lot of opportunity.
1:14:11
Um spent the next several years signing podcasters uh within that agency. >> Mhm.
1:14:18
>> Uh and then in 2018, I met with 11 different agencies and ended up joining CAA.
1:14:24
Uh started the podcast department here.
1:14:26
Uh and we have a great team.
1:14:26
Uh people focused on podcasts.
1:14:29
Uh the podcast department is within our greater um creators department.
1:14:33
Uh that works with all kinds of different creators like yourselves. >> Yes.
1:14:38
>> Uh and um it's been a really wild ride.
1:14:41
I mean, when I got into podcasting, the estimates I've seen is that the global podcast advertising industry was worth about 45 million dollars and Allen code has put out a report that last year it was worth 9. 2 billion.
1:14:51
So, you know, I I I expected there was going to be a lot of growth in this space.
1:14:56
It seemed like a great growth area.
1:14:58
Thank Uh but I never expected this level of growth.
1:15:02
I mean it's been a really tremendous wild ride.
1:15:06
>> I had started working with some podcasts around the same time, a little bit later I think. You were 2014? Is that >> Yes.
1:15:15
>> Yeah, so I I probably started working with with podcasts in like 2016.
1:15:17
But even then I would meet a show that today was probably like a 10 million dollar a year business and they would have zero revenue.
1:15:25
But they would have this like rabbit fan base and maybe they'd have one sponsor which was just someone in the audience that reached out and was like, "Hey, can I send you some some free stuff if if you talk about it?"
1:15:34
And they'd be like, "Okay."
1:15:36
And then and then, you know, fast forward to today, those kind of properties are are super valuable.
1:15:43
Uh what uh what are you seeing what are you seeing now?
1:15:46
Like what's coming down what's coming down the pipeline?
1:15:49
Net net new shows, you know, we've covered a lot of the evolution of, you know, formats, how podcasting obviously interacts with with live streaming in our case, but what do you see coming down the pipeline?
1:16:05
>> Yeah, I mean, you know, it's been clear for a few years now that video is going to be a bigger part of podcasting and we're really seeing that come to fruition in the past year.
1:16:12
It's a a major inflection point right now.
1:16:17
Um now, to be clear like it's not the videos replacing audio.
1:16:19
Both can continue to exist together.
1:16:21
There was a recent study from Edison that said that the majority of podcast consumers do both.
1:16:27
Sometimes they do audio, sometimes they do video, which that's my experience.
1:16:31
I live in LA so I have a long commute and so I I like listening to podcasts in the car, but I also like watching podcasts at home and in the office, you know.
1:16:40
Um but it it's created a really interesting moment in podcasting where there are some things that are fundamentally different about digital video versus digital audio.
1:16:47
For one thing, advertising looks really different, right?
1:16:50
The ways you can integrate with brands looks very different versus the audio space tends to be much more dominated by your 30-second pre-rolls and your 60-second mid-rolls and so on.
1:17:00
The video space tends to have a lot more custom integration with advertisers.
1:17:07
Um, and it also discovery looks really different in video, you know, it in the audio space in terms of breaking stuff through there's a lot more spend that is required in sort of the same way that you might market a TV show or a movie, right?
1:17:21
whereas in video there's a lot more clipping, there's really seamless integration into social media.
1:17:25
Uh, and so if you're trying to break through with a new podcast in 2026, you better have a really strong reason why you're not a video podcast or else you should probably have a video podcast. >> Yeah.
1:17:38
Um, somewhat related to the video podcasting thing, a big trend out of Can that Colin and Samir and others were talking about was that many podcasts are sort of reformulating as shows.
1:17:49
sort of reformulating as shows. We we we we've done this where we think of this more as a show than a podcast because it's a live show, there's a lot else going on of course like it's available as a podcast and then you also think of like Subway Takes it's Emmy nominated
1:18:04
now and it's a it's very much a show but it's also an interview that's sort of like a podcast and I'm wondering how you're perceiving the definitions changing, evolving, just this idea of what does it actually take to to deliver a show as opposed to just a podcast in the modern era. Yeah, I I'm with you. I think that's the Yeah, I I'm with you.
1:18:24
I think that's the right thinking to not try and put it too much in a box, right?
1:18:28
Because the lines are just getting so blurry between what's a podcast, what's a TV show, what's a series of reels, right?
1:18:34
What's a YouTube channel?
1:18:36
In your case, a live stream. >> Yeah.
1:18:39
>> Uh I do think that the word podcast has a bit of utility for me just because um it's sort of like I know it when I see it, right?
1:18:47
People talk about podcast, people like podcast.
1:18:49
But I mean truly, the line is really blurry.
1:18:51
I mean, you know, I I think we talked a bit in can about Oprah's podcast which just moved over to Amazon, right?
1:18:56
And I mean, Oprah is a queen of television, right?
1:18:58
And and this is where she's putting her energy.
1:19:03
Um and you know, who's to say like if you're watching it on Prime Video versus if you're watching it on your phone, is it a podcast if you're watching it one place and it's a TV show if you're watching it the other?
1:19:13
I mean, I don't know if it necessarily matters, but I think that it's an amazing time to be a media consumer because media is meeting us where we are. >> Yeah.
1:19:22
Talk about the landscape of these uh podcast distribution deals that are happening.
1:19:29
Uh Pat McAfee with ESPN, Oprah you mentioned is doing one.
1:19:34
There's uh Netflix is entering the space, Spotify went with Joe Rogan very early on.
1:19:39
Um some of these platforms just sort of get the video podcast for free.
1:19:43
Like YouTube is uh you know, the default for most creators.
1:19:45
But what are the larger companies uh looking for when they want to go deeper with a creator who might just not be ready to graduate from the self-serve options or maybe they don't have a self-serve option if it's uh linear TV or or platform like Netflix? >> Yeah, it's funny.
1:20:01
I mean, it's a really unique marketplace because as you kind of overviewed, the different buyers in my space are just drastically different in terms of their business models, you know?
1:20:11
So, it's really hard to compare apples and oranges when you're looking at one of our major buyers say is SiriusXM, right?
1:20:16
Which is a really significant satellite radio business.
1:20:19
Comparing them to Amazon, right?
1:20:21
One of the biggest companies in the world and primarily, you know, doing e-commerce, right?
1:20:27
They both have drastically think different things they need out of it.
1:20:29
Let alone now, you have Netflix and Hulu the space, right?
1:20:32
That they have different KPIs in terms of what they're looking for.
1:20:36
So, I think part of the um the challenge and the joy of representation in this space >> Mhm.
1:20:42
>> is understanding who the different buyers are and what their needs are and understanding your client, the talent, right?
1:20:48
And what they want, you know?
1:20:48
And it's it's just there's no one deal that makes sense for everyone and there's no one-size-fits-all in podcasting, which I think is great ultimately because it means we can paint with different paint brushes for different types of shows. >> Yeah.
1:21:03
Can you uh do you feel like you can identify if somebody's going to be a star based on their first ever episode? >> Mhm.
1:21:10
>> Well, I I think that I I think that taste is really important, you know?
1:21:12
And for me, like I I will take bets on stuff that I think is fundamentally good.
1:21:16
I mean, when I was a music agent, I was picking bands based on the bands that I liked and I just had to hope that other people would like those bands as well at some point.
1:21:25
And I I still believe that, you know, in podcasting.
1:21:27
I think that there there is room for taste.
1:21:29
There's certainly stuff that I don't represent that has that does good business, but it just wasn't right for me and that's fine, you know?
1:21:37
I think that as soon as you divorce uh as soon as you divorce your love of the medium from the business, then what's the point, right?
1:21:45
Like if we're if we're just here to make money, maybe we should be working in finance or in tech or something, right?
1:21:52
>> Yeah, maybe we should all be wearing suits.
1:21:54
>> [laughter] >> There you go.
1:21:57
>> You're wearing a suit. >> Look, I look at me.
1:21:57
I'm just here to have fun. >> Yeah.
1:22:01
>> [laughter] >> How are you thinking about uh live streaming and maybe more pure play live streaming?
1:22:05
I'm thinking about uh the video game creators, the Ninjas, the the Shrouds, the folks who spend 12 hours, 8 hours a day live streaming uh political commentary, business analysis, anything.
1:22:19
It's such a different business model, such a different community, sometimes much tighter audiences, but incredibly engaged.
1:22:28
Uh is that a a muscle you want to build?
1:22:30
Is that something you're thinking about uh growing?
1:22:33
What what trends are you seeing there generally?
1:22:36
>> Yeah, I think the live streaming space is is fascinating, right?
1:22:38
And obviously this has been happening for a while, but I think the moment that we're talking about of the merging of all the different things is really playing out for live streaming, too. Yeah.
1:22:46
For yourselves and for other shows, you mentioned political shows, where people need real-time news.
1:22:53
I don't think your fans want to wait an entire week necessarily to get your take on something when it develops. Uh it's a lot of work.
1:23:00
I think you guys know the amount of time and energy you have to put into doing this, right?
1:23:04
It's really impressive and I applaud you for it.
1:23:08
Um but it really creates a lot of different opportunities.
1:23:09
I think something else that you guys are doing really well is that you do the live stream, but you also cut this up into audio episodes and video episodes on YouTube and so on.
1:23:19
So, there's so many different ways to reach your audience and I think as a creator today, the more that you can be flexible to meet your audience where they are, the better chances you're going to have of succeeding and breaking through.
1:23:30
So, I love to see what you're doing and I think that you're on the tip of the spear right now in terms of the new type of experimentation that's happening in the streaming space. >> Yeah.
1:23:43
>> Um talk to me about how you're thinking about working with a celebrity or group of celebrities that wants to get into podcasting.
1:23:52
It feels like there was this crossover moment where podcasting was a backwater, then it became cool.
1:23:59
Maybe it was around COVID, but we got a whole bunch of celebrities crossing over.
1:24:03
Some of them did extremely well, won awards, SmartLess is huge.
1:24:06
Uh and but it feels similar to uh celebrities launching brands, where there's still going to be a power law.
1:24:15
It's not just it's obviously a huge advantage to have an audience already, but not every celebrity is going to have a hit podcast and vice versa.
1:24:24
Not every podcast is going to work on TV or wind up starring in movies.
1:24:30
How how are you assessing the the the reverse transition from you know, Hollywood or TV or films coming over into the podcasting world successfully?
1:24:43
>> Yeah, I think there needs to be a reason to be for any given podcast.
1:24:45
There was a level of experimentation like you mentioned around COVID where a lot of folks launched podcasts that maybe didn't pan out. >> Yeah.
1:24:55
[laughter] >> And I think maybe some of those had to do with the the the the reason behind the podcast and the idea of the podcast you know, wasn't as sought out, right?
1:25:03
I think a a good example of it working would be Julia Louis-Dreyfus's Wiser Than Me where that was driven by her desire [clears throat] to hear from older women that are largely ignored in our culture. >> Mhm. >> Right?
1:25:17
So, she had a reason that she wanted to make it.
1:25:18
It wasn't like she was just coming to the space because, you know, some enterprising agent, myself, told her that she should make a podcast and she could make money, right?
1:25:26
She was there for a different reason.
1:25:27
The things I see succeeding have that reason behind them.
1:25:29
I mean, you mentioned SmartLess, right?
1:25:30
Another COVID, you know, project that that began in COVID and it came together because the three of them wanted to hang out. >> Yeah. >> Right?
1:25:38
And they wanted to create something for people who were locked at home, you know?
1:25:41
And it And that that like genuine friendship between them is the basic building block of what it is, you know?
1:25:47
So, a lot of what I do when I talk to celebrities about podcasting is try and get to the core of what it is that they want to create and what their purpose is for coming.
1:25:56
And then we can craft all the business and everything else around that like central seed of an idea. >> Mhm.
1:26:03
What's the secret to a successful podcast tour?
1:26:05
I mean, back to SmartLess, I feel like they've been extremely successful engaging the community off of the internet, which which interesting cuz they started off the internet, they went to the internet, then they go back into the live tour.
1:26:19
Um but what what what makes for a successful podcast tour?
1:26:24
>> Yeah, I think that um your relationship with your audience is so strong in podcasting.
1:26:28
And it depends a little bit by format.
1:26:31
You know, I think the level of engagement um for more format-driven shows a little bit a little bit less than more personality-driven shows.
1:26:37
But uh something that I saw really early on getting into this space and coming from a touring background, I was looking tours for podcasts.
1:26:43
And um you know, one of the first live podcast that I went to was early client of mine, Stuff You Should Know, which I signed by emailing info@stuffshouldknow. >> No way.
1:26:53
>> [laughter] >> Uh it was a really different time.
1:26:53
Uh but then I I went and saw them do a show in Vancouver, and you they had a Q&A at the end of the show.
1:27:00
People were getting up to the microphone, and this this woman gets up to the microphone, she's like 22, looks normal and nice.
1:27:04
As soon as she gets on the mic, she is bawling cuz she is talking to Josh and Chuck from Stuff You Should Know.
1:27:10
And it was a real like lightbulb moment for me, right?
1:27:13
Cuz that's a educational podcast, right?
1:27:16
But for her, she was explaining on the microphone that they spent so much time in her ears that she has this relationship with them, and it was like she was finally meeting these friends that she sat for so long.
1:27:27
And I think that's that's when you get people who are not just going to buy a ticket, but to travel four states over to make it to a live show.
1:27:33
And you know, we we look at data for um download performance in a market before and after a live show, and we see you know, in some cases like a niche show can have a really solid uh touring business because they're converting like 50% of their listeners in a market are turning out for these live shows.
1:27:52
And we've done analysis too where we look at zip codes of uh of ticket buyers, and the the number of people who are traveling long distances to come to these is is pretty incredible.
1:28:01
So, you know, I I think that's the key thing is if you have that relationship with your audience, if you let your personality be a part of the podcast, people are going to want to come out for that. >> Mhm.
1:28:13
Uh how are you thinking about international just the like like the puzzle because uh if your business is if your podcast is aligned to specific products that are maybe only sold in the United States, it can be hard to monetize.
1:28:25
You might need to go on a tour and the international tour can be more expensive.
1:28:29
At the same time, I'm sure from musicians that you've worked with, you've solved the puzzle of what an international world tour looks like that is successful.
1:28:38
So, how do you think like the the the the the the future eras tour of the podcasting world will play out?
1:28:49
>> Yeah, I I think that um the international touring is I mean, you have Mhm.
1:28:54
[laughter] >> It see it feels like it's early.
1:28:55
Am I right in that or am I just not aware of there was already the eras tour of podcasting and like SmartLess did it and I just wasn't paying attention to their their trip to Japan or something?
1:29:05
>> I would say there's only one eras tour, but you but the you are seeing a bit of international touring happening already.
1:29:10
I mean, you know, many years ago we sent Stuff You Should Know to Australia and it was such a wildly successful tour.
1:29:13
We see a lot of American podcasts doing the UK, a little bit in Europe.
1:29:20
I mean, one of the challenges is when you're crossing over into markets that are primarily not English language speaking markets where you might have listeners, but you might have a different type of relationship with them.
1:29:30
That can be a real challenge, but especially going from one English language market to the next, it can really work.
1:29:36
But you do see audiences becoming really segmented between different markets.
1:29:42
If you look on any given day at the charts in the UK and the charts in the US, they're probably going to look pretty different.
1:29:47
There's going to be crossover for sure, but a lot of different stuff changes out even though we speak the same language and understand each other very well.
1:29:53
There's just different sensibilities from one market to the next, you know?
1:29:56
But I mean, you mentioned advertisers, right?
1:30:01
The US ad market is is a big leader in podcasting, right?
1:30:03
Because this is a consumer that a lot of brands want to reach.
1:30:08
And so, you know, as we've built our business internationally and signed podcasters from all over the world, you know, there's still at this point a real value into having a foothold in the United States in terms of reaching audience.
1:30:20
I think one of the most exciting future change areas in podcasting is going to be some seeing more of these markets come alive.
1:30:26
And if I was, you know, an investor looking for a place to to start up, I would be looking at India right now, for instance, you know. Um right?
1:30:36
Whereas there's just a little bit less saturation and more opportunity for extreme growth.
1:30:41
Um Australia, I think is a really interesting market.
1:30:43
I made it out to Sydney last year for South by Southwest Sydney.
1:30:46
And it it reminded me a lot of the podcast market in the US uh 10 years ago, 12 years ago when I got started.
1:30:53
And so, I think there's a lot more that's going to come online for those markets as the brands locally and um uh the advertisers in those areas start to realize that this is a great way to spend money and to reach audience.
1:31:05
But it was an education process in the US to get brands to to spend here and trust this market.
1:31:11
And it'll be an education process in a market-by-market basis. >> Last question.
1:31:16
Uh how are you talking to or I don't even know if you can talk about this, but um uh folks who work for large media companies, they have been uh around for the pivot to video.
1:31:29
We're going to put you on camera.
1:31:31
We're going to set you up with a podcast.
1:31:32
They build an audience and they're ready to venture out on their own.
1:31:36
We've had Ashley Vance on this show, Joanna Stern, uh Eric Newcomer.
1:31:41
There's been a whole host of these uh folks who sort of grew audiences and and learned the the skills of of content creation, whether it's Advice or Vox or any of these platforms, and then they go independent.
1:31:55
If you're having a conversation with them, what how are you talking to them about why they might want to do that or why they might not want to do that. >> Yeah.
1:32:05
So, every podcaster, every creator is an entrepreneur, right?
1:32:09
Which can be really scary if you're used to getting a paycheck every single, you know, bi-weekly from a big media company.
1:32:17
But, it's high risk, high reward, right?
1:32:19
Because once you launch your own show, you own that audience, uh and no longer are you at the whims of, you know, the executives that you work for.
1:32:28
Uh and actually capturing a smaller audience can be more lucrative for you because you're capturing more of the revenue that that audience drives, right?
1:32:37
This is a conversation I have with folks all the time who are at legacy media companies trying to decide what their next steps look like.
1:32:41
Um especially in this really fast-changing landscape where some folks are forced out necessarily when they necessarily want to make that choice right away.
1:32:48
Um so, it can be a really scary transition.
1:32:51
I'm very empathetic for people who are going through it.
1:32:54
Uh I don't uh begrudge anyone who decides that they want to keep working in legacy media.
1:33:01
I don't think that it's doom and gloom for legacy media, right?
1:33:02
I think that there's still room for great journalists on television and on radio, print journalists, and so on.
1:33:10
Um but, for folks who are entrepreneurial and want to build their own thing and own their audience, there's incredible upside and opportunity. >> Yeah.
1:33:16
Yeah, it's exciting times.
1:33:16
Well, thank you so much for taking the time to come chat with us. >> to have you on.
1:33:21
>> Thank you so much for having me.
1:33:23
>> The godfather of podcasting. >> It's true. It's true.
1:33:27
Uh we will talk to you later. Have a good one. >> Great to see you. >> Thank you so much.
1:33:30
>> Let me tell you about CrowdStrike.
1:33:30
Your business is AI, their business is securing it.
1:33:33
CrowdStrike secures AI and stops breaches.
1:33:37
Up next, we have Jeff Morgan from Olama.
1:33:40
He's the co-founder and CEO.
1:33:40
This is his first appearance with a massive Series B. Jeff, how are you doing? Welcome to the show. >> Thanks for having me. Doing great. So excited to be here.
1:33:50
Big fan of the show as well. >> Thank you.
1:33:52
Please introduce yourself and introduce the company.
1:33:54
I want to get to the bottom of how the hell you got 80% of the Fortune 500 using your product, but first introduce yourself and the company. >> Yeah, I'm Jeff.
1:34:02
I'm the CEO and co-founder of Ollama.
1:34:03
Ollama is the largest network for developers to access open models.
1:34:07
You download Ollama, get connected right away to open models like GLM-5-2 or even download them locally and run them right on your laptop for more kind of edge low latency use cases. >> Okay.
1:34:18
Uh tell us about the round.
1:34:20
Jordi's warming up the gong. How much did you raise? Who from? >> 65 million.
1:34:25
The lead was uh Tamas Tunguz, and we had existing investors participate, too, like Battery >> Awesome. Fantastic.
1:34:35
Um so uh >> [clears throat] >> tons of GitHub stars.
1:34:39
Why uh how how does this fit into like the the value add to uh businesses versus just uh downloading the model themselves from uh from an open platform, getting the the actual uh the weights themselves, deploying things, or just going with an API?
1:34:57
Like uh how are you talking to Fortune 500 customers about uh how you will actually improve their experience with something like GLM-5. 2?
1:35:06
>> You know, the power of open source is how fast it can build trust with developers around the world.
1:35:10
It ends up a large number of developers work at work at companies.
1:35:14
They work at, you know, a lot of those in the Fortune 500 or Global 10,000.
1:35:17
And open source has this special, you know, capability where you can deploy it in your own environment and not have to think about a ton of security and compliance.
1:35:25
And what that means is, look, like you can take an open model, which already has open weights, and that's kind of why open models are perfect for these use cases, and run them without any approval, really, as a developer, and and get real successful results, all without, you know, having to expose your data or, you know, even incur big costs.
1:35:45
So, um you know, that's really been the driving force of being able to to get into these large businesses.
1:35:48
Um and to your point, look, like you know, just the weights being open isn't enough.
1:35:52
You need a way to deploy them, to run them, to make sure it works on your hardware.
1:35:58
For the cloud models, you know, you really need to make sure that you're running them in a secure environment where your company can access them.
1:36:02
You know, a lot of a lot of businesses we talk to, especially Fortune 500, they need these open models hosted in the US and in Europe.
1:36:09
And that's just a requirement and it's it's a need they have.
1:36:13
And so, you know, put put that all together is it's so surprising and incredible how fast it get adopted.
1:36:19
>> So, what does uh I mean, you say you're you're you're talking to these Fortune 500 companies, but I imagine that uh the fact that you have so many GitHub stars means that there's a lot of self-serve activity.
1:36:26
Uh when does the customer cross over to an enterprise relationship with you?
1:36:31
>> Yeah, generally it starts with the individual dev, right?
1:36:32
They bring it They bring it to work.
1:36:34
A lot of them use it, you know, for personal productivity and they bring it to their team.
1:36:37
And once you're using a team, it's not a one one-person story anymore, right?
1:36:40
There There's a team there, right?
1:36:42
And everything from security to technical architects, IT teams.
1:36:47
You know, in in these folks, they they need not just a product that's really, you know, easy to use and self-serve, but they need they need a solution that really kind of end-to-end covers things like safety and monitoring and logging and data uh storage and protection.
1:37:00
Like, these are all components of a successful, you know, agent deployment.
1:37:03
And so, you know, that's when it becomes a multi-party environment and, you know, as we know, like, that's when you need a solution and and on our side, you know, we need a team to be there to help help those customers. >> Mhm.
1:37:15
>> What uh what set of models are you most excited about for the back half of this year in the open weights world?
1:37:22
>> I mean, I think with GLM-52, we just had another massive moment in open models.
1:37:26
And, you know, EleutherAI by far, at least from what we know publicly, is the highest token volume of accessing GLM-52.
1:37:32
And so, I'm excited for that cuz I think there's going to be a series of new models that are long horizon, they're focused on these really hard agentic use cases.
1:37:39
And it's going to unlock so many use cases in enterprise that, you know, the prior generation of open models couldn't.
1:37:44
You know, and the gap between open models and the frontier models is shrinking.
1:37:48
And so, you know, I I think at that point we're able to get to these incredible use cases that just weren't there, you know, three or four months ago.
1:37:58
>> Take me through some of the game theory in the open-source community around those rumors that we heard that there might be export controls on open weights models coming out of China soon.
1:38:08
If If If we stop getting uh frontier or or or near frontier uh open-source models from China for free, uh is is the next step that you would see uh an American company step up, Nvidia, or maybe Meta changes their strategy?
1:38:30
How are you thinking the uh the open-source ecosystem would evolve if China changes their strategy?
1:38:37
>> Yeah, you know, we like to work backwards from from our customers.
1:38:38
What are they trying to do?
1:38:39
And they for by and large, you know, they may have preference on specific, you know, geographies where the models are from, but by and large they're adopting both, right?
1:38:49
They're in some mix of open models and frontier models as well.
1:38:51
And to your point, like I think the US models are absolutely stepping up. They're incredible.
1:38:56
The Nemotron 3 Ultra model is just amazing and is able to accomplish some of these long-running agent tasks.
1:39:02
Um and then also, you know, one of the most downloaded models on Llama is a US model. It's the Gemma models.
1:39:07
And >> you know, this is an like a super amazing team at at DeepMind that's bringing them out.
1:39:11
The new ones are, you know, agent ready.
1:39:12
Like they can run coding agent loops.
1:39:14
They can accomplish much harder tasks.
1:39:15
And so, look, I think it's really up to the customer if they want a US uh entirely US built model designed from scratch, that's there.
1:39:24
Um if they want a Chinese model, which is often the case, they it it's less about where the model's from.
1:39:29
It's like, where does it run?
1:39:30
And is it running next to your data, which, you know, a lot you can deploy it locally.
1:39:34
Um and then are you able to deploy with safeguards?
1:39:36
And ends up a lot of customers they're not looking for like where the model's from.
1:39:40
They just want to make sure that they're running it properly and safely.
1:39:41
Um so I think, you know, they can have a an understanding of what's going to you know, what could go wrong, but what could go right.
1:39:48
And and uh there's a lot of safety tooling that can be deployed to help with that.
1:39:51
Um they have tons of appetite for that.
1:39:54
>> What do you see your role as in terms of benchmarking, reality checking, vibe checking, uh different models, helping enterprises that work with you to make the right decision, pick the right tool for the job?
1:40:10
>> You know, our job fundamentally is to connect the 9 million developers on Ollama to the right model for the right task.
1:40:16
And that's that's step one.
1:40:16
And so just by having that sheer volume and that this critical mass of devs, we're able to already understand just from, you know, our community which models are performing right for the right tasks.
1:40:27
That that's a starting point.
1:40:27
I think from there it's really collaborating with the model labs.
1:40:30
And we're launch partners with every major model lab.
1:40:32
And just making sure that, you know, the best parts of the model are shining through through Ollama.
1:40:37
Um including what are they capable of or what are their benchmarks, how can customer customer benchmark it for their own use cases.
1:40:44
Um it it all comes down to a lot of software tooling and and and and and you know, a community and a and a network.
1:40:50
And that's, you know, what we built and and is what makes Ollama special for developers. >> Got it. Uh $65 million raised.
1:40:55
Is this like what are you using the money for?
1:41:01
Because you don't have the crazy training costs because you're more of a gateway.
1:41:04
Uh is this head count, office >> 2,000 BDRs.
1:41:08
>> Is that what you're hiring? >> door.
1:41:11
Now, it feels like you have also like bottoms-up adoption with developers.
1:41:13
So you maybe need a lighter sort of like sales force to to actually capitalize on that.
1:41:19
>> Yeah, what what Yeah, what do the next uh 12 to 18 months look like for you?
1:41:23
>> Yeah, you hit the nail on the head.
1:41:23
Look, we we put out a, you know, on our site, "Hey, we're launching a teams plan."
1:41:27
We were inundated with thousands of teams that want to use Ollama. Yeah.
1:41:30
And, you know, that's going to that that's the core mission.
1:41:32
It's like, "Look, we've got this critical mass of devs.
1:41:35
How do we go solve problems for businesses?
1:41:37
Back to what we were just talking about.
1:41:38
And um that takes a team, so obviously we're expanding.
1:41:41
We got here with 14 people uh to to company of this magnitude.
1:41:44
Um but there's a much bigger team today.
1:41:47
>> people are around the market? >> Very cool.
1:41:49
>> [laughter] >> And then um and of course, you know, one thing Llama does very special for the larger open models is we host it on US and European servers.
1:41:55
A lot of the consumption of open models is going to China or is going to servers where it's there's there's no data retention guarantees.
1:42:04
And that's so important for companies.
1:42:05
And so that's a compute uh investment we're making and you know, every business in the world to access the most powerful models on compute that's secure and safe in the US or Europe.
1:42:16
>> Um will we ever settle the debate on whether the gap between uh open and frontier models is closing or widening?
1:42:26
Because depending on what sort of group somebody is a part of, they tend to have one one view or or the other.
1:42:31
Um but it but I think in in reality, it's probably always kind of going like going like this to some degree. But what's your view?
1:42:42
>> Yeah, I think you're right. It's oscillating.
1:42:42
I mean, I'm a daily jail on 52 user through Llama right now and it's replaced 80% of my coding work. >> Wow.
1:42:50
>> And I think a lot of that's going to be true of a lot of customers.
1:42:51
Um as for the gap, like to your point, I think it it may widen, it may shrink.
1:42:54
I think overall it's shrinking.
1:42:56
Um but but ultimately customers are going to use a mix.
1:43:00
And you know, for the bulk of their use cases, they're going to reach for these open models cuz they can tune them to be much faster, they're obviously much cheaper.
1:43:07
Um and there's always going to be use cases where you need the frontier.
1:43:09
Um I don't know if we'll ever settle the debate.
1:43:11
I think ultimately the gap will will continue to shift.
1:43:13
I think that's what makes it exciting, right?
1:43:16
It's like every every 3 months we're able to do something new, we're able to run better agents, and quickly open models will catch up and really enable a whole wave of customers that want to run open models to do that, you know, in their own environment or to customize it to the point where like they can even make it more more powerful.
1:43:33
The last thing I'll say too is you know customers are readily taking these open models and customizing them and they're actually getting better results often than you know just a stock uh frontier model.
1:43:41
And you know I think we're just at the beginning of that transformation. >> Very cool.
1:43:45
Well, congratulations on the progress. >> Awesome to meet you. Congrats to the team.
1:43:51
>> And have a great rest of your week. We'll talk to you soon. >> Goodbye. >> Awesome. Thanks a lot. >> Cheers.
1:43:55
>> Let me tell you about Shopify.
1:43:55
Shopify is the commerce platform that grows with your business.
1:43:58
It lets you sell in seconds online, in store, on mobile, on social, on marketplaces, and now with AI agents.
1:44:04
Uh Jordi, there is a story.
1:44:04
Oh, wait, we actually have our next guest. >> Guest of honor. >> Tibo from OpenAI.
1:44:11
He's the head of core products and platform. Tibo, how are you doing?
1:44:15
Congratulations on the launch. >> Hey, thanks. Doing well.
1:44:20
>> Uh give us the highlights.
1:44:21
>> last [laughter] night?
1:44:24
Do you sleep Do you sleep at all? >> Yeah, I did sleep.
1:44:26
Currently we have like you know maybe five to 10 war rooms going.
1:44:30
So, you know, it's just like a little bit uh it's intense. It's a sport. >> Yep.
1:44:34
>> Uh you know, but we heard you guys >> Well, and and uh you guys like to make it hard on yourselves by launching every every time there's a launch day it's like you know 15 new things.
1:44:41
So, it makes sense that there's you know a close to equal amount of war rooms. >> Yeah. Let's start with 5. 6 soul though.
1:44:49
I want to uh I want to I want you to identify for me like what is sticking out?
1:44:55
What are the most uh cutting edge capabilities that really stuck out to you?
1:45:00
The latest unlocks from the frontier of the actual model.
1:45:03
Then we can go into uh codex and voice model and everything else and how things come together and how these are used.
1:45:08
But first just from the raw model capability, uh what was most impressive to you? What was most exciting?
1:45:17
>> Yeah, it's it's it's actually really hard to answer that question personally because when we were looking at the benchmarks, trying it, you know, we were just like blown away by it all.
1:45:23
It's just better at coding, better at cyber, better at everything.
1:45:26
Like long context, producing documents, better at, you know, having taste, website generation.
1:45:33
Um and then for the first time, we also really cracked, I think, uh multi-agent setups, uh which we shipped as um the ultra mode.
1:45:39
And when you just see that going and, you know, you've got like eight agents collaborating together, communicating, and, you know, getting the same work done like faster, it's just you just feel like, "Wow, you know, this is like another way to scale test time compute." >> Yep.
1:45:52
>> Um but overall, just amazing workhorse.
1:45:55
Feels way, way better than 5.
1:45:55
5 and anything else we've produced so far.
1:45:59
>> So, uh in January, there was a project that was getting some attention called Gastown, uh talking about all these different sub-agents.
1:46:06
You had polecats and all sorts of different abstractions.
1:46:10
It felt uh highly technical, and it seems like Soul Ultra is a way to abstract that away.
1:46:15
Is that Is that deliberate?
1:46:17
Is Is there uh I guess the question more broadly is like is the level of like prompt engineering Are we Are we leaving that era or or will there always be some cycle of uh you know, uh if you get really good at using Soul Ultra, you'll have a better experience because you'll be able to give more fine-tuned, fine-grained uh prompt and and direction to the model?
1:46:42
>> So, one thing that you see as well with Soul is it's uncanny ability at, you know, understanding human intent.
1:46:46
And, you know, you need shorter prompts.
1:46:48
You don't need to explain yourself in that much detail.
1:46:51
It's sort of like, you know, it just gets it and then goes and like does like a very complex thing. >> Yeah.
1:46:55
>> Um you know, you saw the prompt for like post-training the Luna model, which is like super crisp.
1:46:59
Uh and then it does that like but it actually worked for many days.
1:47:03
Um and this is also like with us launching Chat GPT work, it's it's about making it accessible for everyone.
1:47:10
And, you know, you don't need to have like a PhD to use this model.
1:47:14
It's just like it should just behave like just like another uh super, super smart human, and you know, just kind of like get you in the moment and that's what we're striving for.
1:47:22
Of course, if you know, you really push it to the limit, you're always going to find new setups and this is like also a very exciting space, you know, we continue to develop like also Codex in the open source.
1:47:34
And then we're seeing like you know, all sorts of all sorts of novel ways to set up these agents and models so that you know, you can get results in like cybersecurity and like you know, all these other more nuanced and complex things.
1:47:47
But for everyone, you know, you should just feel like you know, it's kind of power out of the box. >> Mhm.
1:47:51
Talk about what was important at the product layer.
1:47:54
Fundamentally, what I think people want out of products is to just be able to talk to their computer like really smart coworker and be able to get things done.
1:48:03
But then you're dealing with you know, so many users over here, millions of users over here, trying to combine it and condense it into something that's simple.
1:48:12
And obviously, simple things end up being you know, exceptionally complex to actually uh create.
1:48:18
>> Yeah, so if you look at it, it's deceptively simple.
1:48:20
You can open it on your phone. It's ChatGPT work.
1:48:22
You just toggle it and then there you go.
1:48:24
You connect it to the things that you already have, your email, calendar, you know, your docs.
1:48:31
And then suddenly you're like, okay, wait.
1:48:33
I can ask it to process all of this information that I had over there that I had to manually like do all these things myself and it can just do all of that.
1:48:41
And it's just like on the go and it's like on your phone in the ChatGPT app you already have installed.
1:48:44
I mean, that's the that's the beauty of keeping it very simple.
1:48:48
At the end of the day, we want it to be just a normal conversation between you and the agent.
1:48:54
This is also why we decided to you know, ship it like just in ChatGPT. >> Mhm.
1:48:59
Uh talk about progress in computer use.
1:49:02
What is actually driving progress there?
1:49:04
Is this just something that sort of comes for free with scale and and model advances or is there deliberate data collection that's happening and some sort of flywheel that's unlocking new capabilities in computer use.
1:49:19
>> Yeah, we've done a lot of effort, bespoke effort on Windows, Mac, and like mobile computer use.
1:49:24
Also like phone use as well.
1:49:24
And so there's an entire team working on this.
1:49:29
It doesn't just come for free, but what does come for free is like every time we push the efficiency frontier and the model gets like you know more efficient like at thinking and acting and it just you know costs less tokens and it gets compressed in time.
1:49:40
It also gets better at computer use because it reduces the latency and reduces the cost. And so the two compound.
1:49:46
Like you know we have a lot of gains that we're getting from you know also like visual understanding and every time you know it improves is like the model just gets more precise.
1:49:53
So it doesn't have to correct itself.
1:49:54
And it's like maybe it misclicks a button and it's like oh it's like wait, I have to redo that.
1:49:58
So every time it's like more accurate and you know more token efficient computer use definitely benefits from it.
1:50:03
And when you compare it to 5.
1:50:06
5 is like you know just really like three times faster.
1:50:08
So you know we're not at all hitting a wall here in like how fast we can do computer use. >> Yeah.
1:50:14
Can you talk about how the role of member of technical staff is evolving because you're you're you're talking about uh Soul Ultra going off and working for days at a time and at a certain point your job sort of evolves to if you have a launch tomorrow, don't kick off a task that takes four days even if the model's capable of it and will deliver something great in four days, you need it tomorrow.
1:50:39
And so you have to size your workloads appropriately.
1:50:44
How is the how how are you thinking about uh sizing work and and actually delegating the right the right chunk of work at this stage? >> Yeah.
1:50:54
I find your question very interesting because um it actually highlights like a shift in our thinking over you know since we had 5. 6s.
1:51:02
You don't really instruct it necessarily you know for a task that's going to take four days.
1:51:07
It's like you know you tell it all the information that you have.
1:51:08
So you know you're like hey I have a launch tomorrow, Yeah, And then it's was keep track of the time and like understand that, you know, the PR needs to land like, you know, by midnight or like by 2:00 a. m.
1:51:16
It was just like reason over it. >> Yep.
1:51:18
>> And you're not the one that needs to manage like, you know, all of that extraneous like complexity. >> Yeah. >> Um and so that's it.
1:51:23
What you're seeing as well is like, you know, your relationship with the agent like changes over time as it gets more intelligent.
1:51:28
And you're just like, "Oh yeah, I can just talk to you like, you know, another like, you know, superstar super smart human." >> Yeah. Yeah.
1:51:33
Uh talk about the efficiency of the model, what work went into that, why why it matters, you know, what kind of conversations you're having with, you know, big customers, all that stuff.
1:51:45
>> Yeah, so what matters a lot right now is like sitting at the frontier and, you know, getting the max capability when you want it, but also for your normal average day-to-day task is like, you know, being super efficient.
1:51:55
And not just for latency, just because also we're seeing you know, so like we had this era of token maxing and then, you know, we've been talking a lot with, you know, old older companies and enterprises that we're working super closely with and then they were like, "Oh, it's just a little bit, you know, maybe out of control."
1:52:08
It's like, you know, what we want is like, you know, we want a highly efficient model that is, you know, steerable, controllable.
1:52:13
We want to have the right, you know, spend control dashboards.
1:52:15
And so we also have all of that.
1:52:17
Um you can you can look at your spend and understand the ROI, but also you can uh rest knowing that, you know, this is actually like a super super efficient model.
1:52:25
And so it gets the job job done uh with, you know, way way um with fewer tokens, which, you know, to you uh you know, means that, you know, you have to pay less uh for the same results, which is super important.
1:52:36
And this is really the theme I think of the year is like, you know, that um being on the efficiency of like, you know, performance um and and and cost.
1:52:46
>> What about if you want to spend more for faster uh performance?
1:52:48
What does the future of uh either ASIC-enabled, Cerebras-enabled, uh Spark and and fast mode?
1:52:57
What do you want to see develop there uh either immediately or over the next couple weeks?
1:53:03
>> Yeah, I think what we are um just really working towards like it's a buffet of options, right?
1:53:06
So, um for for your normal like, you know, interactive task is that, you know, you're going to use 5X all like on medium or on high and you're going to have like an amazing time.
1:53:16
If you have a really hard problem and you're trying to, for example, find a cyber vulnerability in something and so you're going to run ultra and you're going to run it like for 2 days and it's just going to like the leave no stone unturned and like, you know, invent novel techniques and, you know, you're going to be like absolutely blown away by what it comes up with.
1:53:31
Um and but a lot of times you also just need speed, you know, like for some of the stuff that we're dealing with is like, you know, we love working off of like the Cerebras version of this, which is like at, you know, about like, you know, 750 tokens uh per second, which is an order of magnitude faster um than the the default version that we have on the uh on the API and in the product today.
1:53:52
And this is just really situational or um if you just want the very best uh and um you're like absolutely no compromise. It does come at a cost. >> Of course.
1:54:03
A lot of people were feeling left out this week uh that weren't in the early access program.
1:54:06
What makes a good early access partner?
1:54:08
I'm sure your DMs are just just filled making, you know, people that want access to the next set of models, but uh what makes a good partner to the to the to the product and the research team?
1:54:20
>> Yeah, we really try to go as broad as possible.
1:54:22
Um it is quite a bit of effort to manage and then also like we're getting all that feedback and incorporating it and we work very closely with uh the the folks in early access.
1:54:32
For us it's just really about um realizing whether it is as good as we think it is, right?
1:54:39
You know, you're like so close to the model, you train it, you know, you've incorporated like all that feedback, all your dreams, visions into this model and then you've played with it for a little bit.
1:54:47
And then when we give access to, you know, folks outside of Open AI, it's like, you know, the first time where we have like a an unbiased look, you know, where people use, you know, all sorts of models and um different different harnesses every day.
1:54:58
And so it's just kind of like this awesome um you know, is it actually as good as we think it is is like you know what are the things that we missed and so it's that you know high high bandwidth engagement good feedback and then you know that sort of um people who have shown to be unbiased in the past and you know talked honestly about you know all sorts of models and all sorts of hardnesses.
1:55:20
>> How do you think how are you thinking about the trade off between mobile cloud desktop the Mac mini that went mega viral last year do you think that we'll stay in a hybrid pattern for the foreseeable future is it a person by person basis do you have a unit grand unifying theory of how agentic work happens in the future?
1:55:45
>> Yeah the way that we think about it is no compromise so you want to be able to use the same your AI partner you know like on your phone on like on the go it's just like you know I go walk in the park I want the exact same thing I want it on my laptop I want it you know at home maybe like running in a Mac mini and it is more that it needs to be able to
1:56:05
have access to all the things that are important in my life and you know not be constrained by the physical you know boundaries of like you know just like hey I started this prompt or I started this conversation on my phone and it's just like now it's stuck on my phone you know we want it to be uncompromising and so your AI your ideal AI partner I think
1:56:21
just you know has access to everything all the time and just you know processes information as needed and then you know can act in a safe way and controlled way you know so that you know you always understand like what it's trying to do and you know if there is like something risky you can you know approve it or you know ask it to change talk and there I
1:56:38
also think you know like the mobile has like a big role to play right so if you're if you know 5 6 hours you know busy like you know working on something and then you know you just go out to dinner like you know it should ask you for permission to do something when you're there you know you don't don't need to be stuck on your laptop. >> Yeah. >> Yeah.
1:56:54
>> Fast forwarding 6 or 12 months how is how how how important is voice to someone's day-to-day experience with ChatGPT work / Codex?
1:57:06
>> I don't think we need to fast track it and we should ship the voice yesterday. >> No, I know. I know.
1:57:10
But but you know, you assume like you know, often times like something ships, you know, and it takes a little >> Everyone has to go on holiday break.
1:57:19
>> Yeah, we need a 3-day weekend. A 3-day weekend.
1:57:21
>> and then everyone can test out the latest and and and integrate it into their workflows.
1:57:26
>> Yeah, you know, open attentive the app.
1:57:28
The username latest voice.
1:57:28
We also demoed it in the live stream this morning and it's very it was super enjoyable like magical experience when you first experience it but also like you know, on the fifth time as well.
1:57:38
It's going to be part of like you know, day-to-day experience you know, of like how you work with these systems.
1:57:45
We don't we don't have it yet in the desktop app but this is something that we're working towards and when when you experience it is it is like a modern-day like Jarvis, right?
1:57:55
It's like you know, you just talk to it.
1:57:57
You just walk in your room and you know, suddenly it's just like doing things on your computer, you know, with the same level of like precision and power, you know, that you currently have over text. >> Yeah, it's amazing. Uh >> Fantastic. >> Congratulations. Thank you so much.
1:58:11
Hopefully you can get some sleep.
1:58:11
I'm sure you're crazy busy.
1:58:17
Have a great rest of your day. Congratulations.
1:58:18
We'll talk to you soon, Tiva. >> Have a great one. >> Goodbye.
1:58:23
Let me tell you about Railway.
1:58:23
Railway is the all-in-one intelligent cloud provider.
1:58:26
Use your favorite agent to deploy web apps, servers, databases and more while Railway automatically takes care of scaling, monitoring and security.
1:58:33
And while we bring in Sean Frank, I'm also going to tell you about Cisco.
1:58:37
Critical infrastructure for the AI era.
1:58:38
Unlock seamless real-time experiences and new value with Cisco. Sean, how are you doing? Welcome to the show.
1:58:45
First time first time in the actual studio, right? >> Second.
1:58:48
I did the Shopify episode. >> Oh, but right. That's right.
1:58:51
>> He took the time out of his Black Friday. >> That was great.
1:58:53
Yeah, but we we yeah, we bounced around a lot.
1:58:55
It was a it was a big it was a big day.
1:58:56
That was a good one, though.
1:58:58
Hopefully we'll do it again. >> Yeah.
1:59:00
>> And and Black Friday was a record for you, right? It's successful. >> Yeah, yeah.
1:59:05
>> Have you have you had a Black Friday that wasn't a record?
1:59:08
>> No, every year it's gone up. Every year.
1:59:09
>> And it's not going to stop this year.
1:59:11
>> I've only over here at the Ridge wallet a Ridge broadly, right?
1:59:13
The portfolio is growing still?
1:59:16
>> Yeah, the wallet business is a great business.
1:59:18
It's like a 100 million plus a year, but the growth is now like all the other stuff we're doing.
1:59:23
So we have um like a travel line.
1:59:24
We have >> We do like 10% of all men's wedding rings in America.
1:59:28
So we have like a huge Yeah, yeah.
1:59:31
It's a >> Double digits of the TAM.
1:59:33
>> Yeah, it's a very boring monopoly.
1:59:33
We sell a ton of like men's wedding rings.
1:59:36
>> And they're coming for it all. >> Oh, yeah. It's true.
1:59:39
Um [snorts] >> You won't be able to get married.
1:59:41
They're actually going for regulatory capture.
1:59:42
You won't be able to get married if you're not you planning to use a Ridge ring. >> Yeah.
1:59:47
I'm I'm like a big pronatalist now cuz I'm like get married and have kids and >> Okay.
1:59:51
So you're behind all of that.
1:59:53
Every time I see some podcaster yapping about the the fertility crisis, it's you.
1:59:58
You're behind it funding it all.
2:00:00
>> And the big thing now is like the tech business for us.
2:00:01
So like we do like uh phone cases and that type of stuff.
2:00:05
>> It's like 12 months old and it's like already like a 100 million dollars a year. So >> Wow.
2:00:08
>> Just adding a bunch of new random little little widgets to sell. >> That's insane.
2:00:11
Uh okay, so take me through the uh demand generation side of the business.
2:00:16
Like what is actually changing?
2:00:17
Uh obviously there's AI generated advertising, AI enhanced targeting.
2:00:22
There's all sorts of different stuff, but like from your day-to-day, from the platforms you're advertising on, like what has been the most material change of the last 12 months?
2:00:33
>> Well, it's a big day to be here, right?
2:00:36
Because the big meta announcements. Yeah.
2:00:37
People were like dancing on Meta's grave like 3 days ago and >> [laughter] >> and now everyone's stoked on it.
2:00:43
>> you have a 1% God candle. >> Totally.
2:00:47
Um >> Is we we is that specifically about the LLM or or the image generation model?
2:00:51
Because the image generation model seems much more impactful to the advertising business than the agentic coding capabilities. >> Well, yeah. >> Yeah.
2:01:00
>> Well, I would say the mannis actually like really helped unlock like a lot of um like stuff inside the ad account.
2:01:07
>> Using the ad manager more correctly. >> Yeah, yeah.
2:01:08
I think like it just democratizes a lot of tools that like the best ad managers are already using. >> Yep.
2:01:13
>> Um but really we just want Meta to continue to get better. >> Yeah. >> Right?
2:01:18
Over the past 3 months they've rolled out a lot of changes of like the actual ad algorithm. >> Yeah.
2:01:22
>> And it was really bad for a lot of people.
2:01:24
We just had like the best Q2 of all >> good for you. >> was awesome. Yeah. >> Interesting.
2:01:27
>> I think they're getting way >> so >> Yeah.
2:01:29
>> think that it was actually bad for some people or they were just going through a slump in their business overall?
2:01:33
They had like business problems that they wanted to blame on the ad changes. >> Well, dude, yeah.
2:01:41
I mean, what's new, right?
2:01:42
It's like >> [laughter] >> It's a If your business is going bad, it's everyone else's fault.
2:01:46
Um but really like, you know, they do like the Meta Performance Summit every May and they've just rolled out so many great changes to the ad algorithm that like I think you're getting better impressions with better people and I the more compute and AI they throw out I think it's just going to get better and better.
2:02:02
And like we're getting to it like a future where like the perfect impression at the perfect time with the perfect ad that's like customized to that person with AI, that's coming.
2:02:08
And click-through rates will go up, conversion rates will go up, and CPMs will go up with it, but I think uh we'll be like a moment of arbitrage there.
2:02:17
>> Do you do you think do you think Zuck is doing enough to actually message that to customers like you?
2:02:23
Because Ben Thompson put out this piece on Monday sort of an earnings call transcript that he wrote in the voice of Mark Zuckerberg.
2:02:30
And his pitch was to investors telling the investors, "Hey, look, we are investing a lot in AI, but it's all in service of the ads business, which is great, which we do take seriously.
2:02:41
We have taken some side quest on some metaverse, some VR stuff, but this investment that we're making right now, you shouldn't beat us up in the public markets because it's going to come back huge.
2:02:51
We're growing really fast on the ad side.
2:02:53
33% is massive at that scale and it's going to continue to double down, but I'm wondering if if advertisers are are receiving that signal from Meta that it is going to get better.
2:03:06
It is some place that they should be spending more time and more dollars.
2:03:10
>> Yeah, we're captive audience, so I don't think he has to message to us directly.
2:03:12
>> He he can just ignore us.
2:03:12
[laughter] >> Yeah, like the best thing he could do is get people to spend more time on their app, right?
2:03:16
And then better understand who those people are and what they're in market for.
2:03:19
And if he delivers those things, the ad dollars will come.
2:03:22
Cuz besides that, like where else are we going to spend money?
2:03:24
Like TikTok shops actually doing great, but it's still a really small business, right?
2:03:27
Like the GMV this year in America might be 20 billion and Amazon does that like every 4 days or something.
2:03:33
So it's like it's still a very very small business.
2:03:37
>> TikTok shops so small?
2:03:37
Is it Is it a separate panel or something?
2:03:38
I feel like TikTok still has so many so many impressions, so many videos that are being served.
2:03:44
Do you Do you have an idea of why that business is so small?
2:03:48
>> I think it's it's not necessarily I think it's primarily that content platforms have been a place you discover products, but not where you transact.
2:03:56
Like Meta has done a lot of had a lot of efforts in shopping in app.
2:04:00
They've never You would have thought that they would have clicked harder, right?
2:04:04
It's like you have this captive audience that uses your product for an hour a day to discover things to buy and they're not buying that many products in the app. >> Yeah.
2:04:16
>> Yeah, and TikTok shop is delivering way more value than 20 billion in GMV.
2:04:21
Like a brand is like Comfort hoodies, I'm sure you guys have seen them.
2:04:24
They're doing a billion a year right now. >> They're 4 years old.
2:04:26
>> Last time Last time you told me about them, I think it was like 400 or something like that.
2:04:30
>> They've They've [laughter] more than doubled. Yeah, it's crazy.
2:04:31
And all it is is like TikTok shop affiliates, so people posting thousands of videos a day.
2:04:37
They take those the best videos of it.
2:04:39
They put it into TikTok Shop GMV Max.
2:04:41
They run that and their TikTok Shop business might do might do 100 million a year, but the spillover is crazy. Everyone goes to Amazon.
2:04:47
Everyone goes to your website. >> Oh, got it.
2:04:50
>> So like they've just done a horrible job actually capturing the value that they're generating.
2:04:54
>> Interesting, but they're getting a lot of impressions, they're driving a lot of purchases.
2:04:56
What's the sweet spot for uh price point on TikTok Shop?
2:04:59
That hoodie company, is that a $70 hoodie? >> It's cheaper, right?
2:05:03
So like it's to a successful product on TikTok Shop is female focused, uh impulse buy, so like you know, $20, $30, $40, something like that.
2:05:12
And then you have people to make outrageous claims, right?
2:05:16
>> [laughter] >> And so uh Comfort Hoodie is like it's a hoodie that cures anxiety.
2:05:18
So like that's pretty good.
2:05:20
Like it's a pretty good value product.
2:05:22
>> to that one at some point.
2:05:24
Yeah, that's a medical claim the FDA >> All the affiliates are making it, so who cares?
2:05:26
>> Yeah, somebody should care, but we'll see.
2:05:30
Uh yeah, a little bit of a gray area. Interesting.
2:05:32
Um are there any Uh I I I I want to know about when you're using AI personally in-house or someone on your team is using AI, like the models or the products, like ChatGPT, Claude, etc.
2:05:46
, versus you feel like you're getting AI for free because you're using Mailchimp and Mailchimp integrated AI, or you're on Shopify and Shopify gave you an AI feature for free?
2:06:00
>> Yeah, you know, I I don't want to talk too bad about any sponsors.
2:06:02
I don't know if Notion sponsors you guys, but okay, great.
2:06:05
Um you would actually hope that more of these companies would be rolling out AI faster and better and more useful, cuz like, you know, we're using it internally a ton, directly with the models.
2:06:15
And like inventory planning and buying is a solved problem.
2:06:18
That was like the biggest problem that has plagued like >> Like demand forecasting.
2:06:24
>> Yeah, like all of that is like >> sending up is sending a proposal to a supplier or or in your case like the actual factory to uh to understand how many you're going to sell by month, when they need to be delivered, what what the shipping timelines.
2:06:35
All of that's just a bunch of Excel work normally.
2:06:38
>> Yeah, and it was huge teams.
2:06:38
And if you got it wrong, it bankrupted your company, right?
2:06:40
Like you had bad inventory or you get like shorts in December, right? Like >> Yeah, yeah.
2:06:45
>> It ruined a bunch of businesses.
2:06:45
AI has totally solved that.
2:06:47
Working directly with codexes of the world, right?
2:06:49
And just putting in all your your business information.
2:06:53
>> a huge unlock for the global economy.
2:06:56
It's like crazy to think >> We got to tell we got to tell consumers who don't [laughter and cough] like AI.
2:07:01
Your favorite brands are doing perfect forecasting.
2:07:04
And that means that you are going to get the perfect article of clothing right before your trip to Hawaii, even though you waited until 3 days before to order it.
2:07:14
>> You'll always have the right sizes in stock.
2:07:16
That is coming because of codex. >> That's crazy.
2:07:19
>> This is this is this is like part of what the sufinator was talking about, right?
2:07:23
Like the one of the >> [laughter] >> Eric sufert mumbled that memo.
2:07:29
He was just talking about like actually if you have like better advertising because of AI, it will create a like like it will drive economic growth purely because you have more and more of these like niche businesses that might have an audience of 50,000 people in the whole world.
2:07:44
Historically you couldn't build that business cuz it would have been impossible to find that 50,000 people out of billions. Now you can.
2:07:49
And then this is again an accelerator of like so much revenue is lost every day.
2:07:57
So much purchasing activity doesn't happen because people just can't buy the stuff that they want because a brand didn't properly forecast and and is still kind of doing like fly by wire. >> Totally.
2:08:09
The wrong sizes, the wrong things at the wrong time.
2:08:11
Like with how expensive it is to get stuff on shipping containers and how long it takes.
2:08:17
And there was a lot of companies that spent a lot of money trying to solve this.
2:08:19
There was demand software that did billions a year in revenue.
2:08:23
As long as you have like a a of records, so like a clean data warehouse, and you have your all of your sales from Shopify, Amazon, whatever else, you port all of that into a Codex, and it will it totally has it figured out.
2:08:33
And if you have to make those small tweaks, like, oh, actually, last year we were on a promo, this year we're not going to.
2:08:38
>> Oh, so the multi-platform thing is big here, because I would have I was just about to ask you, like, should Shopify roll out a demand planning tool for Shopify merchants, but they probably Amazon has sharp elbows and won't give them all the data just via a simple integration.
2:08:53
>> Yeah, it all has to roll up into the harness.
2:08:55
It all has to go into Codex.
2:08:57
>> And so you so you have to get into a data warehouse, and then you have to point an LLM at that.
2:09:00
>> Yeah, so like, you know, we're not like, um, you know, we're not like a large buyer of software.
2:09:05
Like, you know, we we have Shopify, we have a data warehouse, whatever else.
2:09:08
But as long as you have those like basic things, and you put it into a harness, it's like >> But your IT spend's like, yeah, probably like less than 1% of revenue or so. >> Totally, yeah, yeah.
2:09:17
>> Yeah, exactly, which is which is the good benchmark.
2:09:18
You don't want to be like building every system custom from scratch.
2:09:23
>> Yeah, and people are so excited cuz they can vibe vibe code everything, right?
2:09:26
But like, you know, Judge.
2:09:26
me Reviews is like $5 a month, so I'm not going to vibe code my own reviews thing, right?
2:09:31
I'm just going to do this. >> Yeah, just pay that. Yeah.
2:09:33
>> But, um, but yeah, so we >> There are some review plugins that are very expensive, so.
2:09:37
>> Yeah, Bazaarvoice, horrible, like, yeah, Yotpo has a better reputation. >> Yeah, yeah.
2:09:42
I've gotten I've gotten, uh, fleeced a couple times. Um.
2:09:45
>> What's new in manufacturing land? >> Huh.
2:09:48
>> You know, there's >> Cuz you had you guys were trying to develop US manufacturing like way before like American Dynamism was like a category.
2:09:58
Like, this is something that you guys have been like exploring and dabbling with for a long time, but what's what's the latest?
2:10:05
>> Yeah, so like, we actually worked with the government in like 2022 to get something called like a general exclusion order.
2:10:10
So like, we have a we can we can very easily now get people to get, um, banned from importing.
2:10:15
Like, if we if they violate our IP.
2:10:17
But to do that, you have to prove that you're like a important part of the American economy.
2:10:22
So to do that we actually bought the the largest independent watchmaker in America.
2:10:26
It's called like a It's called FTS 5 Time Piece Solutions. They're in Arizona. So we own them.
2:10:31
So if you ever buy American-made watch, I probably made it. >> No way.
2:10:36
>> but it's a really hard business. >> That's crazy. >> Watch Watch it suck.
2:10:38
Um so we do wallet production there. >> Okay.
2:10:42
>> Um you know, we probably spent two or three four million dollars like getting that whole thing set up to actually produce wallets there.
2:10:47
But then like the Trump tariffs made it really hard to get steel because there's a huge global tariff on all non-American steel.
2:10:53
But that means everybody wants American steel, so now it's really hard to get it and we just wait for those things to work themselves out, but um Look, most manufacturing is already very automated.
2:11:05
Like you guys have spent time in China.
2:11:06
I'm going back there next month.
2:11:06
It's like they don't have that many people in factories.
2:11:09
It is robots and assembly and um that can be done basically anywhere. >> Yeah.
2:11:15
>> And then it's just getting the raw goods to wherever you actually want to manufacture stuff.
2:11:18
And with steel and batteries, you we can't do that in America yet.
2:11:21
So we we have to build a supply chain.
2:11:24
It'll be like a 10-year thing, but the future is going to be hyper-local manufacturing for sure.
2:11:29
>> the idea of you getting into steel manufacturing. That'd be electric.
2:11:33
Just make your own steel.
2:11:33
Fully vertically integrate.
2:11:36
>> Is now the best time in history to start a consumer brand?
2:11:42
>> Well, I would say probably like January 2012 when Facebook ads just rolled out.
2:11:47
That was probably the best time.
2:11:48
>> But now is the second best? >> For sure.
2:11:49
I mean one, it's moded from AI.
2:11:52
Like I would hate to be trying to sell software right now, right?
2:11:55
And a lot of services, I'd hate to be in that business.
2:11:57
People are going to buy stuff forever, right?
2:11:59
You have birthdays, you you have Christmas, the the the American consumer is still incredibly strong.
2:12:05
We really just had the best Q2 of all time and that's with a war in Iran.
2:12:09
Like in >> And inflation and all sorts of stuff going >> Yeah, I and like it all actually like looking at >> Consumer confidence is so low even though consumer spending is is holding.
2:12:20
You There's always nervousness about will there be a pullback?
2:12:24
>> Yeah, there's been nervousness for like 6 years, but I'm telling you the Ukraine war in 2022, there was a noticeable decrease in e-commerce activity.
2:12:29
And right now it's not happening.
2:12:31
Things are actually ripping.
2:12:32
So, I think it's a great time. >> Total vibe session. >> Yeah.
2:12:36
>> Total disconnect between what people say in the Pew research and then what people actually do.
2:12:40
>> If I had my Shopify notifications on, it would just be chiming all day right now.
2:12:44
I think people are Yeah, it's it's definitely a vibe session.
2:12:47
Um So, it's a great time to be selling stuff. >> That's great.
2:12:50
What are you thinking on the creative side?
2:12:52
Are you Higgs field maxing with like all these workflows to generate endless AI videos?
2:12:57
Are you seeing progress in AI images? What what's working?
2:13:03
We just talked to the Soufanator about the this study that showed that when AI is clockable, it underperforms, but when it's not identifiable as AI, if it's just a product image and it just looks indistinguishable from CGI or photo, it over performs sometimes.
2:13:22
>> Oh, dude, pull up my Facebook ads library and it is tons and tons of AI generated static ads. >> Okay. >> Right? >> Static?
2:13:28
>> Yeah, cuz the the statics are actually like you can build like ad factories like totally automated.
2:13:32
So, you know, you take a Higgs field, you use an MCP, you bring it into like your harness of choice like a Codex, and you can generate 10,000 static ads if you wanted to, right?
2:13:41
And we just have that running 24/7. >> Yeah.
2:13:44
>> They get pumped into a Facebook ads library to test it, and then the winners go to a different ads library to like a different ad account to actually scale those up. >> Yep.
2:13:51
>> Um so, the static stuff is totally solved and >> I would hate to be trying to do ads without it right now, right?
2:13:57
I actually built a spreadsheet yesterday or like a presentation slide by hand and I felt like I was a caveman.
2:14:02
It's like that's like what working with like the ads of like the past were were like.
2:14:06
Video, we still do a lot of it.
2:14:08
It's mostly just uh it's cut scenes in like a hyper So like we want to add motion or somebody talking or whatever, we'll do a lot of that. >> Yep.
2:14:16
>> Um but it went viral yesterday the like the Seinfeld fully AI episodes. >> Yes.
2:14:21
>> That is really really good.
2:14:21
It's getting very close to actually being indistinguishable.
2:14:26
>> it looked really good.
2:14:26
I thought the editing pacing was wildly off.
2:14:29
>> Yeah, I agree with >> I thought there was like the gaps in the humor, like the pacing is so key to that.
2:14:33
It was like but from a fidelity perspective, it looked indistinguishable from the show.
2:14:39
>> Yeah, so it's like it's very inhuman and like the the way they talk or whatever.
2:14:42
But like that's going to be fixed.
2:14:42
Like that's coming in two more model updates, I'm sure.
2:14:46
And then and then it's like yeah, all video will be there.
2:14:48
>> What about using AI either to write deterministic scripts that can assemble hypercuts in different because if you have a picture of the wallet, a picture of a person putting in their pocket, a person stepping out of a car, a person on the beach, you might want to sequence those 1 2 3 4, 2 1 3 4, 2 3 2 1 4 and get every possible variation on the the those different video clips. Are you using AI?
2:15:14
Do you already have a system for that?
2:15:16
Is that already automated?
2:15:16
Is there what's the future of that?
2:15:19
>> Yeah, that's still very human in the loop, right?
2:15:20
So we have like, you know, two amazing editors who would do all the hypercutting themselves.
2:15:24
And now they are using Hexfield and just getting, you know, hundreds of more variations, going through those and then uploading them to the app. >> Got it. Got it.
2:15:31
And they can probably even do some like style transfer and filtering on top of the raw footage that they have to like >> Yeah, and sometimes like, you know, it's a robot.
2:15:38
It'll it'll make stupid decisions.
2:15:39
It's like it'll it'll go like, you know, wallet to something, you know, falling in the ocean.
2:15:43
It's like, what the hell are we doing? >> Yeah, yeah.
2:15:44
It's just complete hallucination. >> 5.
2:15:46
6 people are reporting that it's it's working on video editing workflows now in a way that a lot of other models haven't, so >> Yeah.
2:15:54
I wonder how that will actually play out because there's one world where you're just literally opening Premiere Pro and saying like move the mouse cursor and make the cut in the footage.
2:16:05
And And there's another one where you're like editing the underlying file, and then you're watching it in Premiere Pro because most of these video apps, they sort of represent the file as like a structure of folders or, you know, a bunch of JSON or something.
2:16:17
So, you you you can manipulate things multiple ways, but uh we'll be interesting to see where that where that goes even if uh it's just for like those the the the re- reconfiguring sequences and whatnot.
2:16:27
Are there any uh vibe-coded e-commerce plugins or add-ons or tools or software that have stuck out to you as wow, like this thing it was in the $1,000 a month category, now it's in the $5 a month category, or or this is a new capability that's unlocked that's still something you wouldn't roll your own, but you would buy outside?
2:16:56
>> Uh you know, not really, but like a lot of the the playbooks that people share, it's like, you know, how to build landing pages in one prompt or whatever.
2:17:02
And like that is going That's like going to one-shot companies like Shogun or like there's all these like landing page builders.
2:17:06
And it's like they're just, you know, drag-and-drop tools, but you're getting way faster, way more responsive, way better stuff just out of Codex.
2:17:13
And it's like it's completely on brand with your assets.
2:17:15
And it's like, "Look, that is vibe-coded, right?
2:17:17
It's and we are we're going to launch 50 landing pages next week, and it's all vibe-coded stuff like that."
2:17:23
>> the value of landing pages these days?
2:17:25
Is that Is that critical to the funnel?
2:17:27
Is that critical to the Yeah, it's not.
2:17:30
>> guys had Hermozian here like 2 days ago.
2:17:32
I listened [laughter] to him.
2:17:32
It is uh >> He's got 1 billion.
2:17:33
Now, he's going to get a a landing page for every human on Earth.
2:17:37
That's actually where we're going.
2:17:39
>> Yeah, basically, right?
2:17:40
Uh >> The census records?
2:17:42
>> But it's like you you need the offer to get the click, right?
2:17:43
And then you need the landing page to inform them, and then the like as fast as you can get them to make the purchase as possible, right?
2:17:50
>> Uh in sales, they tell you to use the person's name a lot.
2:17:54
I'm really happy that you're here, Shawn, cuz I want to talk to you about this.
2:17:59
Uh do you think we get to the point where like digital platforms end up, you know, in ads and landing pages are like using the individual's name?
2:18:08
>> Uh >> To an extreme degree.
2:18:08
Like cuz you get down to targeting one person where I would be like >> long time ago.
2:18:13
>> Yeah, no, but would would eventually there's like a there's a there's a flipping point where maybe it just is so effective that it makes sense for Facebook to enable.
2:18:24
>> Well, all AI cold email right now already puts your name in there.
2:18:26
And there's been beta tests rolled out where they're using people's faces in the ads.
2:18:32
It's like like >> Yeah, we saw that on remember? >> Yeah.
2:18:36
>> Like they I think they were trying to sell a pair of the meta glasses where it's like showing like you calling your you know, and they can tell like who your significant other is just based on your activity >> Yeah. >> on Instagram.
2:18:48
>> Yes, so Facebook has all that information.
2:18:49
It has your face, it has who you're talking to, it has who your wife is.
2:18:52
So it's like they they Like hyper-personalization is coming for the entire web and you know, if you're going to be shopping, it's going to show you what you're going to look like in the clothes, you know, what your dad's going to like when he gets the wallet.
2:19:03
I definitely think that's happening.
2:19:05
>> Yeah, I mean if the price of cold call goes to a penny, do you think you'll be cold calling people?
2:19:09
I something just came across my desk.
2:19:10
I got a wedding ring here for you. >> We talked about it.
2:19:15
>> We talked [laughter] about it.
2:19:16
>> Yeah, there's a thing called like ringless voicemails.
2:19:17
Well, like you know, they'll just mass drop those off and they put people's names in there.
2:19:23
It's like hey John, we got this thing for you, right?
2:19:24
And it's like you know, it's not it's one way, but yeah, the two-way is definitely coming. >> Interesting.
2:19:30
>> What's happening in luxury?
2:19:30
LVMH and Kering are down like 25 and 20-ish percent.
2:19:37
>> Well, we talked about that last time I was here.
2:19:38
I was like I'm like oh yeah, Gucci's getting crushed and and it's like yeah, it's going to continue to happen.
2:19:44
>> [laughter] >> Why is that?
2:19:46
>> You know, it's I think it's just a generational change.
2:19:50
It's it's really what it comes down to.
2:19:51
Like each one of those brands have good assets, but like Richemont is is still tearing, right?
2:19:56
Hermès is doing great, Coach is on like a generational run.
2:20:00
Like the best performing stock of the past like 2 years, right?
2:20:01
I think last year it beat Nvidia in performance. >> No way. >> Um >> Coach? >> Yeah.
2:20:06
So, it's uh public traded under Tapestry. You should pull them up. >> Okay.
2:20:09
>> Um they own Kate Spade, too, but Kate Spade's like a nothing business. >> Yeah.
2:20:12
>> Um so, it's just it's just a generational rotation, right?
2:20:13
Ralph Lauren also on a tear.
2:20:15
So, um you talk about like American dynamism, there's American luxury.
2:20:19
Like America wasn't old enough to have luxury brands, but like LVMH bought Tiffany's and now it's like look, Ralph Lauren's crushing, Coach is totally crushing, and I think the old stodgy LVMHs of the world, the Guccis of the world, one, they got overexposed.
2:20:32
Like they really rely on the middle class.
2:20:35
And if you ever look at like their sales demographics, it is like 80% of the revenue come from people making under $150,000 a year.
2:20:40
And it's like that's just a big disconnect when you're trying to sell whatever, right?
2:20:44
Um so, there's going to be a shift towards like, you know, um true luxury.
2:20:49
There's also a some concern that like L Catterton is just personally investing in all the great assets and not bringing them into the portfolio.
2:20:56
Like they own Chrome Hearts, and they they haven't brought that into LVMH. >> Oh, interesting. >> Yeah. >> Okay.
2:21:02
Well, but when you say personally investing, it's L Catterton is investing. Got it. Okay.
2:21:07
>> And L Catterton is the investment vehicle >> of LVMH. Yeah, that makes sense.
2:21:09
And I know that they do a lot of deals in the private equity world, but >> And then >> Oh, yeah.
2:21:13
Why why doesn't LVMH like >> They're actually >> imagine the Arnaults are like, yeah, we want to own the hottest Like you would think they would do everything they possibly could to own Chrome Hearts.
2:21:26
>> There's a divestment of brands right now.
2:21:27
They're actually trying to push stuff out of their portfolio.
2:21:28
I think they just liquidated Off-White, right?
2:21:32
And um Off-White ended up being like a Target like Costco brand.
2:21:34
Like there's Yeah, there's like photos of like Costco Off-White uh like big pallet and delivery.
2:21:40
So, they're actually doing a divestment right now from brands.
2:21:43
Um and, you know, Tapestry just divested from Stuart Weitzman.
2:21:46
So, like they're actually trying to go like bigger down on winners.
2:21:48
But yeah, then there's just like a whole other like, you know, Madhappy's a great brand.
2:21:51
It's invested via L Catterton, not inside of the portfolio.
2:21:56
And the whole idea was it was supposed to be a scout fund to then bring it into the main portfolio, but if you're the L family and you own 90% of L Catterton and only 45% of LVMH, it's like, what's what's the incentive?
2:22:10
But look, I mean, you could like Richemont owns Cartier.
2:22:11
Cartier's having a great time.
2:22:14
You know, they own Van Cleef.
2:22:14
Van Cleef's still having a great time.
2:22:16
But those are both of those are way more true luxury brands than an LVMH. >> Yeah. So.
2:22:21
>> Have we hit peak Temu and Shein? >> Oh yeah, dude.
2:22:25
I mean, uh Trump got rid of uh section 321, de minimis.
2:22:31
That totally crushed those brands.
2:22:32
It's like, I I mean, you know, Temu >> do you go if you want to shop like a billionaire?
2:22:38
>> [laughter] >> There's nowhere else to go. you're right, man. I don't know. rich. com.
2:22:43
[laughter] No, uh Look, I mean, they they're still huge.
2:22:48
They they run a lot of ads.
2:22:48
And like Temu's publicly traded under like PDD or whatever.
2:22:53
So like they have a huge business in Latin America.
2:22:55
They have a huge business in Southeast Asia.
2:22:56
But like you their whole arbitrage was just flooding into America with with low-cost goods.
2:23:03
>> Direct from the factory, right?
2:23:04
>> Yeah, and it just got kind of that kind of blew up. >> It's gone. Interesting.
2:23:08
Uh have you been surprised that live shopping has been slow in America? >> No.
2:23:15
>> Everyone was predicting this from China.
2:23:16
Oh, if it's big there, it's going to be big here in a year.
2:23:17
And it feels like it's been very slow.
2:23:19
But what what what have you seen in the live shopping? >> Yeah.
2:23:23
Look, people are very excited about Whatnot.
2:23:24
Uh and they are putting up impressive GMV growth, but it is very much dependent on like the trading card bubble.
2:23:31
It's like that's that's what it is, right?
2:23:33
So it's live shopping in America.
2:23:36
>> is what I was saying yesterday.
2:23:36
It's like it's effectively I'm not going to call it gambling, but there there's some speculation happening on uh there's a there's a massive I would say spec you would want to figure out what percentage of GMV is driven by speculation and I would expect that it's significant.
2:23:56
>> Yeah, like people aren't going on there to buy their everyday essentials, right?
2:23:59
They're going on there for the excitement.
2:24:01
>> Where is in China there's live shopping where you can buy a tomato and that's just not happening.
2:24:06
>> And women are buying dresses and it's the whole thing.
2:24:07
>> Yeah, of course the whole person picking up one piece of clothing put it down and then the next one like we've [snorts] seen that video and we have yet to see that in America. >> Yeah. >> Why is that?
2:24:17
>> I think I mean dude Americans can't watch anything at the same time, right?
2:24:22
Like we are also like consumption based on my like on demand everything, right?
2:24:29
Everything is catered to us all the time.
2:24:30
I just don't think there's any value to it actually being live.
2:24:31
We're we're doing Tik Tok shop lives and we drive like three to $400 in revenue per hour that we're live.
2:24:37
So it's like some people are buying stuff when it's live, but the real value is like getting all that content and just running it whenever you have 4 hours at night to to go live on to something. Right?
2:24:46
And I really just think it comes down to there's a lot more people in China and they're spending a lot like that the screen time per person is still way higher over there and just you know, it developed over there.
2:24:57
It's more of like a native sport. >> Yeah.
2:25:00
Have you looked at advertising on Netflix?
2:25:03
>> Uh we have looked at it.
2:25:05
Uh the CPMs are not very good and the here's the thing is we buy a lot of TV ads and we buy them like like directly through the network.
2:25:12
So like you know, Fox will have something like we just got an email for the World Cup and it's like you could run a 30-second spot during the World Cup the USA game and it was like a $25 CPM, right?
2:25:21
Um and Netflix they you will get a lower tier of consumer because it's price gated, right?
2:25:29
It's the lowest tier of consumer because of the ads and they want a $45 CPM, right?
2:25:32
It just doesn't make a lot of sense.
2:25:34
>> Yeah, that makes a lot of sense.
2:25:34
Uh what about um uh infomercials?
2:25:39
Have you ever thought of running one?
2:25:41
It's like a full hour middle of the night type of thing.
2:25:44
I've >> That's the new meta.
2:25:47
>> wanted to get I wanted to It feels like a bucket list item on an entrepreneur.
2:25:52
>> Well, you have the studio, bro. Let's shoot one. >> We should shoot one.
2:25:56
>> [laughter] >> I'm I'm I'm down to be a >> 1 hour pitching you a 3 hour live stream.
2:26:01
I interest you in hitting the subscribe button, potentially leaving us five stars.
2:26:05
>> I want it hyper personalized infomercial though, where somebody's just like falling asleep on their couch at 1:00 a. m.
2:26:11
and you're like, "Hello, Sean."
2:26:14
>> I I I >> [laughter] >> feel like I've seen some of those YouTube experiment ads where uh they they they didn't gate it and so if you didn't click skip, you'd wind up watching like 12 minutes of an advertisement.
2:26:24
>> you know I IKEA did that, but it was like 12 hours.
2:26:27
IKEA was like, "We're going to read every product we have, but please click skip." Right? >> Fantastic.
2:26:31
>> Um >> But you've never done an infomercial?
2:26:33
>> No, but I met a guy one time who made like, you know, his whole business, $80 million a year, was selling hoses on infomercials.
2:26:37
And he's like, "Yeah, I have a great hose.
2:26:39
I just sell it on infomercials."
2:26:41
Then he gets it in a Home Depot or whatever.
2:26:42
So, look, you can make a lot of money in a lot of different ways.
2:26:44
So, >> But you got to you got to focus a little bit.
2:26:47
You got to get things done.
2:26:48
>> Yeah, right now I'm trying to sell like a bunch of random accessories to men, and my goal is to get to a billion a year in annual revenue.
2:26:54
I'm like three or four years away.
2:26:56
And if I do that, my life's good. >> Fantastic.
2:26:58
Well, thank you so much for coming on the show.
2:27:02
>> You want to close it out with us? >> Yes. >> All right.
2:27:04
>> We got one last advertisement. It's for MongoDB.
2:27:06
What's the only thing faster than the AI market?
2:27:08
Your business on MongoDB.
2:27:09
Don't just build AI, own the data platform that powers it.
2:27:12
Thank you so much for coming on the show.
2:27:16
Leave us five stars on Apple Podcast and Spotify.
2:27:17
Sign up for our newsletter at tbpm. com.
2:27:18
And we will see you tomorrow at 11:00 a. m. sharp. Goodbye. >> Yeah.