searchlore

0:03

[Music] [Music] Hey. Hey. Hey. Heat. [Music] Hey, Heat. [Music] down.

1:14

Hey [Music] [Music] Trust the experts. Draw fire code. [Music] Double glaze. Infinite glaze. Triple glaze.

2:34

The glazing freaking glaze. [Music] Heat. Heat. N.

2:41

[Music] [Music] [Music] We came to this world to reach the stars. We came to this world. to see a future.

3:34

[Music] We came to this world [Music] to shape our future.

3:50

[Music] Reaching to feel the new If you let me Feeling [Music] [Music] [Applause] me.

4:42

[Music] You're watching TVPN.

5:03

>> Today is Thursday, August 21st, 2025.

5:03

We are live from the TVP Ultradum, the temple of technology, the fortress of finance, the capital of capital.

5:09

Um, we have a post from Jeff Dean, the legendary programmer.

5:13

Tyler, do you know who Jeff Dean is? >> Dumb question. >> Dumb question.

5:20

>> Of course I know Jeff Dean.

5:21

>> Of course Tyler knows Jeff Dean.

5:21

Uh, do you know uh uh the list of Jeff Dean jokes?

5:27

So Jeff Dean is a legendary programmer, been at Google for a very long time.

5:31

Um, one of the best to ever do it.

5:34

And there's a list on it was on some random website for a while.

5:38

Somebody finally put on GitHub.

5:38

Uh these are these are the facts about Jeff Dean.

5:43

Jeff Dean proved that P equals NP when he solved all NP problems in polomial time on a whiteboard.

5:49

Jeff Dean's pin his ATM pin.

5:52

It's the last four digits of pi.

5:54

Uh this is this one's true actually.

5:56

When Jeff gives a seminar at Stanford, it's so crowded that Donald Kuth has to sit on the floor.

6:01

Apparently that actually happened.

6:05

>> My favorite one is uh to Jeff Dean NP is no problemmo. >> No problemmo. Yes. Uh, another one.

6:07

Jeff Dean Cup got promoted to level 11 in a system where the max level is 10.

6:15

Apparently, this happened at Google.

6:17

They had 10 levels of software engineering.

6:18

He was so good they were like, "We got to make a 11th level.

6:22

>> Let's take this one to 11." >> Yeah. Uh, yeah, it's great. He's he he's a legend.

6:26

Uh, you don't explain your work to Jeff Dean.

6:28

Jeff Dean explains your work to you.

6:30

Uh, there was another one that I thought was funny, which is uh Jeff Dean uh his his keyboard only has two characters, one and zero.

6:37

He just programs in binary.

6:39

I think that's hilarious.

6:44

Uh Jeff Dean has exactly two keys on his keyboard, zero and one.

6:46

Uh errors treat Jeff Dean as a warning.

6:49

Uh compilers compilers don't warn Jeff Dean.

6:52

Jeff Dean warns compilers.

6:54

Anyway, you could go through these for all day. Google.

6:59

It's basically Jeff Dean's side project.

7:00

Uh anyway, he's a legend. People love him. Uh I love him.

7:02

I would take a bullet for Jeff Dean.

7:04

One of the greatest to ever do it.

7:05

Seriously, he's he's amazing.

7:08

Um, but anyway, he posted something very interesting.

7:10

So, he said, "AI efficiency is important.

7:12

Today, Google is sharing a technical paper detailing our comprehensive methodology for measuring the environmental impact of Gemini inference.

7:20

We estimate that the median Gemini apps text prompt uses 0.

7:26

24 W hours of energy, equivalent to watching TV, average TV, for 9 seconds, and consumes 0.

7:33

26 26 milliliters of water, about five drops.

7:37

Uh so they did all the averages, >> it's just a few drops.

7:41

Uh and so this of course is to is to counter the narrative that was growing online uh in the media and just in the general like anti- AI anti-tech world that every time you fired off a prompt on any AI system you were lighting the Amazon rainforest on fire using enough power to power an American city stealing water out of the homes of American people.

8:05

Uh and that does not appear to be the case.

8:09

And to be clear, many of these new data center projects are you will use the energy equivalent of millions of single family homes.

8:16

>> So this is still energy intensive, but what matters is that the underlying products still get more and more efficient. >> Yes.

8:24

>> As they get more efficient, we'll probably use them more. Yep.

8:27

>> Hence uh you know the massive energy consumption. >> Yeah.

8:29

Somebody was comping one of the latest uh one of the latest buildouts as um I think it was in the journal actually today.

8:37

Um it's like enough power to power a small US city and they think they maybe said like Austin, Texas or something is like one data center is enough to power that.

8:46

Um but again this is training a model that will be used for years probably and the training run only runs for a couple months and then it's distributed.

8:56

Yes, it's enough to power a million homes but a billion people wind up using the stuff.

8:59

people wind up using the stuff. So um you so you have to think about the overall value that is created by these systems and obviously uh everyone wants to optimize them not just the environmentalists also the companies that if they don't if they don't reduce

9:14

the amount of energy they use their margins are terrible and so there are it's not just Jeff Dean and the and the crew at Google who enjoy optimizing systems and I'm sure got a ton of joy out of squeezing such performance out of uh Gemini on the inference side. Uh but

9:27

Uh but it's also just the CFO uh who says, "Hey, why are we why are we spending so much energy on a single Gemini Apps text prompt? Let's get this down.

9:38

Let's optimize this thing so we can actually make some money." Um which is great.

9:44

>> And Bubble Boy over on X had a good post.

9:46

They said, "AI inference is the new highfrequency trading and the companies that serve models and don't realize that will die."

9:52

Thought this was a solid take.

9:55

It's just that performance and efficiency and speed are going to matter more and more. >> Yep.

10:02

>> And uh if you're in the business of serving up >> Yep.

10:06

>> farmtotable inference, better make sure you're efficient. >> Yeah.

10:09

Jeff Dean was like the original superstar AI researcher, just software engineering researcher.

10:14

Honestly, built a lot of the systems that Google built GCP on top of.

10:17

And what's interesting is we just we just realized or came to terms with the rise of the superstar AI researcher and that was underwritten probably at least in part on the back of super intelligence.

10:32

If you find a researcher who creates an insane trick that unlocks trillions of value >> inapp purchases for your companion >> maybe maybe but but no XAI has not has has notably not been competing in that space.

10:48

We have not heard about uh billion-dollar aqua hires at XAI, hund00 million offers at XAI.

10:53

Uh Elon has said, "I'm focused on engineers, not researchers."

10:58

But the reason you would underwrite paying a researcher $und00 million is that if there's a 1% chance that they create a new uh advancement in how large language models are designed, trained, implemented, well then that could drive $10 billion in value overnight, just like what happened with uh GPT4, RLHF, post- training, pre-training, all that all that stuff.

11:19

Any anyone who was involved in that created a ton of value, right?

11:23

Um and so basically Meta has been saying hey we're going to comp the researchers against that uh optimistic scenario where they come up with some breakthrough and that drives a ton of value.

11:32

Now the question is are we going to shift into an era where the superstar AI inference optimization engineer is worth a hundred million?

11:44

>> Well I will say that I you know the big labs are poaching from the high frequency trading firms.

11:51

Yeah, Mark Chen worked at Jane Street. >> Yep.

11:54

>> I mean, that wasn't a poach, but uh there are there are lots of people that come over.

11:58

My my question is is if I come in >> So, the poaching is just not happening within the labs.

12:02

They're looking to areas. >> Yeah.

12:06

Sam Alman came out and said, "Hey, you're working at Yeah.

12:08

If you're working at a hyper trading shop, you should come over to AI research."

12:11

Um, but my question is if if I'm if I'm a superstar inference optimization engineer and I go to one of the cloud providers that's that's going to be generating, you know, tons and tons of tokens.

12:24

They have these token factories, these massive AI data centers.

12:26

And I come in and I say, I can make them I I can make them use 5% less energy, 10% less energy that could drive billions of dollars points on the board.

12:38

>> Put some points on the board.

12:38

And so, are you going to get comped on that? I don't know.

12:40

Will will Jeff Dean be getting a maxed out contract?

12:41

I'm sure he's had a maxed out contract this entire time, but can he go further? We'll see.

12:47

Um the uh the other interesting thing here is just um how Google exactly pushes back on the media is very different than what we see from other tech folks.

12:59

So um obviously there there have been uh there have been posts about water usage in AI inference um that have gone out in the mainstream media.

13:11

Uh Sam Alman did address uh GPT4's water usage specifically. They did an analysis.

13:18

He dropped it kind of in the middle of a long post.

13:19

I think in the soft singularity or something.

13:22

He just mentioned it randomly. Yeah.

13:24

>> Um >> the uh I like this approach of not being combative, not calling anyone out for being wrong, just saying, "Hey, we take this seriously.

13:33

We have this uh this doc. They ran the numbers.

13:35

We ran the numbers and they're willing to talk about here's the here's the actual here's the actual post. We have a blog post.

13:40

We have a paper uh you can see our full methodology where we're we're we're committed to this stuff.

13:44

Um and uh and also it it it serves as like a big flag for hiring saying hey we're the type of company that's dropping our energy footprint on Gemini apps text prompts by 33x like that's got to be exciting and that's got to be on the back of some really solid engineering.

14:00

So maybe you want to come work there.

14:01

I mean it it's it's worth noting that despite the wind surf debacle, >> yes, >> Gemini and the Google team broadly have been threading the needle avoiding a lot of controversy.

14:13

You don't hear people about getting getting oneshotted by Gemini.

14:17

You don't you don't see them posting, >> you know, 3D renders of scantily clad Yeah. >> uh AI characters.

14:23

Uh they're they're just uh focused. Yeah.

14:26

Clearly >> uh I do think that we are going to see a new genre of not necessarily hit piece but negative AI story.

14:34

It will be before bad person did bad thing here's what they talked to chat GPT about or here's what they talked to their AI assistant about.

14:48

And we've seen this with uh there's been a few articles of what what were people ch talking to chat GPT about before they ended their lives or anything like that.

14:57

But basically these tools are going to be so diffused within society that everyone's going to be using them constantly.

15:04

And for a long time, if there was a high-profile crime, uh, the courts would look up and subpoena Google search records and say, well, they were googling how to sell drugs across the world and they were how to start a drug cartel.

15:20

And so, that was kind of a smoking gun.

15:22

Now, it will be phrased in a different way, not just it won't just be in a knowledge retrieval context.

15:29

At least the the a lot of the articles won't be. >> Yeah.

15:32

They won't just say before before Pablo the next Pablo Escobar started a drug cartel he googled how to start drug cartel.

15:40

It'll be before Pablo Escobar started uh was it the Medí cartel? Is that the one? Yes. Yeah. The Medí cartel.

15:48

He was talking to Chachi VT about starting a drug cartel.

15:51

And did Chachev talk him into it? Was it being Pablo?

15:57

That's the best idea ever.

15:57

you're no one's thought of starting.

16:00

The latest South Park arc is all about sick and there's a scene where I don't I don't even know the character's name, but there's a husband and wife and they're talking and the and the and the uh husband is in a Khole and for the wife to try to break through, she's acting, she's pretending to be Chad GBT and uh >> and apparently David Sax made it in there.

16:23

Little cameo from Sax South Park.

16:25

Well, if you want an AI agent that won't be sickopantic to you, get on RAMP. >> Ramp. com. Time is money. Save both.

16:30

Easy to use corporate cards, bill payments, accounting, and a whole lot more all in one place.

16:35

RAMP isn't going to tell you, "Oh, yeah, that expense is fine. Don't worry about it." Yeah.

16:39

Like, "Yeah, go out for the state dinner."

16:41

It's No, it's going to say, "Let's put Let's save some time and money. >> Let's save both. >> Let's save both. Let's save both."

16:46

Uh, anyway, we should do the inexurable rise of drug cartels.

16:50

I've been waiting to do this story for days and we and we put it off.

16:54

But there is a big read in the Financial Times that I want to read through because uh if you're if you're tuning in for the first time and have been living under a data center, uh Jordi Hayes is is a bit of an expert.

17:05

Bit of a bit of an expert when it comes >> certainly not an expert, but a but a but I >> you famously fall asleep listening to stories of drug cartels. Correct. >> Yes. It's relaxing to me.

17:13

So >> it's it's it's it's business businessoriented content that's just far enough outside of my world that I don't that it doesn't keep me up.

17:23

You're not like wait should I invest in this? >> Yeah.

17:27

So so if you >> it doesn't turn like should I have the founder on the show like doesn't turn into work.

17:31

>> I've maybe listened watched like two episodes of of Silicon Valley TV show.

17:36

>> Oh it's way too >> And it's because it's not relaxing to me.

17:39

Totally >> because it's just prompting I got to respond to that email or I I I got to >> Yeah. Yeah. Yeah.

17:44

So, we have uh the inexurable rise of drug cartels in the financial times.

17:50

With the cocaine with the global cocaine business booming as never before, organized crime groups are diversifying into a swath of other illicit activities.

17:58

Political leaders across the region warn about the threat to stability.

18:01

We're going to find out what America is going to be doing about this.

18:05

Do we have a particularly American sound effect today? Thank you, Jordy.

18:10

Over the past century, the Tikuna, the biggest tribe in the Brazilian Amazon, have fended off threats from loggers and illegal miners in some of the most remote tracks of the vast rainforest.

18:21

But the latest challenge is more intense than anything they have experienced before.

18:25

Drones were flying around this area last year, says Major uh Jonata Jo Jonatas Suarez, the regional military police commander speaking in the village of Ori about 1,000 kilometers west of the city of Manau.

18:39

Drug traffickers, he adds, were stopping off there, storing their cocaine, and then putting drones up to check what was going on before continuing their journey. >> Picture this. They're in a jungle. They're hanging out.

18:52

Normally you can't get a lot of visibility, so they just pull the they just they just send drones up.

18:58

They look around, make sure everything's all clear before they make any moves. Right.

19:02

>> That's super interesting. Wow.

19:04

Informants said there were 200 kilos of the drug stored there, but police were unable to locate the stash.

19:08

Uh, one of the world's largest and most inaccessible stretches of rainforest, the upper Amazon, has today become a superhighway for the export of cocaine to Europe, its fastest growing global market.

19:21

I didn't know that Europe was down like that.

19:23

Uh, every week, Brazilian law enforcement officials say tons of cocaine make their way from illicit production labs in the jungles of neighboring Peru and Colombia through the Amazon to Manouse and the port of Balem for export to Europe and Africa.

19:40

Sometimes the traffickers pay locals to smuggle just a few kilos down river or they hide larger amounts under the floorboards of motorboats.

19:46

At the other end of the scale, semi-submersible craft dubbed narco subs capable of carrying several tons of drugs have been detected on rivers, feeding the Amazon from Colombia to and Peru.

19:59

Imagine buying directly from the cocoa producers for 300 a kilo and then selling a refined kilo for 60,000 euros. Wow.

20:06

In Europe, uh that changes the life of of the guy who's trying his luck bringing it.

20:13

So it's high risk but high reward at least financially. >> Yeah.

20:19

So the the the if you haven't uh gone down this rabbit hole it's f the the history of the way they make these uh submersibles >> Oh yeah >> is fascinating.

20:29

So started out they would obviously use regular boats. Those are still used.

20:33

Eventually they realized we could make these kind of homemade submarines that are not like traditional submarines and that they're going deep under the water but they're just hovering right at the surface.

20:44

>> They might have like a snorkel.

20:44

So you could see one like if you were looking out at the ocean five five miles away, you would just not pick it up.

20:49

Yeah, it's like >> But if it was sitting like right if you were on a dock and it was right there, you would just see it right there.

20:54

It's just like one foot under radar. >> Got it. Interesting.

20:57

Um and it probably makes it a lot less risky.

20:59

You don't have to deal with pressurization.

21:00

You don't wind up in um an ocean gate scenario because you're not putting it in any sort of extreme scenario.

21:05

in any sort of extreme scenario. So the more recent trend is they're now starting to leverage Starlink to maintain connectivity and they are no longer having to put passengers and actual pilots >> because they can be remotely remotely control them

21:20

>> which just reduces I mean again these the cartels don't want to lose um lose out on shipments like if one gets captured but uh it's certainly better to just lose the product over a team member if that's uh if they want to go so far as to I don't know if they have these people in Slack uh or WhatsApp groups. Uh but uh

21:37

Uh but uh >> chat chat is uh proposing spacking a cartel.

21:43

We are firmly against that here at TBPN.

21:46

We will not be uh >> Don't give any ideas, folks.

21:51

>> Um the global co you asked for a business with free cash flow. I brought you one. >> Yes. Stop complaining.

21:56

Uh the global cocaine business is booming as never before. I had no idea.

22:00

Uh Europe's drug habit has grown so fast in the past two decades that it has overtaken the US as the biggest cocaine market. That is crazy.

22:11

They really are like 50 years behind everything.

22:13

They don't get their movies for a few years.

22:15

They don't AC yet, but and and they're just getting the drugs that Americans were doing in the 80s, I suppose.

22:20

Um uh while the narcos are hooking new users in the Middle East and Asia, flushed with money, Latin America's cartels are diversifying from drugs into a swath of other criminal activities.

22:31

We think 2024 was the most lucrative year for organized crime in Latin America, says Jeremy McDermott, co-founder of Insight Crime, which tracks illicit activity in the region.

22:42

This was driven principally principally by three criminal economies. The first is cocaine.

22:49

Lagging not that far behind uh lagging not that not far behind that is gold. Gold. They're smuggling gold. Okay.

23:00

Number three is human smuggling and human trafficking.

23:04

>> I would imagine the the gold activity is tied in to the first one. >> Sure. Sure. Yeah, that makes sense.

23:09

Uh drugrelated crime and violence used to be concentrated in the narcotics producing nations of Peru, Colombia and Mexico with nations such as Argentina or Chile uh largely untouched.

23:20

Today such violence has become a feature of in virtually feature of life in virtually every country in the region reaching even former havens Costa Rica and Uruguay uh a nation sometimes as the Switzerland of Latin America.

23:33

This is why Pam Bondi is putting out 50 million for Nicholas Maduro. That's right.

23:39

Because Maduro being accused accused of narco terrorism. >> Well, yeah.

23:42

Effectively having the government of Venezuela >> Yes.

23:46

>> run >> basically runel not even getting a cut.

23:50

>> Oh, they're actually >> like that they're being accused of just actually running a a drug cartel.

23:53

Well, that might actually be the answer to the drug war.

23:57

Because if you put the government, if you put the folks who did the DMV in charge of getting drugs, you have to, oh, I'd like some cocaine.

24:03

You go, you take a ticket, you wait four hours, they say, "Ah, we're out. You got to come back.

24:08

You got to mail in this form.

24:10

You got to fill out that form."

24:12

All of a sudden, you're like, you know what? I've cleaned up.

24:13

I'm just going to stick with I'm just going to stick with caffeine. >> Yeah, >> I'm good. I'm good.

24:17

I don't I don't actually need any cocaine.

24:21

>> Um, anyway, really quickly, let me tell you about Reream.

24:25

one live stream, 30 plus destinations, multiream and reach your audience wherever they are.

24:28

Stick to live streaming, folks.

24:30

You don't need to go into narco terrorism.

24:35

>> Pivot to live streaming.

24:37

>> Um, >> so, uh, drugrelated crime and violence, um, organized crime has become the main threat to institutional stability of our nations, uh, says the former president of Costa Rica.

24:47

No Latin American country today can escape this.

24:49

Some of the >> I wonder what's going on with El Salvador, right?

24:52

They locked up all their They locked up >> the one that has Bitcoin, too.

24:55

He's a Bitcoin max, right? >> Yes.

24:58

They've locked up all the criminals. Yeah. >> Safer than ever. >> Yeah.

25:03

>> You are they, you know, through that are they able to prevent trafficking happening within their borders? I don't know. >> I don't know.

25:09

It's probably like anything else.

25:11

Probably like a cat mouse game that is unending.

25:13

I went uh I I've I've been on a couple surf trips to El Salvador back in the day and it was >> historically known as one of the danger most dangerous countries to visit. Yeah.

25:25

>> And then the cartel at the time realized that it was terrible for business if you didn't get tourists coming through the country.

25:31

They couldn't sell drugs to tourists. >> Oh, sure.

25:34

And so by the time I got there, there had been some unspoken deal which was that if anybody harmed >> a a tourist in any capacity or stole from them, they would get like a insane beatd down from the gang.

25:49

>> Yeah, it does feel like I I spent over a month there, probably like six weeks in total, >> and never had an issue at all.

25:55

Could walk around at night, could go anywhere.

26:00

And uh and so they they had their sort of own cartel enforced safety mechanism. >> Yeah. Yeah.

26:07

It does feel like there's a uh there's a bit of a truce. Yeah. Gold Rock nailed it.

26:11

Cartels are just CPG companies with better PR.

26:14

Not necessarily better PR. Uh no PR or negative.

26:19

They they they build in silence.

26:21

Um but uh >> CPG companies with no brands, right? >> Yes. With no brands.

26:26

>> They have the distribution. They have logistics. >> Yeah.

26:29

They they have very sophisticated operations.

26:31

They just don't put uh a brand there's no branding on the product. >> Yeah.

26:36

Uh some of the larger players are so powerful they challenge even the biggest states forging links with longestablished organized crime groups in in Europe and Asia and generating billions of dollars in profit.

26:45

At the heart of Latin America's vast criminal enterprise is cocaine.

26:49

By far the most profitable illegal business.

26:51

production, seizures, and use of cocaine uh hit all-time highs in 2023, making cocaine the world's fastest growing elicit drug market.

27:00

Illegal production skyrocketed to 3,78 tons.

27:05

>> It's funny that they had to say illegal production because I guess >> there must be some legal production. >> Yeah.

27:11

What's the What's the fastest growing non-illegal uh market? Probably uh AI tokens. Um >> well, yeah. Yeah.

27:19

>> well, yeah. Yeah. the the question here because the cartels obviously generate massive amounts of cash flow and profits >> and then they have the choice of reinvesting that in the illegal business or reinvesting it in legitimate business

27:30

which is kind of laundering right >> gas and you can imagine you could imagine Mexican data centers being built out with cocaine dollar >> neocloud >> in Colombia the world's produ the world's biggest producer cocaine production rocketed 53% from 2023 to uh from 2022 to 2023. That is crazy. A That is crazy.

27:49

A senator and a presidential hopeful was shot in June at a rally in Bogota and later died of his wounds, raising fears that the nation might relapse into cocaine fueled political violence that scarred it in the 80s and 90s.

28:02

But cocaine is only part of the picture.

28:05

Uh experts say Latin America's organized crime groups now run a portfolio of diversified businesses robust enough to absorb cyclical downturns in one area like a like a legal conglomerate.

28:18

Criminals are now starting to accumulate revenue at the scale of national GDPs with none of the burdens of a state.

28:24

They're operating across borders in countries with legal systems that were built for a past age.

28:27

The combination of illicit businesses.

28:29

Um they're no longer drug trafficking organizations.

28:32

They will move anything that transits their area of control.

28:36

You have all of these new markets opening up. Incredible.

28:40

Anyway, you should go and read the full uh the full deep dive in the Financial Times.

28:44

Well, here here's the here's here's the thing that's more interesting.

28:47

So, Trump recently designated eight Latin American cartels as foreign terrorist organizations and secretly signed a directive to the Pentagon to begin using military force against them.

28:56

The New York Times reported this month and as part of that, so now there's spy drones flying over Mexico, like gathering information in real time.

29:05

And the US has deployed uh naval forces into the Caribbean, basically the Gulf of America, formerly known as the Gulf of Mexico. Yep.

29:15

>> And um it seems pretty it seems somewhat unlikely that they would start striking the cartels.

29:22

I don't know if we need to open up a new front at the moment. >> Yep. Yep.

29:27

>> Uh but it's certainly not uh totally off the table.

29:30

>> Well, this is part of a three-part series.

29:32

Uh part two will be on Syria's drug war and part three will be on the illegal gold trade, which is very interesting. Um any Okay.

29:39

So, you're a drug dealer, right? >> Yes.

29:43

So, are you going to start a cartel? >> Uh, no.

29:46

I was talking to Jordy about that.

29:47

Uh, the the the drawback of selling nicotine pouches is that there there really are no cartels.

29:52

Uh, >> closest thing to cartels yet are the big uh are the big uh big nicotine. >> Yeah.

29:59

The big tobacco companies have uh have really really solid control over their supply chains and distribution.

30:05

>> They won't kill an upstart like you, but they do try to make startups lives uh startup life a bit harder. Yeah, a little bit.

30:13

It's more just like they they have a monopoly over the distribution in retail, which is where people buy nicotine containing products.

30:22

And so, um, they don't they can just kind of sit back like they're chilling.

30:28

It's kind of like Google's search monopoly.

30:29

Like, they're just they're just good.

30:32

They don't really need to be overly aggressive.

30:34

Um, and Google doesn't really need to, you know, take shots at Bing or like try and like tie Bing up in court or do anything because like people just use Google and they're good.

30:45

Anyway, um, let's The chat says Tyler unlocked Grock Premium for sure. Uh, I missed something. What's with Tyler Fit?

30:54

Uh, it is Thursday and it's summer and it's a mid August Thursday afternoon. >> Yeah.

31:03

Uh anyway, let me tell you about public.

31:05

com investing for those who take it seriously.

31:06

They got multiasset investing, AI powered analysis, and they're trusted by millions folks.

31:10

We have a very important post here from East Village guy.

31:16

He says, "Being a salesperson that doesn't drink or golf is like playing Scrabble with dyslexia." Uh yeah, the golf.

31:26

>> Uh the real the real uh fighting with one hand tied behind your back would be uh being a vegetarian and not being able to go out to those steak dinner.

31:33

I think you can get by without drinking.

31:37

>> Golf is important, but you know, there's other things to do. You could go track days. >> Track days. >> Do something else.

31:43

There was a fun article in the Wall Street Journal.

31:46

>> But if you can't go stto with a prospect, you're gonna have a tough time. >> Yeah.

31:50

Oh, this this article on the front page of the Wall Street Journal really tells you about uh what what slow and news day it truly is.

31:56

Uh nine beers, nine hot dogs, nine innings. Challenge accepted.

32:01

A gutbusting race against the clock has taken off among baseball fans.

32:06

I tried it, says Xav uh Xavier Martinez in the Wall Street Journal.

32:09

On the front page, this is front page Wall Street Journal news.

32:12

Uh I lean back in my nosebleleed seat.

32:15

Seven empty cans of Kors Light littered around me and eight hot dogs marinating in my stomach.

32:19

I tried to remember how I ended up in this place.

32:24

So sick and yet so determined to keep going.

32:26

I'm not a big drinker, nor can I remember ever craving a hot dog. Oh, right.

32:31

I was attempting the 999.

32:35

Uh uh now a a ballpark tradition.

32:39

The challenge entails eating nine hot dogs and drinking nine beers in the course of a baseball game.

32:45

I had found out I had found a group of friends who planned to take it on during a game between the New York Mets and the Atlanta Braves and asked if I could tag along.

32:53

Against the advice of every cardiologist ever, I decided to try it myself.

32:57

Joining me in attempting the challenge were Tim uh a 29-year-old software engineer.

33:01

Let's hear it for the software engineers, folks.

33:04

Uh, and William Lee, a uh, a 38-year-old analyst.

33:07

Shirona Lynn, a 31-year-old writer.

33:10

If things went wrong, we could share the janitorial duties.

33:14

Wild wild story in the journal just about this.

33:18

>> Just the important stuff.

33:19

>> Just the important stuff.

33:19

And he truly like wrote he wrote he wrote this piece fully. Like it's uh it's great.

33:25

>> Does it kind of trail off to like he's clearly feeling the effects? >> Yeah.

33:30

Uh, I was thinking about th this reminds me of the 999 that a lot of people in Silicon Valley do.

33:33

Nine glasses of Screaming Eagle, nine servings of Beluga Caviar, nine different horses competing in DR. >> Yes.

33:43

>> And so that's the when I heard he was doing the 999, I thought that's what he was doing.

33:48

>> Venture capitalists heard about the 996 and they decided to adapt it for >> Yes. the 999.

33:52

Uh, our fears our first beers had been a respit from a sticky New York afternoon.

33:57

Those Nathan's hot dogs had their had surprised me with their juicy snap.

34:01

It's now or never, I recalled Lee saying when the game began and we naively clanked our 12 ounce cans together in celebration of gluttony.

34:08

But now with time running out and my body at war against itself, doubting doubt was creeping in.

34:15

I'm limping, but I'm getting there, said Digel. I was feeling worse.

34:20

The 999 has existed for at least a decade, though for much of that time it was largely the domain of frat bros and competitive gorgers.

34:26

Recent social media attention and spin-offs have given it new popularity.

34:30

NFL great JJ Watt uh completed the challenge last month.

34:33

Bar stool sports this year debuted a 9999 uh challenge with a nine member team that included competitive eating champion Joey Chestnut.

34:45

Skateboarders gather each year in Brooklyn to uh park to consume nine hot dogs, nine beers while completing nine specialized kick flips.

34:54

That sounds extremely difficult.

34:56

>> Specialized kick flips.

34:58

>> Probably like a kick flip, a heel flip, uh a pop shove it, right?

35:05

>> That's not a kick flip.

35:06

>> It's I think that's I think he's using kickflip as an abstraction on top of just flip trick generally. >> Yeah.

35:13

Well, he should he should respect >> he should respect your culture.

35:15

Uh, how how advanced can you do?

35:18

Can can you do a double kick flip?

35:20

Can you do a triple kick flip?

35:22

>> I probably couldn't right now in loafers, but I could land a kick flip in in uh five minutes if I was if I was.

35:30

>> But but at your peak, at your peak, could you do a 540? Could you do a 720? >> All of them. >> You could do a 900. >> A skateboard company. >> You could do a 900. >> 900. No. Okay. So, there is a limit.

35:41

Those are those talking about halfpipe now. >> Yeah. Okay.

35:44

I I'm just I'm trying to understand. >> I had a mini ramp.

35:45

My dad built me a mini ramp with scraps.

35:48

My dad and my uncle built a mini ramp with scraps in our front yard.

35:52

It was probably the worst thing ever. >> What about a varial? >> Uh varials. Yeah.

35:58

Switch nollie >> quad heel flip. >> Quad heel flip.

36:04

>> That's a special trick from Tony Hawks game. >> Yeah. Yeah. Yeah. XX.

36:06

What are the What were the the >> SSX?

36:09

SSX was the uh was the snowboarding one. Great game.

36:12

Uh you know what else is great? Vanta.

36:14

Automate compliance, manage risk, and accelerate trust with AI.

36:17

Vanta helps you get compliant fast. And we don't stop there.

36:19

Our AI and automation power everything from evidence collection and continuous monitoring to security reviews and vendor risk.

36:26

Whether you're starting up or scaling. >> Oh, guess what, John? A little bit different. What's up? >> Investment alert. >> Investment alert.

36:32

>> The White House at White House has just posted investment alert.

36:35

Adrian is investing 200 million to build an advanced manufacturing facility in Arizona.

36:42

I said it before, America is cultivating Chris Power and uh >> they really are cultivating Chris power.

36:51

>> The new the new startup meta launch videos are done. Don't do launch videos.

36:55

>> You got to get >> you got to get the White House to post uh your announcements from their main handle.

37:01

or you can get Patrick Oshanaugh to post uh and do a whole profile on you.

37:06

Uh so there there are a bit of dueling dueling articles today. Yeah.

37:08

Uh because um the the Wall Street Journal is talking about a school that embraces AI learning.

37:17

We'll run through this then we'll do the the Patrick Oana Colossus profile.

37:22

Uh billionaire Bell Aman has a new fascination.

37:25

in a fast growing private school that assues lessons on diversity, equity, inclusion and uses artificial intelligence to speed teach children in 2 hours.

37:33

Alpha School is launching a New York location in September and the investor and social media commentator has been acting as something of an ambassador for the institution.

37:42

Um on Friday, Aman is set to appear with Alpha's co-founder, its principal, and others on a panel discussion in his Hampton's home.

37:50

Uh the panel on K12 education will be moderated by former financier Michael Milin as part of Milin Institute's Hamilton Hampton's dialogues.

38:01

Uh Alpha School, which calls its teachers guides, says it uses AI enabled software to help students complete core subjects in just two hours daily.

38:10

It claims students learn twice as much as those in traditional schools despite condensed days.

38:15

The schedule allows students to do hands-on activity in the afternoon, which the school says help them build real life uh skills.

38:23

These include five mile bike rides without stopping uh for kindergarteners. >> No stops. >> No stops.

38:29

Five mile bike ride without stopping for kindergarteners. That's pretty sick.

38:32

And exploring personal hobbies through AI generated plans.

38:34

Uh they should learn some pit flips. Learn the 900.

38:38

I would send my children to alpha school. I was guaranteed a 900. Yes.

38:43

>> Kids to do a bike ride and not stop.

38:48

>> If you could actually pull that off.

38:48

I don't know how I don't know how your little guy does, but my little guy >> stops constantly. >> Stops constantly.

38:53

Yeah, it's a stop machine.

38:55

>> He's not a kindergartener yet, though, so anything's possible.

38:56

Maybe Alpha School, maybe AI will make it possible.

39:01

Uh, the school also strives to keep the hot button social issues that have divided grade schools and colleges across the country out of its classrooms entirely.

39:08

uh co-founder Mackenzie Price said in an interview, "We don't let anything political, social issues come in the way.

39:15

We stay very much out of that, uh, Alpha's guides come from a variety of backgrounds and don't typically have teaching degrees."

39:22

I wonder if there will be a backlash to this or like a response to this, like a hyper political school that teaches your child how to navigate the modern political landscape for maximum power.

39:37

So you go there, you learn how to how to really pour the fire on the culture war and create a movement around yourself and get into the White House.

39:44

Could be some interesting alpha there.

39:46

Uh do they not is this why do they call it alpha school?

39:51

Are they trying to like develop alphas? Is that the idea? It's like no bet.

39:54

>> Well, John, if they wanted to develop betas, you can imagine what they would call it.

39:59

>> I would be interested.

39:59

So we need to start a competitor to this. Sigma school. >> Sigma.

40:05

>> Sigma school >> alone.

40:06

It's a school for lone wolves. >> Lone wolves. Yeah.

40:07

And it just plays It just plays literally me songs on repeat.

40:12

It's like class is in session.

40:14

You're going today you will become Ryan Gosling from Drive.

40:19

>> You're going to be operating outside of social hierarchy school. >> Yes. Uh class.

40:24

>> You develop any friends.

40:25

>> We live in a society and you will be the Joker today.

40:27

Here's how to become Joker.

40:31

>> We're training We're training you up to to to join the manosphere one day. >> Yeah. joined the ministry. Yeah.

40:35

Uh >> when you when you enter the blade Yeah.

40:39

When you enter the first grade, you have to choose between Tim Ferrris. >> Yeah.

40:43

>> Andrew Huberman, Lex Freriedman, Chris Williams.

40:45

You get sorted into the houses.

40:47

>> The houses of the manosphere this sigma school.

40:51

>> We're on to something here. >> Yeah.

40:53

We need you to study the blade.

40:56

We're sure >> purely so that in in two decades time you can say so you come to me in your time of need. >> Yes.

41:04

Um the new school aims to launch with 30 or so founding families.

41:08

>> So you still on Sigma school? >> Sigma school. >> Yes.

41:12

>> So imagine you have a 5-year-old, right? They join Sigma school.

41:14

So these are our principles. >> Individualism, right?

41:18

So Sigma operates independently and doesn't seek to be at the top or even a part of a social hierarchy. Yes. >> Like an alpha male. Yes. Solitude. Yes.

41:28

>> So >> you need to teach the 5-year-old find strength and solitude.

41:32

So Uh, you know, you'll be taken to like the base of a mountain.

41:37

You say like, "Okay, class."

41:39

>> Five mile bike ride with all your friends with no stops.

41:41

How about five mile bike ride by yourself? >> Summit this peak.

41:44

You have 8 hours to get to the top of this.

41:46

This is just like a monastery. >> Yes.

41:50

>> You're describing monastery. Yeah. >> Monastery. Okay. No, not yet. Not yet.

41:54

So, so the other aspect of the sigma archetype mystery.

41:58

>> So, teaching a 5-year-old got mystery.

42:01

teaching a 5-year-old to be mysterious. >> Good morning, Mark.

42:03

Uh, Rigav says, "Alpha school replaced teachers with clankers." >> The backlash.

42:11

>> I think the clankers are are uh are >> But alpha school responsible to the teachers.

42:16

I think >> test the hormone profile of the of the the human teachers to make sure that they are also alpha level. >> Yes.

42:24

Um, so anyway, um, let's continue on this. They talk about Aman.

42:29

Um Aman who regularly weighs in on everything from fine dining to the war in Kasa on X had uncharacteristically not said a word about Alpha School on the social media platform until the Wall Street Journal revealed his involvement.

42:42

I don't know why this is such like a revealing thing.

42:46

This is like this is this is just like a a startup that people are doing.

42:50

I don't know >> these parents >> I guess it's because like schooling is important.

42:54

It's like culture war issues and stuff.

42:56

Oh, it really comes down to these parents want their children to get a good education. >> Yeah.

43:01

On Wednesday, he wrote a fawning post that ran like an ad.

43:03

Uh, a person close to the school said Aman is now poised to become a de facto ambassador.

43:10

Uh, he is already imshed in the New York's education scene and sits on the board of trustees of Horus Ma man, the school's most storied, one of the school, one of the city's most storied private schools.

43:20

Aman first learned about Alpha School this year at at this year's Bergkshire Hathway annual meeting after speaking to Sahil Bloom uh author of the five types of wealth.

43:28

Some of the people familiar with the management.

43:31

>> What are the five types of wealth again John?

43:34

>> Cars, watches, horses, merchants, real estate. >> Oh yeah.

43:39

>> Yeah, I think we got up to like 10 sources of wealth. >> 10 of wealth.

43:43

Um anyway, uh he has since been hyping up Alpha School to parents and his social circle.

43:49

Aman has been a spoken critic of DEI programs.

43:52

Um he will appear on a panel Friday alongside Price and Joe Lamont, a Texas-based billionaire who Alpha says has served as its school principal for the past two years.

44:05

Lamont is the founder of Trilogy Software, which became part of his private equity firm, ESW Capital.

44:09

Do you know what ESW Capital stands for? Tyler, pop quiz. >> Pop quiz.

44:18

>> ESW stands for enterprise software >> group. >> No, ESW capital.

44:22

Enterprise software capital.

44:24

Nothing better than that.

44:24

That is a fantastic name. >> Yeah.

44:28

>> Uh he has develop over to the Colossus.

44:30

>> We should we should >> Patrick has a and Jeremy Stern have a new profile in Colossus today. It is available online.

44:37

You can also subscribe to the print edition.

44:39

Um, but the profile is all about Joe Lamont and Alpha School and uh the >> I guess his co-founder in Alpha School, McKenzie, Mackenzie Price.

44:50

So >> crazy lore in here in deeply underrated.

44:55

We've heard a lot of people have known about Trilogy and known about uh Joe Lamont for a long time, but uh to get this level of >> you don't know about Joe and uh it wasn't because you weren't you were living under a data center.

45:07

It was because he is pretty much completely silent for the last 25 years.

45:11

So >> he was uh he's an alpha >> but then he was an alpha transitioned more of a sigma sort of mysterious outside traditional hire.

45:23

>> Extremely sigma extremely sigma.

45:25

>> So anyways I'll give you some background here.

45:26

So uh this is this is from the profile.

45:29

Even after agreeing to break his silence after 25 years Joe Lamont is still reluctant to talk about himself.

45:34

It's hard to get him to relive Trilogy the enterprise software company.

45:36

Let's just give it up for Enterprise Software real quick.

45:40

The enterprise software company he founded in 1989 which by age 27 put him on the cover of Forbes twice as America's youngest self-made scenti.

45:50

So if you are if you are if you've been going through a period of silence and mystery and you've been a sigma and you're trying to become an alpha trying to get your brand mentioned in chatbt go to profound reach millions of customers who are using AI to discover new products and brands. Continue Jordy.

46:05

So he isn't keen to expound on sales builder, which is an amazing name for software company.

46:13

Trilogy's flagship expert system from the 1990s and the world's first billiondoll artificial intelligence product in all but name.

46:22

Ditto ESW, the investment arm of Trilogy that's acquired hundreds of software companies since 2000 and helped make him a deca billionaire. Wow.

46:28

Yet the mention of which makes the otherwise inexhaustible Lamont, who always seems to be straining at at some invisible leash, seem drowsy and bored.

46:35

Nor is he interested in dwelling on his reputation as a tech pioneer in everything from expert systems to manufacturing configuration, college recruitment, corporate boot camps, tech bro culture, remote work, monitoring employees with surveillance software and replacing American employees with overseas contractors.

46:53

uh whether his uh reticence can be ascribed to an inhibition about the things that made him rich or because he's refreshingly adverse to gilding them with hypocritical virtue signaling or because he simply doesn't give an hit anymore is hard to tell.

47:09

The one thing Le Man will talk about for hours on end is Alpha School, the teacher list, homework list, K12 private school in Austin, Texas, where students have been testing in the top.

47:19

1% nationally by self-directing coursework with clankers for 2 hours a day, of course, with AI tutoring apps for two hours a day.

47:27

Alpha students are incentivized to complete coursework to quote unquote mastery level scoring over 90% in only two hours via a mix of various material and immaterial rewards including the right to spend the other four hours of the school day in quote unquote workshops learning things like how to run an Airbnb or a food truck, manage a brokerage account or Broadway production or build a business or drone.

47:50

And you know why you have to go to uh uh uh and actually um subscribe to Colossus Magazine because uh you can get this beautiful picture of this horse >> price. >> Yes.

48:05

Uh the co-founder of Alpha School and Flair the whole >> So when you think about you know thinking about these these courses that they're taking, you know, these K through2 students through how to run an Airbnb.

48:16

I wonder if they found any course bros online that are that and and and trained an AI on on their programs of how to build a a nine figure Airbnb empire.

48:28

>> First first trick there would be to just get those assets on wander.

48:30

Uh find your happy place >> book a wonder inspiring view hotel great many dream beds top your cleaning and 24/7 concier service.

48:37

It's a vacation home but better folks.

48:39

Uh and we have our first guest of the show Miller is here.

48:43

>> Welcome to the stream Chase. How you doing?

48:45

calling in from a log cabin set.

48:48

>> That's a beautiful background. Where are you? >> Great. Television broadly.

48:54

>> I'm in I'm in Jackson Hole right now.

48:54

I have an event that I'm attending out here. So >> amazing.

49:00

>> Found a found a corner of the log cabin to call in to to hang out with you guys. >> I love it. Uh give us the update. What's the news? Break it down for us.

49:10

Um, you know, we are excited and and thrilled to share that, uh, you know, Cruso made an acquisition today. Uh, uh, we Let's go. >> Let's go. >> Get the gong ready. What'd you buy? >> What' you buy?

49:27

>> Cruso acquired a business called Aero. Um, >> sorry.

49:34

Sorry to keep interrupting. I get a little excited.

49:37

Um uh so Cruso Cruso acquired a business called Aero and we're really thrilled to welcome the entire Ato team to Crusoe.

49:46

Um you know Ataro has been a a company that's mostly been operating in stealth um but they uh have been focused on optimizing the very lowlevel pieces of the high performance computing infrastructure stack um particularly around memory optimization.

50:00

So um you know there's a lot of investment being made today uh for these very large clusters these large scale AI factories um that are bringing you know intelligence to the masses and and embedding intelligence in in the economy.

50:16

Um but uh you know the a lot of that infrastructure is not used nearly as efficient efficiently as it as it could be and uh a lot of that has to do with like getting the data into the right place at the right time.

50:28

Um so that your GPU utilization rate can go way way up.

50:33

Um and you know sort of our commitment in in uh this acquisition is uh you know building both very very large AI large scale AI factories and then building the operating systems um and uh you know the software systems to basically operate those AI factories with incredible degree of efficiency and uh scale.

50:53

So >> amazing I I want to go into uh I I want to go into trends about inference cost and and kind of what's going on in the broader market.

51:02

There was this amazing Google paper today by Jeff Dean about saving water and energy.

51:06

It seems like there's a lot of progress.

51:08

Um, but first let's hear uh from Alon about uh the actual journey.

51:13

When did you start the company?

51:14

Uh what what what made sense about teaming up with Crusoe here? >> Okay.

51:22

We started the company last July.

51:24

So about a year ago now >> with the understanding that inference is going to be the next huge frontier in this space like training was all the rage for the last few years >> people all of the sudden realize that they need to make money off of AI and inference is the only place that you actually make any money building a model. >> Yep.

51:47

What we realized in addition like one of our members of the founding team Beness he came out of open AI he led the infrastructure in open AAI for about 5 years.

52:01

This guy knew what it means to run inference at scale.

52:03

He knew what are the challenges.

52:05

He knew the challenges of scaling.

52:06

He knew the challenges of orchestrating different workloads on the GPU clusters.

52:10

And our realization was that it is not enough to run models at a really high efficiency when you're running a single model on a single GPU.

52:20

But the challenge comes when you actually have a real world workload that breathes and changes size like changes the amount of resources it consumes, changes what models are being served, what work, what type of prompts are being asked of it to do.

52:33

And what we realized is that memory is the main bottleneck for this infrastructure for inference.

52:41

Like first you need to shift the model into a GPU. Models are huge.

52:46

They're about a thousand times larger than standard containerized applications.

52:49

Like just how much time it takes >> to put it into a GPU. >> Sure.

52:54

>> Secondly, user assets are also huge.

52:57

Like typically when you have a user session in a standard like consumer application, you have a few photos, a few text assets.

53:03

Those weigh few kilobytes, megabytes.

53:06

KV cache volumes which are the equivalent of user sessions in inference. Those weigh gigabytes.

53:14

So it's another times like it's it's another increase of three orders of magnitude of how much memory you need into a process.

53:26

So that's what we have solved with our technology.

53:27

We have the ATOM technology, the unified memory layer.

53:29

This allows you to utilize the GPU cluster to its fullest, shift memory assets fluidly and in the highest possible speed throughout the cluster.

53:41

And this allows you enormous performance improvements in everything related to inference.

53:47

>> What layer of abstraction are you operating at?

53:49

There's obviously uh a host of different large language models, some open source, some closed source.

53:58

There's a variety of different chips and and and and you have stuff from Amazon and Google like are you sitting above CUDA in the NVIDIA ecosystem?

54:08

Are you agnostic to any particular piece of the stack?

54:13

Just help me kind of understand when uh a an AI company would pull this particular product off the shelf.

54:23

So we have built throughout the last year few different ways of structuring this offer and this also relates to your question >> but the way I like to look at it I have two two part two two analoges in mind when I think about our tech you can think about us as an in-memory high high throughput cache. >> Mhm.

54:47

So it's it's kind of like a database, a radius for your AI application that gives you the assets as fast as possible into the process.

54:57

So I like to think about it as something that is sort of orthogonal or in parallel with the with the CUDA stack and the VLM.

55:05

So we integrate on the side and allow you to shift in and out assets from the existing stack very rapidly.

55:11

This actually allows us to very easily integrate to everything like we can integrate to VLM but we can integrate to any inference runtime.

55:17

And while we currently only support Nvidia officially or Nvidia hardware officially um we are by no mean shape or form constrained to Nvidia hardware.

55:32

>> Yeah, that makes sense.

55:33

>> That's one one one way I want to look at it.

55:35

The other way that I find more inspirational and kind of like tells about this partnership with Cruso is I like to think about as the sort of native virtualization for this new world.

55:47

>> So it's virtualizing the memory which is the main asset.

55:49

So I think about it as the equivalent of VMware for the wave of AI. >> Mhm. That makes sense.

55:55

Uh, Chase, uh, when I think of Cruso's business, I think of like the big projects that get cinematically filmed in Bloomberg documentaries and then I also think about Cruso Cloud um, and the reports from Dylan Patel over at semi semi analysis around cluster max.

56:12

Um, what does this allow you to say to customers that you couldn't before?

56:18

Is this purely going to help with pricing in some sort of commodity offering?

56:23

Is this a differentiator?

56:25

Is this something that uh will increase reliability and uptime?

56:30

I think a lot of people are maybe underpricing what it takes to actually not just yes, I have a chip and I'm selling it at a cheap price, but if it's offline all the time, people aren't going to be happy.

56:40

Um, so what's the shape of how you're uh how you're sharing this with customers and where you hope this goes from like a sales position? >> Yep.

56:50

Um well, you know, we we feel like a lot of the technology that Alain and his team have have built um at Aero is incredibly complimentary to our infrastructure software stack at Cruso Cloud.

57:04

Um and a lot of this is uh really about accelerating performance and enhancing performance across everything from, you know, h how do you actually get the most out of the GPUs that you're you're paying for?

57:15

you're paying for? Um and and and how do we do that through uh things like uh you know more optimizing or u how do we do that by by things like uh you know investing in the storage layer so that you're you're basically able to

57:33

use multiple different layers of storage very cohesively to you know have your data in the right place at the right time to utilize your GPUs more effectively and so I think from an efficiency standpoint it's a major investment From a reliability standpoint, it's a major investment. And

57:47

And then, you know, leveraging uh the scale that Crusoe has uh and being able to operate that infrastructure um at, you know, gigawatt scale, I think is like a a very important aspect to um you know, this this investment we're making.

58:02

Um you know, when I think about like benchmarks and metrics, um I think like, you know, one of the things that people are are frankly looking at is like dollar per token, right?

58:11

dollar per token, right? And if you you actually you know if you double your utilization rate of your GPU uh simply by you know optimizing how you manage your KV cache how you >> you know manage your uh you know you know h how how the model actually gets managed um >> then uh you can actually drive down the

58:34

cost per token by half right so when we think about it it's like what is the cost of intelligence running on cloud and like the intelligent result um that folks are actually you know coming to us for at the end of the day um and how do we actually drive massive efficiencies for all sorts of different high performance workloads. >> Yeah. It goes from cost per token to the >> Yeah.

58:52

It goes from cost per token to the for the end customer uh in the end profit profit per token would be the >> exact Yeah.

59:00

I want to hear about >> Yeah.

59:02

I mean, if you're if you're if if you're you know, people get very focused on the dollar per GPU hour, but if your GPU isn't stable, it's not reliable, you know, you can't move data into it effectively, you don't actually know how to utilize it effectively for your own, you know, hosted inference workload, um you're not going to be able to create that much value for ourselves.

59:19

So, you know, our goal here is to, you know, invest in the platform and to drive massive performance gains that will ultimately enable our customers to drive more value for themselves by using Cruo Cloud.

59:31

There was a there was a post we covered earlier on the show from an account called Bubble.

59:34

They said AI inference is the new high frequency trading and the companies that serve models and don't realize that will will struggle.

59:42

Is how what's your reaction to that?

59:44

Is that is that what it >> kind of feels like right now internally in terms of just like it feels like speed, performance, cost?

59:52

>> Yeah, you hear all these crazy stories about hypery traders like you know Jane Street maintains Okamel like a special language just for hyperre trading.

1:00:00

They're all they they'll build their own e uh their own like uh backbones if they need to uh just to move data more uh more quickly.

1:00:07

Um it feels like there's a there's potential that we're going in that direction.

1:00:13

>> Uh yeah, you know, speaking as someone who spent 10 years working in the high frequency trading sector, uh I don't know if you guys knew that, but uh I uh >> that's lower that's lower. >> That's great.

1:00:23

>> Uh you know, I I do think that there are elements of this that uh rhyme with kind of what's happening in high frequency trading.

1:00:29

you know, they're they're um you know, we we I view it very much as a a massively expanding pie though as opposed to kind of a fixed pie that you know, the high frequency traders are competing for.

1:00:39

Um and you know, as performance gains increase, utility of you know, this infrastructure will go up and demand will go up with it.

1:00:50

So it has kind of this like reflexivity that you know if we can actually as an industry help drive better performance we can actually grow the pie and make you know make make the impact of what we're doing way way larger.

1:01:03

>> So I think it has like a you know positive force that you know high frequency trading doesn't in that regard.

1:01:08

Um, that being said, you know, I I do see like hosting of, you know, hosting of more uh widely used open source models to be something that will commoditize over the course of time.

1:01:22

And, you know, things like speed, cost, and, you know, a bunch of different tricks to optimize performance are definitely going to be instrumental in um, you know, providing a great market product to the marketplace. >> Yeah.

1:01:34

>> Yeah. I mean on that on that note of like tricks uh do you feel like we're on a Moors law type curve for cost per million tokens where not necessarily that it mirrors Moore's law in in getting half the price every 18 months

1:01:51

but it just that it's somewhat predictable versus the progress on the usefulness of LLMs felt extremely spiky and you get these crazy branches in the tech tree and research paths that just all of a sudden produce a ton of value and it kind of goes from zero to 100. Um, but on the cost side, do you think

1:02:09

Um, but on the cost side, do you think that we're going to be chipping away at this, making solid gains on a predictable cadence or do you think there there might be like one weird trick that that uh, you know, every AI lab will love?

1:02:25

>> Um, look, I think it's both uh is the answer.

1:02:28

answer. uh you know we're definitely like riding the you know Nvidia roadmap curve to like you know better more performant you know chips that um we're able to drive down our you know dollar per flop uh by just kind of you know continuing to grow and build on you know a lot of the innovations that are taking place in the silicon companies but

1:02:49

certainly there's a lot of low-level software optimizations in ter and as Alain like frankly put it you know I think there's this massive opportunity to kind of have this you know VMware for for for memory where you know that ultimately is this critical piece of the stack to to produce valuable results from these chips. >> Amazing. Well, congrats on the uh >> Amazing.

1:03:09

Well, congrats on the uh announcement.

1:03:11

Thanks so much for hopping on.

1:03:13

Uh enjoy the rest of your trip and uh safe travels. We'll talk to you soon. >> Appreciate it, guys. >> Great chatting, guys. >> Goodbye. >> Cheers.

1:03:20

>> Let me tell you about graphite.

1:03:20

dev, the AI developer platform.

1:03:22

The AI developer productivity platform.

1:03:25

from Graphite helps teams on GitHub ship higher quality software faster. Get started for free.

1:03:31

Uh I I'm very much enjoying the soundboard today. Missed dearly on Monday. It's back in action.

1:03:36

Uh we have >> Think about a new The thing about a new sound effect, too.

1:03:40

It's like it's like discovering a new favorite song.

1:03:41

You play it like a hundred times and you got to find a new favorite song because it doesn't hit quite the same >> once did.

1:03:50

But for now, we're enjoying >> We We We have an alert noise.

1:03:52

I'm pulling for that to come out during this next interview with Phil from Draq. He is in the studio.

1:03:59

He's in the reream waiting room.

1:04:00

Now he's in the TBPN Ultradome. Welcome.

1:04:02

I like that you're tenting your fingers in in anticipation of a big announcement.

1:04:08

Uh what you got for us, Phil?

1:04:10

What's new in your world?

1:04:13

>> Gentlemen, it's a pleasure to be on. Good to see you all.

1:04:14

Um I'm uh calling in live uh currently forward deployed on site uh with the customer.

1:04:19

No, we are uh, you know, small burgeoning Defense Prime in Costa Mesa, you know, having a great time over here as you can see from the the phone booth that I'm in. Great lighting.

1:04:29

Great lighting over here. Fantastic.

1:04:31

>> Uh, we've uh we have some pretty exciting stuff that we're talking about today.

1:04:34

Um, two big announcements.

1:04:34

One of them is a spectacular technology partnership with our favorite great giant Semens.

1:04:41

Uh, so that means a team center integration for the uninitiated, but it means that I got to hit the for Seammens. Let's go. We lost for a while.

1:04:54

>> We won't take the team slander.

1:04:56

>> Semens overnight success. >> Totally. >> What else? >> We're helping. We're helping.

1:05:00

So, uh, with the with the seams partnership, this means they're like bringing us into their customers. Um, it's been great.

1:05:06

They walked us in some pretty great automotive giants lately and some defensive companies.

1:05:09

So, we're rocking and rolling. Great partnership. Love the Seams team.

1:05:11

Secondly, uh, we are announcing our our primary financing brought to us by Founders Fund and CO2.

1:05:19

favorite Trey and Thomas Leont fell over. >> How much? How much? How much? >> How much? >> The number. >> The number. >> Oh, sorry. 10 uh 10. 7 million. 10. 7. >> There we go. >> Yes. >> There we go. >> Thank you, gentlemen. Thank you.

1:05:36

You It's It's a pleasure. It's a pleasure.

1:05:37

I I thought what better place can I uh tell the world than my favorite technology podcast and news network, the TVN villas. Yes. >> I appreciate it.

1:05:46

Um >> we still we still have to do the the live episode from uh the Empire State Building. >> Yeah, we do.

1:05:53

Uh walk us through the the most concrete example of like uh what uh what could be better about automotive manufacturing?

1:06:01

Like are are are there patterns from Tesla that you think could be ported back to the legacy OEMs?

1:06:07

Are there entirely new things?

1:06:10

Are there particularly broken processes in automotive manufacturing that you've discovered?

1:06:16

any anecdotes you can share them anonymously but whatever's interesting.

1:06:20

>> So it's actually kind of interesting the a lot of our insights are not specific to industry.

1:06:24

So it's actually this realization that you end up having a very bad context problem.

1:06:29

So what we're this paradigm that we're trying to push through is what we call contextaware production planning.

1:06:34

This means us basically taking uh you know every piece of information in a manufacturing facility is one node in a very very large and deeply interconnected graph.

1:06:46

And what build of the west is today is basically a point solution for a manufacturing engineer.

1:06:49

One person in that whole process who is helping with the beginning of the production planning process and a good chunk of the production planning process.

1:06:56

And what we find at an automotive facility, at an aerospace and defense facility, at an A like a tractor manufacturer facility is that they have contacts that might live on the shop floor, they have contacts that might live in the design world.

1:07:07

And they often aren't connected.

1:07:09

They often leave the facility once somebody retires with all that tribal knowledge.

1:07:14

And so a lot of what we found is there is there exists this interesting like spectrum of volume versus mix.

1:07:20

So a aerospace and defense manufacturer does everything very manually.

1:07:25

And so that might look like uh very very high mix.

1:07:27

They have a high a lot of variability in their production and very low volume.

1:07:31

So they call that high mix, low volume.

1:07:33

On the other side of the spectrum, you end up having like uh consumer electronics manufacturing.

1:07:38

A little bit before that, you end up having automotive where it's pretty low mix.

1:07:41

So like, you know, they're having like one one line for one type of car, but it's very high volume.

1:07:47

And because of that, you know, because of that predictability, right, you end up having the ability to have a lot of automation.

1:07:52

And a lot of what we want to drive as a company is taking th those two trade-offs of high mix low volume.

1:07:59

So you like have a lot of uh flexibility in your production, but like you can't do a lot of volume really quickly versus the other thing which is like you can turn out a lot of stuff really fast but you can't reconfigure your line a ton.

1:08:10

We want to basically uh swap and democratize a lot of that value.

1:08:12

And so we want to basically allow an automotive manufacturer to have very very high reconfigurability and mass produce with a lot of flexibility.

1:08:21

And then on the auto on the aerospace and defense side, we want to allow them to be able to massproduce that scale while maintaining that flexibility that they have.

1:08:28

And so that's a lot of what we've seen.

1:08:30

That's kind of interesting. Yeah. >> Yeah.

1:08:32

I feel like Tesla's gone like except with the Cybert truck like ve like all the all the cars are kind of like very clearly like, okay, this is the same platform, the Model 3, the Model Y, like there's a ton of shared parts and that's been what's allowed them to scale volume. >> Color of the seats. >> Yeah, exactly.

1:08:47

There's like two colors, three colors.

1:08:50

Uh whereas uh a company like Mercedes that's been in business for so long, they have like everything from like a small sedan to the the the AMG1 which is probably handbuilt.

1:09:00

Um what uh what is uh um what is the biggest like bottleneck to getting more information out of the heads of like the old guard?

1:09:14

like uh is it is it just bringing on a new system that they are forced to use or are there other kind of like pieces of data collection?

1:09:22

like pieces of data collection? I'm thinking about like like you know GitHub cursor cursor for bio the electronic note lab the lab notebook just this idea of like if you can have someone use voice notes in medical context all of a sudden you're starting to get that data

1:09:39

into something that can be parsed with uh AI or just software uh and it feels like that particularly if someone says oh today I realize that this tool needs to be lubricated a little bit more in order to get good yield that's staying in their head they're not saying it out loud they're not writing getting it down anywhere. Is there is there like is

1:09:55

Is there is there like is there a solution to that?

1:09:57

Is that a real problem or is that just something that people say when they're marketing?

1:10:02

>> A a very very real problem.

1:10:02

Um it's a very real problem.

1:10:04

I I'm not a huge fan of approaches that like take that don't structure that information in a very useful way.

1:10:11

>> Like I I believe that like context is king and structure is key, right?

1:10:13

Like a manufacturing facility can't just be like a dune buggy going off the rails being really crazy.

1:10:20

You need a manufacturing facility to be extremely uh fast, repeatable.

1:10:24

You need to basically have a high-speed rail system as a manufacturing facility.

1:10:28

And so in order to do that, you need to be able to take that tribal knowledge in what might be like in the back of, you know, some dude who is in their 50s and about to retire might be in back of his head a little bit like unstructured, but you need to be able to make sure that you're capturing that information in a really structured and intentional way.

1:10:45

And so a lot of the way that we end up serving as a platform for that is allowing folks to like associate, you know, specific like standard notes and different pieces of context with specific components.

1:10:54

So like that's one way to do it.

1:10:55

But I've seen a lot of really interesting approaches in other places, right?

1:10:58

The key is not just the tooling though. The key is the culture.

1:11:02

key is the culture. And a lot of folks on the manufacturing side are actually like uh unfortunately for the past like 15 20 years some Silicon Valley folks have been just like flying into some facility in Texas or Michigan and saying you know what like here's just some

1:11:17

gobbledygook SAS you know it'll take your manual pen and paper process and it'll still be manual but now it syncs with the cloud like whoa crazy thanks yes um so that the the problem is that like a lot of folks in the manufacturing world like they see a new tool coming in and they're like, "Oh, this is BS." Like, "This is another one

1:11:35

Like, "This is another one of those."

1:11:36

And so, >> I've seen this.

1:11:39

>> And so, what you have to do super intelligence. >> Exactly. Exactly.

1:11:41

And so, it's uh what you end up having to do is build trust and love and affection from like the guys on the shop floor.

1:11:48

We've been very very fortunate to be able to do that.

1:11:51

Like we have really great relationships with the folks who actually like doing the work.

1:11:54

And so if you were a company trying to change the way the west builds, you can't just go top down, you know, go golfing with the C, you know, with the CO or the CEO and like force the top down adoption.

1:12:04

You need buyin from the guys doing the work.

1:12:06

Otherwise, they're not going to use the thing.

1:12:09

Tools going to sit on the shelf collecting dust. >> Yeah. And that's very common.

1:12:11

I mean, that's like the that actually is like the adoption pattern for like cursor.

1:12:16

You know, it doesn't come as a as an enterprise sale mandate.

1:12:17

Uh it comes by just one developer picking it up and then it grows internally and you have your internal champion.

1:12:23

Um, well, I'm very glad everything's going well.

1:12:26

Congrats on the new round. So bullish. We'll talk to you soon. >> You're the man.

1:12:30

>> Showing off the guts for us. Thanks, man. >> Yeah.

1:12:32

Can we get a quick Can we get a quick uh >> There we go. You know what I want?

1:12:37

I want I want your take on those shirts that allow you to bench press more.

1:12:42

Are you familiar with these shirts? The compression shirts? >> Yeah. Shots. I think it's cheating. It's cheating. It's It's awful. It's awful. >> Yeah.

1:12:49

But what if it What if it gets you comfortable and gets you to that next level? >> Yeah.

1:12:52

What if what if you're not ready to bench 315?

1:12:54

This is the thing that gets you out of the out of the valley of uh despair.

1:12:58

>> If you're using a if you're using a slingshot shirt for if you're if you're using a slingshot shirt for 315, you got other problems.

1:13:04

Like somebody like >> like that's crazy. That's crazy.

1:13:08

Just take some creatine, get some protein in, take like a healthy a healthy helping of pre-workout and rip it. Like you'll you got it.

1:13:15

You don't need a slingshot. >> You heard it here.

1:13:18

>> Uh thank you for stopping by.

1:13:18

We'll talk to you soon, fellas. A good one. >> You're the man. Cheers.

1:13:20

If you're looking to design parts, obviously go over to DRA.

1:13:24

If you're looking to design anything else, go to figma. com.

1:13:26

Make anything possible all in Figma.

1:13:28

Figma lets you turn big ideas into real products.

1:13:33

Brainstorm, design, and build with your team at figma. com.

1:13:37

And we have our next guest in the studio. >> Looking sharp. >> Looking sharp. How are you doing?

1:13:44

>> We caught up yesterday, said he was going to come prepared, but you're you're blowing us away here. >> You look fantastic.

1:13:48

Uh, give us an introduction.

1:13:50

Give it explain the company.

1:13:52

Give us any news that you got to share with us. >> Of course. Uh my name is Garam.

1:13:56

I'm one of the co-founders of FAL.

1:13:58

Uh FAL is a generative media platform for developers.

1:14:04

We provide fast readytouse APIs for image, video and audio models.

1:14:12

And then application developers in enterprises like Shopify, Canva, Adobe, they they use these APIs to build AI enabled products in their in their own product.

1:14:24

And then I do have some news as well. >> Some news.

1:14:29

>> So two weeks ago, we announced our CC announcement.

1:14:32

to be raised $125 million.

1:14:40

>> It's not announced until it's announced on TV.

1:14:42

>> We just broke the news.

1:14:43

>> We just broke the news.

1:14:46

>> I have other news as well. So, this is >> great.

1:14:50

What else you got for us?

1:14:50

Hit that soundboard, Jordy.

1:14:52

>> And and this month, >> more news coming in. >> Okay, give it to us.

1:14:55

>> Okay, give it to us. cross 100 million run ra >> we have 10 more days but yet we still we still crossed it so >> that's crazy >> it's been an incredible summer for us and the whole industry to be honest yeah >> congratulations

1:15:11

>> that is absolutely insane >> amazing amazing uh what what uh what's the key to the high growth uh what marketing channels are working uh do you have are you are you buying an incredible amount of steak dinners is there some bottom up adoption is self-s serve option. >> Uh are you are you secretly brothers

1:15:27

>> Uh are you are you secretly brothers with a hyperscaler CEO or something? What what's going on?

1:15:31

How you how you selling this thing?

1:15:33

>> Yeah, I would say all of the all of them.

1:15:35

>> You are brothers with Tim Cook. Interesting. Except that one. Uh that's great.

1:15:39

Uh >> but yeah, >> we are taking enterprise sales really, really seriously.

1:15:46

We work with some of the biggest enterprises, but also like our developer experience, the way to get started is so easy.

1:15:53

We are getting a ton of like indie developers.

1:15:56

Level zio for example is is a big customer and you know >> amazing.

1:16:02

>> So we are getting a lot of enterprises and indie developers and indie developers they go work at bigger companies and they bring fun with them.

1:16:10

So both of the sides of the goto market strategy is working.

1:16:14

>> So >> what trends what trends are you seeing on the uh on the cost side for your customer?

1:16:20

I'm sure people are saying, "Oh, yeah, you're making this much money in revenue.

1:16:24

What does the gross margin look like?"

1:16:26

We saw amazing news from Google today that they dropped their inference cost by 33x. >> Yeah.

1:16:32

And the other thing on the media side with with with textbased prompting, it's it you can use a a non-frontier model and get a great result if you're just using it in sort of like a Google search function.

1:16:43

I would consider last generation video generation to not be on brand for a lot of use cases.

1:16:50

>> Um, so you got >> so we we see you're absolutely right.

1:16:54

Some people they want to use the best newest model and they're willing to pay whatever it takes for that.

1:17:01

>> But we we make more money actually on like the cost effective models, people who are who want to use it for higher volume but the model is good enough.

1:17:08

So that that's a very interesting sweet spot that we actually make more revenue.

1:17:16

But you're right for for a certain type of customer, they they're always looking for the best and the newest.

1:17:20

Maybe they are trying something.

1:17:23

Maybe they just want to get it out the door so the the cost is not a problem for them.

1:17:28

>> Are you seeing opportunities on distilling models for specific use cases?

1:17:33

I could imagine um like a an image or video generation model that's just good at really really good at logo animations or text or just product photography or just beverages and then that being just cheaper to inference. Is that a crazy idea?

1:17:47

Is there something there? >> No, I don't think so.

1:17:50

I I wouldn't say it's cheaper to inference, but it's maybe better to inference.

1:17:54

So for image models, fine-tuning is actually a really really big use case.

1:17:59

not enough people pay attention to it because I don't think it works very well for LLMs but for image model for the past year.

1:18:05

Um we've done like really good work on fine-tuning research and you know you can basically fine-tune on a product and you now can generate better images of that product.

1:18:17

It's actually a really big use case for us and same same thing is going to happen for video.

1:18:22

Uh we we are just getting open-source video models that are fine-tunable.

1:18:27

Uh so a lot of companies are working on post training strategies to create different camera camera angles, different styles, you know, different angles of a certain product or a person.

1:18:39

Uh so I expect a lot of developments on the video fine-tuning side going forward as well.

1:18:47

>> What are you seeing out of ch what are what are the best uh image video uh audio models coming out of China on the open source side?

1:18:55

Yeah, video hasn't really had a deepseek moment, have they? >> I wouldn't say so.

1:19:00

Like there is the best open source model is is video model is Chinese right now.

1:19:05

It's it's coming out of Alibaba.

1:19:07

It just came out couple weeks ago and people started using it right away.

1:19:13

I wouldn't call it the deepseek moment.

1:19:15

I don't think the deepseek moment for the whole industry hasn't happened yet.

1:19:18

hasn't happened yet. um you know we haven't really hit that inflection point that video generation or image generation is mainstream it's it's hidden inside products if you go to Canva Adobe all these products it's still not the main feature it's like an additional thing we still have some time to hit fully mainstream but yeah we have great models coming out of China and we have

1:19:43

>> and also just for I mean this is the classic story of enterprise software like there are open source databases And there are still companies that build amazing businesses on top of them providing enterprise tooling and and support and infrastructure and all the different pieces to actually make a product useful instead of saying, "Okay, yeah, let's go provision some servers to do this ourselves." Uh, fascinating. Um, Uh, fascinating.

1:20:04

Um, George, do you have another question right now? I I >> No, for now.

1:20:09

I'm just blown away you guys are cooking.

1:20:12

>> Yeah, I I do I do want to talk about uh hybrid deals or hybrid models.

1:20:14

I somebody had me demo a AI image generator that uh I I had it take a picture of me turn me into a professional bodybuilder.

1:20:26

It was the most remarkable thing because typically when you go to CHP and have that do that it kind of gets the face a little bit wrong but very clearly and I could tell this from the grain pattern.

1:20:35

What it had done is it had cut out my face and then only operated on the body.

1:20:41

And so the face was a pixelto pixel perfect representation.

1:20:47

And for a certain thing, oh, you just want to change the logo on the shirt.

1:20:50

You don't want to change anything else about the shirt.

1:20:52

It seemed like a good value.

1:20:53

Now, I was being gaslit by the founder who's like, "No, we're not doing that.

1:20:57

It's all in one network."

1:20:58

And I was like, I don't care. I just want good images. It's cool.

1:20:59

Um but but where do you see the different like value added like maybe less sexy less pure endto-end AI but value creative opportunities in in image and video generation.

1:21:12

Um I could imagine like a multi-layer diffusion model that has a specific a specific layer just to add text on top of images.

1:21:21

So you're not trying to confuse those.

1:21:23

I don't know if that's actually how these models work, but what do you see as as valuable in terms of like merging multiple models to create a better pro end product like the way you know 03 Pro has a web browser, a Python shell, and then also the ability to search the web and and pull stuff together and reason. >> Yeah.

1:21:43

I mean one of the reasons why uh you know our business has higher margins than some LLM insurance providers is because of this this fragmentation of different kinds of models different kinds of use cases.

1:21:57

People are trying to chain these models together and create workflows.

1:22:02

Um both image and video models they haven't really generalized yet.

1:22:07

So you can't write a long prompt and expect uh you know the the image and video to edit itself.

1:22:13

It's still a lot of model chaining and workflow generation.

1:22:18

So like the use case you you just said like putting your face onto something else.

1:22:23

It's like chaining couple models and we have a lot of uh that behavior on the platform.

1:22:27

The models are getting bigger and better.

1:22:29

So some of that is actually happening you know by by the model itself but but still unlike LM where it's like a single prompt and you expect it to you know write a book but also generate code and like generalizes very well it doesn't it doesn't happen for image on video yet.

1:22:49

>> Yeah I can imagine that there's tons of interesting enterprise like features and harnesses on top of the models that will be extremely value creative.

1:22:56

So, seems like you're in the good in a good business.

1:23:00

But, uh, everyone already knows that from the revenue and the fundraising.

1:23:03

So, congrats on all the progress, Jordy. Anything else?

1:23:06

>> We'll have you back on again soon.

1:23:07

>> We'd love to have you back. This is fantastic. We'll talk to you soon. Legend. >> Bye.

1:23:11

>> Let me tell you about Julius, the AI data analyst that works for you.

1:23:13

Connect your data, ask questions in plain English, and get insights in seconds. No coding required. >> Sorry, Tyler. You're out of the job.

1:23:22

I'm going to be using Julius. Try Julius for free.

1:23:29

>> Uh they're loved by over two million users and trusted by individuals at Princeton.

1:23:32

>> And you can connect Google your Google Ads account to Julius now.

1:23:36

>> Run some numbers there. >> Love that. Love that.

1:23:37

Anyway, we have our next guest in the reream waiting room.

1:23:41

He's here now in the TBPN Ultradome. Welcome to the stream. How are you doing?

1:23:46

>> I am doing well, Jordi.

1:23:46

And John, thank you so much for having me here. >> Thanks for joining.

1:23:50

Would you mind kicking us off with an introduction on yourself, your company, any any news you got for us? >> My name is Sudish.

1:23:56

I'm the CEO of Tiny Fish.

1:23:58

First of all, I'm very happy to be here.

1:24:00

First time I got to know you was saw Parik by the way. That was a great guest. >> That's amazing. That was a great guy.

1:24:06

What a wild day on the internet.

1:24:06

Uh >> and how fast it's all gone too though. I mean it's all gone. So >> yeah. Yeah.

1:24:12

I mean people were we never got the full postmortem. We'll need to check in.

1:24:16

But he said he was going exclusive.

1:24:18

There's somebody named Sohan Periq.

1:24:20

Exact same spelling that's been appearing on AI research papers. >> Really?

1:24:25

>> People are trying to figure out is it the same guy or just >> I saw that too.

1:24:28

Yeah, I think DBP needs to have an investigative arms.

1:24:32

>> Yeah, I actually did talk to someone that was thinking about hiring him after the fact and it seemed like it was not uh not enough of a red flag and they said they were going to give it a shot and hope that he >> I thought he was going exclusive. >> Yeah. Yeah. Yeah. Yeah.

1:24:45

And so and so as far as I know he is still exclusive but I haven't done any research recently.

1:24:51

>> Anyway guys thank you so much well you both have built something in the last 10 months entrepreneurs investors. We appreciate it. >> Thank you. Thank you >> Tiny Fish.

1:25:00

We just announced our 47 million funding series which >> I was not ready for that.

1:25:06

I was not ready for that. >> Absolutely. >> Needed a big number.

1:25:10

I needed a big number warning >> board. Absolutely huge. >> Congratulations.

1:25:14

7 million is like a seed round, right? >> Mango seed.

1:25:20

>> It's a seed round for a tiny fish these days. >> Tiny fish. But congratulations.

1:25:25

>> I'm sure you can do a lot with it.

1:25:25

So what are you going to what are you going to do with it?

1:25:29

>> Look, I think we are going to invent a category, build a category that makes sense and build a great company. That's basically it.

1:25:34

Prior to this uh you know I was at uh thought spot and nutanics.

1:25:40

We have done some large enterprise companies >> co-founders.

1:25:42

They all came from Meta and companies like large.

1:25:44

We want to build a large company doing one thing which is to make web browsing like humans do but a superhuman scale for solving enterprise problems.

1:25:55

>> We don't want to be a consumer company.

1:25:57

I know you both have built companies.

1:25:59

There are a lot of AI companies doing things in in in consumer space browser agents and all of that.

1:26:03

We want to do uh a massive scale web browsing and turn the internet into a structured database for enterprise customers to analyze and act on.

1:26:15

>> Do you want to sell this to other software companies that are going to piece it together with other uh APIs to create maybe even more enterprise SAS?

1:26:26

Like how deep do you see yourself in the stack?

1:26:28

Look, I think nowadays there is no difference between enterprise business and developers/builders because every enterprise product in the AI world starts as a a builder product, somebody trying it out.

1:26:38

We 100% want to be the best way anyone to connect to connect code to internet because look these models can solve math problems at international math Olympics level but when it comes to browsing it's a mess because browsing is so complex that our brains make it look easy.

1:26:54

Yeah, >> like think about going to a movie.

1:26:56

The the seat that you pick is a function of complex reasoning.

1:27:00

If I'm going by myself, I can sit in front and read the movie.

1:27:04

But if I'm far away with my family, I want to make the right place, right, you know, contiguous seats.

1:27:10

>> These decisions we make without even thinking.

1:27:11

Not to mention the complexity of the website itself, how the pictures are popping, how there's like a pop-up here, these things need to be solved really well. We do that.

1:27:19

So enterprise customers who want to browse mass 60% of all time of knowledge work in enterprise is on browsing. Yeah.

1:27:29

>> And we think we can automate that through what we are calling enterprise web agents.

1:27:31

So, we we heard reports that uh some of the big labs were building RL environments with verified rewards, basically cloning Door Dash, creating clones of Amazon.

1:27:42

com so that they could train an agent on the full Amazon.

1:27:45

com web app and then an agent would be able to check out for you.

1:27:50

Are you doing something similar?

1:27:52

What does data collection look like?

1:27:54

What does training look like?

1:27:55

What does differentiation look like in the world where uh where these agents uh might need to be rldled on a specific website to really get good results?

1:28:07

>> This is why I love talking to you guys.

1:28:08

You know, it's it's you geeks ask the right question to geeks.

1:28:11

So there are two models when it comes to uh these.

1:28:16

One is synthetic websites like web arenas and others have built this.

1:28:18

The other uh is uh paying money to say let us go learn Door Dash workflows.

1:28:22

Both are good but both have downsides because both really create significant challenges with respect to scale when they come out of that environment.

1:28:33

>> What we are doing is something very we have a proprietary architecture that we have built that frees us and takes a snapshot of a data website in a way that becomes that exhaust feeds ourselves in a very scalable way.

1:28:45

a very scalable way. So yes it is synthetic data that we are generating but it is out of actual web and more importantly most of our use cases we launched our company with a use case with Google hotel as a customer dash is a customer >> great great I try to give this

1:29:00

>> I try to give this advice to to early stage founders if you if you're going to get a customer early just get Google >> start with a mag >> start with a mag 7 why not right >> they built this AI technology I mean it all started with all you need is attention Yeah. >> So these customers are giving us access.

1:29:16

>> So these customers are giving us access.

1:29:19

There's a customer who has 22 million uh web SKs in the product.

1:29:22

There's another customer the 40,000 uh subshops as part of them.

1:29:28

Class pass we announced them right.

1:29:30

So these sort of things while we are solving for them is giving us a lot of data and that exhaust is mined by our products for uh annotation and creating labeling that makes art.

1:29:39

The second thing I will tell you though models are good at a lot of things but they usually perform really well when the goal is straight and the reinforcement can be built on something where the context doesn't change.

1:29:50

Web is exactly the opposite.

1:29:53

Everything is context that is constantly changing.

1:29:55

The goals are changing the websites are changing and the context is changing.

1:29:58

So what we have done is we use reinforce learning and models for understanding and like sort of see what the heck the we are trying to do here.

1:30:07

Then we separate that and execute that code in CPU at massive scale.

1:30:12

And then we use alpha evolve a version of that to make the code better.

1:30:16

>> So separating exploration and execution is another fundamental thing that we've invented that is going to make this >> browser agents you know perplexity comet and operator and others that is like forcing and feeding us.

1:30:27

But this is where the execution framework.

1:30:29

So that's an infrastructure problem.

1:30:31

Think about uh the amount of browsers we have to spin up, the fingerprinting, the anti-bot, all of that infrastructure need to be solved. >> Interesting.

1:30:41

Um how do you think uh the competitive landscape will evolve?

1:30:46

Amazon's launching a bunch of stuff.

1:30:46

Uh I'm sure all the clouds have something that looks like a browser agent on the shelf somewhere.

1:30:56

But um we've seen you know Silicon Valley has you know probably hundreds of stories about uh companies that have built things that have been competitive with the big hyperscalers and been wildly successful.

1:31:10

But what's your specific plan?

1:31:12

>> You know there is a quote that uh often attributed to Chasty Puller Marine that we are surrounded from the north, southeast, west better watch out.

1:31:22

This is one of the space where it is so amazing because there are three groups of companies that we will be competing with from a tech space.

1:31:29

I mean obviously there is sharehing of dollars.

1:31:31

One are the browser agents who are doing one thing like jod going on a vacation 20 minutes he'll book a hotel for you.

1:31:38

We want to sell to xedia not to jodi because xedia now want to simulate 100,000 jodi transaction from 50 states or 50 countries. How do we do that?

1:31:46

So browser agents to a certain extent will be trying to expand to that space.

1:31:51

I think we want to compete.

1:31:53

The other is the infrastructure.

1:31:54

So the browser base and others who have been building things we want to build something enterprise grade with the surveillance you know the the governance the control. So that's the second.

1:32:04

the second. Third is going to be another interesting area where the codegen and the developers who builders who just want to connect their code to internet so it can do amazing things there again there are parallel you guys interviewed parag

1:32:18

>> I do think that we'll compete there on some level not because today we compete but all of these will converge because there will be a few massive companies built on how browser access how internet can be turned into a structured database so exciting time. We've raised enough

1:32:34

We've raised enough only because we want to make sure that we build it in a way that is where we can control our destiny.

1:32:40

I have a lot of respect for all of these companies, but to me, you know, if you're not surrounded by competitors, you're not in the right place. >> I love it. >> Yep.

1:32:48

You're not going after a big prize, but congratulations. Fantastic answer. Good luck out there.

1:32:52

Have a good rest of your day. We'll talk to you soon. >> I hope so.

1:32:56

This was such a short conversation. I enjoyed. I'll keep watching. Thank you so much. >> We'll talk soon. Talk soon. >> Bye. >> Bye.

1:33:02

Let me tell you about Adio customer relationship magic.

1:33:04

Adio is the AI native CRM that builds, scales, and grows your company to the next level.

1:33:09

We have our next guest already in the reream waiting room.

1:33:13

We will bring her in now. How are you? >> Welcome. >> Hi. Nice to be here. >> Great.

1:33:21

>> Great to meet you, Julie. What's happening? Great to >> Yeah.

1:33:23

Would you mind kicking us off with a little bit of an introduction on yourself, uh, your company and maybe some of your background as well? >> Sure.

1:33:30

So I am one of the founders of Sundial.

1:33:32

We are trying to build the cursor for data analysis.

1:33:34

So we're trying to automate a lot of what data scientists will do um so that we can help companies make better decisions with data.

1:33:43

We work with great companies like OpenAI character captions.

1:33:45

Before that I uh led design and research for the Facebook product.

1:33:53

I grew up at Meta back then on Facebook.

1:33:56

Uh could you explain for the audience what Facebook is?

1:34:00

They've been living under a data center. No, I'm kidding. Sorry.

1:34:04

>> So, you know, if if you if you guys become friends and your relationship gets to a certain level and you want to, you know, let the world know can >> amazing broadcast. Amazing.

1:34:15

Well, I think it it feels like the absolute perfect background to build Sundial because when I think when when I think people think about optimizing digital products and making decisions with massive amounts of data, people think Meta and they think Facebook, right? >> Totally. >> Yeah.

1:34:34

>> I mean, you say that because you probably know more than most people, but I do think what makes Meta so good is that it was so freaking operationally rigorous.

1:34:42

It was so good at actually capturing and modeling what made it a great business.

1:34:47

Um, and and being like, you know, just so on top of every little thing so that we could optimize, iterate, experiment, test, and eventually, you know, get to that scale. >> Yeah.

1:34:58

>> And and did you guys, I imagine, had to rely on a bunch of internal tooling that you built in order to do that that was proprietary or or what did that look like? >> Absolutely. Yeah.

1:35:07

Everything was just homegrown because like actually the whole concept I would say that Meta probably introduced this idea of growth.

1:35:13

I mean now everybody's got growth teams but back then that was like a new thing and it was only created you know with people like Chimath and Alex Schultz and Naomi because >> Facebook was growing like crazy and then one day growth kind of stalled and you know every I think this happens to every successful company right in the beginning you're like we're we're we're

1:35:33

great our intuition is on we're solid like everything hits and then one day it stops hitting and then everyone's like oh my god what happened and there's a lot of questions you know board board members are calling and then it becomes like well we have to figure it out and the only way you can do that is with >> good observability. >> Yeah. Can you take us through like a >> Yeah.

1:35:51

Can you take us through like a concrete example of uh like one of the like a case study of like design at meta and and and how um like some of the decision-m linked to some of the data analysis you did?

1:36:02

Like do you have anything that you can share?

1:36:04

I I know it's been a while but some of the stuff's probably private but some of it you can probably share. >> Yeah.

1:36:10

Um I'm trying to figure out um what's an example I can pull out.

1:36:12

Well, I will say that like a lot of the process so I I think we kind of segment things into two worlds, right?

1:36:17

The first is like look in order to build something new, you do need that spark of inspiration and it usually comes from intuition.

1:36:24

Like you need to kind of have like a vision >> and so often that would be like the spark of the idea.

1:36:28

spark of the idea. But then as soon as you realize that you have like a hypothesis you know like I think what people want is you know better ways like if we give people more custom ways of sharing on Facebook for example they will share more that's an example of

1:36:46

like a particular hypothesis that somebody would have and usually that comes with like an idea you're like if I build you know a really awesome text editor or I give them a way to record video or you know I make it super easy to like pull in a quote Right? Then

1:36:57

Then people will go and share more.

1:37:00

But the thing that you then need to do is to figure out well how quickly can you test whether that hypothesis is true?

1:37:06

Like what's the what's the kind of smallest thing that you can build that gives you either more or less assurance that this thing is is the thing to bet on.

1:37:15

And so we would try and figure out what's like the smallest framing of that experiment.

1:37:21

Um and then if it seems like you know okay it's inclusive or we actually think it might be promising then you just double down right then you like add more you know you keep building and I think that um that that's been like you know very this is like a super early example

1:37:37

but I think like hackathon project like video was like a hackathon project that somebody did um and then very early on like video came out after the hackathon and like just very quickly you can see um what what will you know like that that people are actually using. In fact,

1:37:51

In fact, maybe a better example now that I'm thinking about it is actually feed.

1:37:54

So, feed is one of those very early examples where people hated it. So, newsfeed came out.

1:38:02

Before that, the way that people went to profiles is like, you know, I would go to your profile, John.

1:38:06

I would go to your profile, Jordy.

1:38:08

I'd see if anything changed.

1:38:09

But there was no system that automatically told me if you updated something.

1:38:13

And so we in launch feed, this was like back in 2006, and there were like huge protests.

1:38:18

In fact, like a million people within a week joined this Facebook group called I effing hate Facebook newsfeed.

1:38:25

Um because people thought it was like it was creepy.

1:38:29

It was like, you know, um too much surveillance.

1:38:31

But the reason I think Mark and the team like we kept on it is the fact that like through the numbers like through the data you could see that people were spending a lot of time on newsfeed and in fact the reason they knew to join this group I effing hate newsfeed is because of newsfeed like it was they were like you know so it just was like very obvious to us that this was working despite what people were saying about it. >> Yeah.

1:38:55

Uh can you talk about preference falsification more broadly?

1:38:57

Was that like a hard one lesson that carried through?

1:39:01

It feels like Meta has really um like absorbed that lesson and it continues to inform the company culture.

1:39:10

Have you carried that idea of preference falsification forward into your career?

1:39:15

Uh how do you think about that disconnect between what people say and then what they actually do?

1:39:20

>> I think it just like you have to gather the data.

1:39:23

the data. You just have to look at you know like you can say okay look this is what I see I see it like data their data is that people tell us they don't like it or they tell us they're feeling one way or another right so I think about qualitative feedback as data customer

1:39:38

surveys are you know or data like anything is data but also behavior like what do people actually do what do they click on where are they spending their time and attention and the more we can have all of that in one place the better like a a better more realistic and accurate view of reality we have. >> When should a company sign up for

1:39:54

>> When should a company sign up for Sundial?

1:39:57

Because I I I think there's a there's it's almost a meme in Silicon Valley.

1:40:02

If you if you launch a product early and you're obsessing maybe too much over the data, you could you could go end up actually going down the wrong path and and there's points in which uh maybe you disagree with this where the actual just like founders intuition is what what's going to get you to that um sort of uh takeoff with a product. But what's your read? I agree with that.

1:40:20

I do think that it's there's such a thing as like obsessing too early.

1:40:24

Like when you don't have product market fit, you probably have to take bigger swings.

1:40:29

You're probably not in the business of trying to optimize.

1:40:31

But I think if you feel like you have product market fit, you're clearly scaling your team.

1:40:35

Like you're like, "Okay, I probably need a growth person. I probably need a data."

1:40:39

Like the reason you want a growth or data person is you want to form more fuel on the fire.

1:40:43

Like you want to understand what works and what doesn't.

1:40:47

And so what all of these roles are trying to do is start a series of experimentation and just like again get to better uh better visibility, right?

1:40:55

Better observability for like what we all want to do is just build a mental model like this is how the business works and if you tweak this thing more money goes up and you tweak this more users, right?

1:41:05

We want to build a very real and um influencable model and that is what like all of that work of trying to build uh and and and do analysis is is there to help us do.

1:41:15

Did you overlap with Laura Dana at uh Meta?

1:41:17

Can you tell me a little bit about her? Well, yeah.

1:41:22

What everyone she was in the Wall Street Journal earlier this week.

1:41:23

Um what what what's made her uh you know what's the secret to her impact?

1:41:31

>> So Laura Dada joined as um a design lead and I think she took over a messenger.

1:41:37

So, Facebook Messenger and I think the team was pretty small then, maybe about eight designers and she helped scale that up to like 200 designers and eventually she actually become the product group lead.

1:41:48

So, the product group lead is kind of like the GM role and in fact it's there's not that many examples of design leaders who become GMs, you know, like engineers can become GMs like um I think Adam Oseri, head of Instagram is one.

1:41:59

I think Lordana is probably the other that I can recall.

1:42:03

Um, so she ended up uh actually leading all of Messenger.

1:42:05

I think recently she went to go work on AI and I just saw the news that she is joining as Figma's new uh chief design officer, which is fantastic news.

1:42:14

I absolutely adore Laura. I think she is fabulous.

1:42:18

>> Yeah, we'll have to get her on the show.

1:42:20

That would be uh it'd be it'd be fun to dig into.

1:42:21

Um anyway, Jordy, anything else?

1:42:24

>> I have a bunch more questions, but I don't think we have time.

1:42:25

Yeah, >> um and uh yeah, congrats on on all the progress so far. >> This is fantastic. We'll talk to you soon.

1:42:34

>> All right, talk to you guys soon. Bye. >> Bye. >> Cheers.

1:42:36

>> Gabe in the chat says, "Whoa, she has a headset on.

1:42:38

How can I long this company?" Totally agree. >> That's great.

1:42:41

Uh, let me tell you about Numeral HQ. Sales tax on autopilot.

1:42:45

Spend less than five minutes per month on sales tax compliance.

1:42:47

You can get started for free. Benchmark series A. Let's go. >> Let's go.

1:42:54

>> Um, our next guest is in the reream waiting room.

1:42:57

We will bring them into the PBPN Ultra Dome.

1:43:02

>> Jacob, Jacob, welcome to the show.

1:43:04

Finally waiting for this moment. >> It took us very long. >> We're pretty good. How are you guys? >> We're doing great.

1:43:12

>> Give us the breakdown. What'd you launch today?

1:43:16

>> Cohort 3 applications are live. >> Boom. Hit it.

1:43:20

>> So, the the the best part here is the video.

1:43:22

I I assume you guys may have watched it, but it's I think the best Gundo hype video made.

1:43:25

Wait, can we pull Can we pull it up? That's a tall order. Can we pull it up? Yeah.

1:43:29

Production team, Ben, can you pull up uh the the latest uh discip discipulous ventures post?

1:43:36

Maybe we can put it in the chat.

1:43:39

Um can you send it to them?

1:43:39

I'm trying to find Jacob's throw it in the chat.

1:43:44

>> We we we will pull this up.

1:43:44

We might need to watch this after you uh hop off, but but give us the update on on the on the previous cohorts. Who's in those?

1:43:53

What companies are you in?

1:43:53

What's the general thesis?

1:43:54

What like what's the why now for this particular uh organization?

1:44:00

Break me break it down for me. >> Yeah.

1:44:02

So, cohort one was in Brain Maker and Valor at the time, their office. They were sharing it.

1:44:07

They were both preceded stage.

1:44:09

>> I think it was still very much like, is this going to work or is this not going to work?

1:44:13

>> Um, I think now we're at the point like, okay, this something's working here really well.

1:44:15

And I I think the the big kind of thesis and point here is like we want to be like that entry point to what's going on in Elsagundo. Yep.

1:44:23

>> The culture is very attractive to to mostly young people want to build companies that really do matter a lot for the country that have kind of been pushed away from other ecosystems that don't really care about that and then and two practically like if you want to build in hardware or aerospace defense, you should be in Elsagundo. Yeah.

1:44:36

>> SpaceX started down the street for where I'm based.

1:44:38

Um and the best talent the supply chain the whole supply chain um and a lot of the customers are in Elsundo area.

1:44:44

So it's like if you want to build an Elsagundo, if you're building an aerospace, defense and hardware and want to be part of the ecosystem, we want to be that way to do that.

1:44:51

>> Um, companies wise, >> Augusta said Augusta said earlier Rockefeller built Elsa Gundo. I I never knew that.

1:44:57

What What's the lore there? >> The oil company. >> Okay. >> Yeah.

1:45:01

It's Chevron like that that there is a massive oil and gas extraction oil refinery there.

1:45:06

When you're in >> the whole town, the whole town was built out to service. >> Yeah. Yeah.

1:45:10

You you can tell the story.

1:45:12

>> I I'll give the the rundown.

1:45:12

So Elsando means the second.

1:45:13

It means the second Chevron refinery in the state of California.

1:45:16

Um, so that's the reason why.

1:45:19

So it's very it's very cool.

1:45:19

But on the >> Yeah, we got the video if you guys want it. >> Yeah.

1:45:25

Yeah, let's play the video. >> Uh, play it.

1:45:27

>> We're going to we're going to switch over to the video.

1:45:28

This is the fall cohort applications are live now. Yeah. Hit the soundboard.

1:45:33

Why not just add an extra layer on top?

1:45:37

>> Are we going to get taken down for this?

1:45:39

Is this a open source song? Hopefully. We'll see. This is this is good.

1:45:46

>> It is necessary fast >> that every in the United States that are like just by their shoulders.

1:46:02

[Music] That's a cool presentation. Let's go. Holy bass.

1:46:17

[Music] Good sound editing. There's Isaiah. There's Augustus. We got cameos on cameos. >> America. [Music] >> There we go. >> With a with a >> Yes. Yes. The gun back.

1:46:41

We're definitely getting demonetized.

1:46:44

>> Thank you for demonetizing us. It was worth it. >> It was worth it. It was worth it.

1:46:46

Uh, take me to the economics.

1:46:51

>> Now, so so how much money do you raise for each cohort?

1:46:53

How much money are you putting in?

1:46:55

What what's the deal structure?

1:46:57

Like how does uh what do the what do the companies actually get out of going through the cohort? >> Yeah. Yeah.

1:47:03

So I guess just top big picture fund one end up being right around uh the $9 million mark.

1:47:09

We're basically doing five cohorts out of this.

1:47:14

So three next year, one end of year in October. Yep.

1:47:17

>> Um the basic kind of structure each cohort we do 10 to 12 founders.

1:47:19

So in a 100k check um per company will be around 1.

1:47:25

1.2 2 mil for cohort and then that'll go across the rest and a little bit for follow on for kind of proa >> um >> but yeah the kind of standard deal cohort itself is like main thing again you want to be an also a weird way to do that you have all the big name people everybody knows like we can give you access to them just like that and we

1:47:40

think that's very valuable early on um to kind of become like I guess consensus bet in a way um just to have those behind you is very helpful and then we have a demo day we bring like 120 I think we had 120 partners last time from top firms like just like that access to all of them so um that's kind of the breakdown we have house. It's very fun. It's very fun. You have a chef. We have a gym.

1:47:57

Everything you need to to build.

1:47:59

>> Are you Are you officially taking over from Augustus's job as also going to a tour guide?

1:48:04

Because every sing There was like a solid six-month period where every single VC and founder would hit up Augustus and be like, "Okay, time for the tour. Let me go around."

1:48:15

>> He doesn't have time for that.

1:48:16

>> I don't think he has time for it anymore. He's on the road a lot. >> Yeah. Yeah. It's funny. Yeah.

1:48:18

I I'm getting I have too many texts from from BC saying, "Hey, I want to see show me around."

1:48:25

So, I I think I've taken that I I think it's more of a job for me than him. But one >> Cool. cool.

1:48:30

Here's the wire details and then I'll take you on the tour. >> Exactly. Exactly.

1:48:33

But also on that point, we had the most recent bonfire which a couple of you guys were at >> last week, last Friday.

1:48:40

>> Oh, Tyler is like, "Okay, you're taking this over."

1:48:42

Um, and we had like I think close to 200 people come out, which is pretty awesome.

1:48:45

Which is like I think a record I hopefully for one of the bonfires in >> Wow.

1:48:49

venture capital firm absorbing an iconic uh uh networking series.

1:48:51

Uh where have we seen this before? >> It's interesting.

1:48:57

I I'll I'll be like completely honest.

1:49:00

When I saw Dipilus launch >> Yeah.

1:49:03

>> I I wasn't >> I like I've I've backed a bunch of, you know, I've backed Rain Maker, I backed Shink, bunch of bunch of the Gundo companies, but I it didn't immediately click for me.

1:49:12

But hearing you hearing you talk about it, it's just like uh and obviously we're we're super close with uh with Gary and the whole YC team and they've backed a bunch of hard tech companies and so my immediate thought was like do we need something like this?

1:49:25

Do we need a new institution?

1:49:25

And >> hearing you talk about it and seeing, you know, the progress from the port coast so far, it just makes a ton of sense to have like an institution in Elsagundo because we can't just rely on the city to give tour guides to people that want.

1:49:39

A couple years ago, I went I went and did the tour of Elsa Gundo.

1:49:41

Uh couple you helped facilitate the meme.

1:49:46

>> Uh there were a bunch of folks uh that were that were highlighting it.

1:49:49

Uh and it the energy was very similar to like you know Mountain View in 2009 or 2012 in my opinion.

1:49:59

in my opinion. like it had that like university a feel like it's sort of it's just boring enough like you're in LA but it's not like this hot town that you're going to get distracted by like it's really just like a lot of cool companies building and then you walk across the

1:50:15

street there's another founder they have some tool like it actually makes sense to have people sharing stuff and moving stuff around yeah it's very very cool community makes a ton of sense >> congratulations on the launch of the new uh application window uh I haven't run this by John But hopefully he agrees. We'll sponsor the next bon bonfire. So

1:50:31

We'll sponsor the next bon bonfire.

1:50:31

So count to come to the purple orchid as well afterwards.

1:50:36

That's >> I've been to the purple orchid. It's a lot of fun. It's a good spot. >> Awesome.

1:50:41

>> They do the the dragon bowls there that you drink out of. >> Dragon Balls. >> Dragons on fire. They light them on fire. Yeah. Yeah. Yeah. Yeah.

1:50:48

>> Anyway, well, congrats, Jacob.

1:50:50

>> Thanks so much for hopping on.

1:50:50

We will talk to you soon.

1:50:52

Have a great rest of your day. >> Awesome. Thank you, guys.

1:50:55

>> In the meantime, let me tell you about Linear.

1:50:56

Linear is a purpose-built tool for planning and building products.

1:50:58

Meet the system for modern software development.

1:51:01

Streamline issues, projects, and product road maps.

1:51:04

>> Cursor, head over to runway, perplexity, >> retool, mercury, ramp. They all use it.

1:51:11

>> We we have another three hours of content uh to go through.

1:51:14

I like this post from Growing Daniel.

1:51:16

The opposite of FOMO is when you skip an event and then everyone who went says it sucked.

1:51:23

Vindication of missing out.

1:51:23

And Tyler Cohen, Tyler Cowen says, "VOMO." >> Vomo.

1:51:29

>> I think VOMO is a good one. Vindicated. >> It's simply the best. >> It is the best.

1:51:33

>> It simply is the best. >> It is the best.

1:51:34

I'm trying to think of the last time I had Vomo.

1:51:36

I don't want to put anyone on blast, but uh uh I have certainly I've certainly said no to things.

1:51:42

>> Well, let's keep going through the timeline.

1:51:43

Uh John, can you text Gabe? >> Yes.

1:51:47

>> And see if he is still uh coming on. He confirmed one.

1:51:50

We moved him up 10 minutes.

1:51:52

We might need to do some timeline.

1:51:53

>> I'm happy to hang out. Yeah, we can hang out.

1:51:55

So, uh, Sophie Netcap girl says is showing a post saying, "There must be some mistake.

1:52:02

I thought I'd be working on digital super intelligence."

1:52:04

And it is a angry uh man in a military un uh uniform saying, "Build the Goonbot."

1:52:10

And uh so there's a lot of that going on right now.

1:52:15

Um, uh, Luke Glacier has another post, uh, in response to the UK independent space agency is entirely scrapped to cut costs.

1:52:28

And, uh, we have Matthew McConna here saying, "We used to look up at the sky and wonder at our place in the stars.

1:52:34

Now we look down and worry about our place in the dirt."

1:52:40

Um, KPMG wrote a 100page prompt to build a Gentic Tax bot produces advice in a single day instead of two weeks without job losses.

1:52:49

And uh, X Nelio says 100page prompt is crazy.

1:52:54

You ever ripped a 100page prompt, John? >> I've gotten close.

1:52:58

I've used a lot of those like repurposed prompts that people have online.

1:53:02

Oh, this is the super prompt.

1:53:04

Uh, never never got in the habit of like daily driving one of those.

1:53:08

Uh Doresh Patel had a good prompt uh in Gemini that would take a transcript and clean it up and it gave a really like the hundred pages was really like a 100 pages of examples basically like here's what a messy transcript looks like out of a uh out of just like a transcription service where the speakers aren't labeled and there aren't necessarily paragraph breaks.

1:53:29

you give it a ton of extra um you give it a ton of extra context but ton of extra examples and then that makes it perform a lot better because it knows it understands the job. Yeah.

1:53:42

>> Uh but yeah 100page prompt is pretty wild.

1:53:45

>> So CFKA Esquire says uh just one more model bro just one more funding round bro please.

1:53:50

And uh this post from Walter Bloomberg esteemed journalist.

1:53:55

>> Where did this come from?

1:53:55

So this was from some interview. Okay.

1:53:57

And and to be clear, yeah, taken out of context, the the headline was open AICO Sam Alman conceds GPD5 was a misfire bets on GPT6.

1:54:08

And this has been going pretty viral.

1:54:10

People are poking a little fun.

1:54:10

I think I I think uh in some ways, you know, um this the the the Death Star kind of set them up for this.

1:54:20

>> Double Bird Capital says just one more trillion, bro. Honestly, I swear.

1:54:22

I mean, my take on this is that like the numbered releases should just go.

1:54:25

Uh, when Facebook introduced the feed, they they they stopped saying we're changing things.

1:54:32

It's a good take because when you say >> it's a good takes here. >> Yeah.

1:54:36

When you say when you when you tell everyone we're changing everything, people are like, I don't like change.

1:54:41

But if instead you just say like, yeah, like every day Facebook gets a slightly different update and the algorithm is slightly different.

1:54:46

And then some days people are like, ah, the algo is not really for me right now.

1:54:49

and then they complain enough and then you get the feedback and then they tweak and they make it better and better and better.

1:54:55

Like I feel like uh we should just be in the era of GPT is what powers chat.

1:54:57

com and it's constantly getting better and stop focusing on the size of the pre-training bubble.

1:55:05

Stop focusing on the parameters if you're a consumer.

1:55:09

Stop focusing on the numbers.

1:55:09

Start focusing on your sleep number, your sleep score. Go to eightsleep. com.

1:55:16

Exceptional sleep without exception. Fall asleep faster.

1:55:18

Sleep deeper, wake up energized.

1:55:20

You can discover podfatsleep. com.

1:55:24

And I believe we have our next guest, Gabe from Mischief. How you doing, Gabe. >> There he is.

1:55:29

>> Welcome to the stream.

1:55:31

>> You are live at the TBPN Ultradome. What's new with you? Uh how you doing?

1:55:34

I think everyone knows you.

1:55:37

Maybe just launch into uh the the announcement this week about uh the the private military corporation that you're starting.

1:55:47

>> Oh, you saw the Twitter.

1:55:47

I didn't think anybody looked at that. Interesting. >> Everyone saw that. Everyone saw that. >> Yeah. Yeah. It's uh Sure. Sure.

1:55:52

What What do we do this week?

1:55:55

Well, we launched an agency called Applied Mischief. >> Yep.

1:56:02

>> The agency honestly like to to rewind where where does that even come from?

1:56:06

And a lot of people don't actually understand this about mischief because probably what you see is like videos of people wearing boots or like press headlines or whatever.

1:56:14

Uh but the way we're actually organized is we are essentially a holding company of different creative enterprises based on categories.

1:56:24

So there's like a handbag division, there's a footwear division, there's a fine art division.

1:56:28

There are other divisions I can't really talk about yet, but with like more bigger permanent structures.

1:56:33

Um >> well, you talked about a division you can't talk about, so I hit the >> Oh, nice. Nice.

1:56:45

So we we've been operating like that for for a second now.

1:56:47

And to make that efficient, there's been an internal back office that's like normal, right?

1:56:51

Like finance, legal, manufacturing, customer support, but also like design and marketing.

1:56:58

Y >> and that's a group that not only does design, but also applies that sort of like mischief magic like viral juice or whatever. Yeah.

1:57:07

>> Um so the thinking here was we might as well open up that to external clients.

1:57:13

it can become a profit center, which is great.

1:57:14

But the other part is like maybe it can make our world a little bit bigger, too, because we're hungry for more formats.

1:57:21

We've made so much stuff over the last like six years.

1:57:23

What have we not touched yet?

1:57:25

And like maybe we won't get sued this time around.

1:57:27

Like maybe there's just stuff that we can do that's like cool, we can uh >> Well, it's it's there what what seems obviously exciting to me is partnering with with global companies.

1:57:38

I mean, sure, I'm sure you'll partner with with startups and and scaleups and things like that, but what what gets exciting at a certain point has to be scale.

1:57:47

So much, hey, instead of selling like a thousand of this thing, can we figure out can we partner with a >> company does an Apple ad?

1:57:53

It's like that's just not there's spe there's something special about that about just like giving it.

1:57:59

It's almost like patronage in some ways.

1:58:01

It's marketing, but it's just like it's still art and it's cool and it's like un unbridled. >> No, totally. Totally. Right.

1:58:07

Like the way that we look at these is like it's not a service that we're providing necessarily.

1:58:13

Like there are brands and entities in culture that we perceive as cultural material.

1:58:17

The same way an artist looks at their material is like paint or sculpture or marble or whatever.

1:58:24

Like for us, >> Coca-Cola is material.

1:58:25

Give us that and we'll make something new with it instead of fabricating a marketing story to sell more units that people see through now anyways. Right.

1:58:36

the opportunity is like there's just so much cultural material being left on the table to make new things with.

1:58:42

>> So talk to me about inbound versus outbound.

1:58:45

I feel like >> in that vein of like there's cultural material if you come up with just like I have a great idea.

1:58:53

>> It it would only work for this particular brand because the nature of the idea is tied deeply to the lore of that brand.

1:58:59

They're never going to think about this, but I need to I could the only value I could get out of this is selling it to them.

1:59:05

Uh are you going outbound and pitching big companies and saying I have a I have the idea for you or is it more like there is an actual process where you can sit down do >> I'm doing outbound outbound right now.

1:59:20

This is a message for Gabe Newell from Valve.

1:59:22

We want to design your submarine. >> I love that.

1:59:27

>> That would be an amazing project.

1:59:28

>> That's that's the outbound.

1:59:28

That's what I want to do.

1:59:30

Or uh extraterrestrial space research organizations.

1:59:33

We want to design that you're putting in space. >> I love it.

1:59:38

>> We were look, we were fast company's number one design in the world in 2023.

1:59:43

People don't think about that cuz they think about viral, but if you actually look at like what we do, it is unparalleled.

1:59:48

So, let us uh let's like put some stuff.

1:59:51

>> We got a we got a space play for you with with a company that we're very close with.

1:59:54

I won't say anymore, but uh >> yeah, we we'll talk after the stream.

1:59:58

>> Um I I have a bunch of uh I have a bunch of questions.

2:00:01

one quickly on the business model of the new agency.

2:00:03

business model of the new agency. Is there a world where you would you would uh you know the standard agency model will be like hey let's do this project and we'll charge you like a million bucks or a couple million bucks but there's some scenarios where you could

2:00:16

actually I imagine if it was a one-off more like drop style you could take a percentage of of the sales even is that something you would ever explore to get to really bet on yourselves and say like hey we think we could sell like a $10 million of this and we'll only price based on on success. Is that is that

2:00:32

Is that is that something you would consider doing?

2:00:36

>> It's totally on the table and it's it's not that's not even uncommon from like the collaboration route.

2:00:40

Like some of the collaborations that we've done in the past have been that kind of model where it's like you make a bunch of product together, we both invest on our own ends and then you split the cut afterwards.

2:00:50

I think what's more interesting though is because of Mischief's position as like this artistic entity like in a way like we have a personality, we have an audience >> and so we're actually like building a pipeline of joint ventures. >> Yeah.

2:01:07

>> With Yeah, that's what I was getting at.

2:01:10

>> It's not even about 10,000 units.

2:01:10

We're talking about millions of units. >> Yeah. Over forever, basically. >> Concern. Yeah.

2:01:16

Like I want I want the I want a mischief themed line of of kids toys, right?

2:01:21

Like I would happily buy those.

2:01:23

I I would buy a bunch of those, right?

2:01:26

And so that's like if you partner with the right company that should be in the the you know my personal lifetime value would be in the hundreds of dollars and I would even pay a premium. >> Yeah.

2:01:35

>> Because I know it's you guys behind it.

2:01:38

Um another question I have >> uh you know I've I've leveraged the the drops kind of model to do marketing.

2:01:43

I was inspired uh by you guys uh uh to do marketing for my last company, Party Round.

2:01:51

And something that I've noticed recently, a lot of companies in tech say they want to do drops, and it's because they're inspired by mischief and they just see like drops get attention. I want attention.

2:02:01

I >> One drop, please, sir. >> One drop, please. >> One drop, please. >> The usual, sir. >> Sign me up.

2:02:06

No, but the thing I usually push back on and the thing I'm curious to get your opinion on is like I think when companies think they want to do drops, what they really want to do is true advertising and they don't even know the difference.

2:02:18

And so like where where's that line for you and how often are you know would you talk to a company and it's like what you really want is to do advertising. >> Yeah.

2:02:27

Just like a consistent message delivered via every channel forever might be the right thing for >> not even forever or like a few months for a quarter. Yeah.

2:02:35

I would I would take it even further and it's it's like that is the final goal which is like get a message out into the people get them talking whatever but like the >> the focus on the drop is is misplaced.

2:02:51

>> Drop is just a delivery mechanism for something new.

2:02:56

>> What these guys are actually looking for like what they actually want at the end of the day is how do you break through the noise?

2:03:01

And right now you can either just make more content, which honestly the ROI, I think, is going down on that.

2:03:08

There's too much content and not enough demand. >> Yep. >> Sorry. It's just how it is.

2:03:11

But the opportunity is you can make new things that fit within your brand parameters, that use your brand as cultural material, that are actually interesting, that are tactile, that have a story to them.

2:03:26

And if they happen to be ephemeral, then it's a drop.

2:03:27

But don't get hung up on the drop.

2:03:29

just make something interesting.

2:03:31

That's >> all are ideas valuable and and and and if you had to you know what kind of think about like a a making something new, how much is it is it the idea versus the execution?

2:03:45

>> Do you hire Do you hire idea guys and then like operational excellent people and they're like a dividing line? >> Yeah.

2:03:53

And and to me I I John and I will sit down with companies that we're friends with and we'll generate like 20 ideas.

2:04:00

It ends up being super frustrating because you're like >> we're get we're feeding you this like incredible idea.

2:04:06

It doesn't even cost a lot to do and if you do it right like it could be like a million dollars of like you know brand value, however you want to measure that.

2:04:15

But it it it feels like you can give people these these things that in your hands they might be million-dollar ideas but but uh >> poorly executed >> poorly executed negative value. >> Totally.

2:04:27

I think it's like uh like squeezing a balloon on two ends.

2:04:29

Like there's the conceptual power and then there's the execution.

2:04:35

>> Uh there are things where like the concept is really high.

2:04:36

And the execution doesn't have to be crazy.

2:04:38

You don't need a huge budget.

2:04:40

It's just a good idea and it's going to work.

2:04:41

What you typically see a lot of brands do is they're afraid of good ideas because good ideas have not been done by definition.

2:04:50

Um so they lean more on the execution and you see like large budgets and like huge bloated teams and production.

2:04:57

So it's sort of like now if you can dial in both that's the magic right that's but that's really hard to do because more money more risk you want to it's you just you don't you want to derisk on the concept side.

2:05:08

John, to your question, uh, I don't hire idea people.

2:05:12

Everyone here is an idea person.

2:05:15

We did, I'll give you an example.

2:05:15

A couple of years ago, we made handbags, right? They're very expensive.

2:05:19

The smaller they get, the more expensive they become.

2:05:22

So, we made a microscopic handbag that sold at auction for $63,000 at Paris Fashion Week, at Corell's whatever coronation of Louis Vuitton.

2:05:33

That idea came from a summer intern.

2:05:36

Anyone here can have an idea, right? the lawyer.

2:05:38

Our general counsel joins brainstorms.

2:05:40

Our ops lead joins brainstorms because being creative doesn't mean you have good ideas.

2:05:44

Being creative just means you're human.

2:05:46

You have insights and you have reactions to things that are going around you. That's it.

2:05:51

And then also like being creative means you're not afraid to like look stupid.

2:05:55

And I think that's like a big defining trait here because >> anybody can look stupid here including myself.

2:06:04

What uh have you ever seen highly complex ideas work well as marketing advertising stunts?

2:06:11

This is something that uh we we end up >> I think David Oggov said like simple ideas are powerful ideas.

2:06:18

That's certainly something I've seen in Mischief's work throughout your entire career.

2:06:23

But is there some sort of contrarian or nuance to that concept?

2:06:29

>> I think there could be.

2:06:29

The the line that we use here is a good idea should slap in one sentence and then slap even harder in three.

2:06:38

>> And that just means there there are layers, right?

2:06:40

Like you appreciate it for the headline, but if you like really dig into it, you're like, >> "Wait up.

2:06:45

Is this thing making fun of me? And do I still like it? And am I gonna it?"

2:06:52

And if I buy it, am I a chump or am I a patron of the arts? Right?

2:06:57

It's like uh the layers are good.

2:06:57

The ambiguity is good, but at the end of the day, like from a marketing perspective, just being practical right now, like people's attention spans are so low and honestly like people are just we're all like very distracted.

2:07:12

So, you don't have a lot of window to like nail it.

2:07:14

So, you do got to have that one liner like pretty dialed in.

2:07:21

>> How uh how dialed in is your crystal ball when it comes to releasing new things on the internet?

2:07:26

Like do you have a do you feel like you usually have like a pretty accurate sense of how much attention something will get?

2:07:34

>> Do you get surprised >> or do you still get surprised often?

2:07:38

>> Totally still get surprised.

2:07:38

Like you you always get surprised.

2:07:40

I think um it's easy for us to fall in a comfort zone where we're like oh yeah like this will work because what that ends up doing is you start coming up with ideas based on a framework of what has worked. Mhm.

2:07:53

>> So we actually put in a lot of effort to come up with formats and mechanisms that don't really fit into our existing framework that we're like I don't know if this is going to work and is this offensive? I don't know.

2:08:07

Are people going to >> buy slices of this giant foam baby that we're cutting up in Brooklyn? I don't know. >> They did.

2:08:17

No, you know, like >> is there a project that you think is like criminally underrated?

2:08:21

People are obviously familiar with the boots.

2:08:24

They're familiar with the uh the the the shoes and the footwear stuff and some of the other projects.

2:08:29

But what what what's your what's your example of a project you're like, "Ah, that's still like one of my favorites." >> Good question.

2:08:35

And the reason >> it didn't really get seen as much is because to be honest, we were a little bit scared. Okay.

2:08:42

And so we end up buried it in an art show that we did.

2:08:47

But uh are you guys familiar with the paradox of the ship of Thesius? >> Yes.

2:08:51

But explain it for the listener. >> Totally.

2:08:53

So it's like uh it's this old paradox where imagine you have this like large wooden ship and every day over seven years you're coming up with a new piece of wood identical to a piece of wood on the ship and you are swapping it out.

2:09:05

The question is after 7 years is it the same ship or is it a different ship?

2:09:12

We found that uh a very interesting question and so we decided to uh apply that in our own way to a sink located in the American wing of the Metropolitan Museum of Art. >> As one does.

2:09:27

>> How do you as one does? >> Yeah. Explain.

2:09:29

>> So we we found the sink. We did recon. We uh not me.

2:09:32

I'm not I it wasn't me personally.

2:09:35

It was somebody here names.

2:09:38

But we found, you know, we we tracked down the serial numbers, the parts, uh, ordered exact like replica.

2:09:43

We we ordered the exact same parts, built it here, tied off the dimensions of the bathroom here so that we could practice swapping things out.

2:09:54

And then over many months, we would walk into the Met and do a quick swap and leave.

2:10:01

Then the next day, do it again and do it again.

2:10:03

Bolt by bolt, screw by screw.

2:10:07

>> Just an art enthusiast.

2:10:07

I just I just like art and I I go to the bathroom. That's all I'm big.

2:10:12

How did you get the big piece through?

2:10:14

Isn't there like one piece?

2:10:16

>> That that so that we decided to leave. >> Okay.

2:10:19

>> That we did not swap out.

2:10:19

Although we had um >> a plan.

2:10:23

>> We we we we fabricated a wheelchair that could have hidden in the seat.

2:10:26

But that I'm telling you, we got we got a little bit we we were honestly scared.

2:10:32

>> You got a little carried away.

2:10:34

>> We still we still did it. we got everything else.

2:10:35

And uh so yeah, they they they have our sink and we have their sink and ours is now um part of like our art gallery. So >> that's amazing.

2:10:45

>> Although although I believe they figured it out and they've replaced it cuz >> as soon as we announced the art show, even though the sink was sort of like hidden in all of the other pieces, um people definitely found out and there was a line in the Met for that bathroom. people.

2:11:03

>> They should have they should have sectioned it off, turned it into an installation.

2:11:07

>> So, if anyone if any of your if any of your viewers are on the board of the Met, uh the coolest thing you could do right now is acquire that sync from us and put it back. >> Let's make it happen. >> Let's make it happen.

2:11:18

Army, >> how how many millions of dollars do you think you've left on the table by not doing crypto projects? >> The painful question.

2:11:30

Is it like a It has to be like a hundred million like at least. >> Yeah, probably. Honestly, probably. Probably. Yeah.

2:11:38

>> Um and is there is there um uh like there I'm sure at some point there will actually be a way to do this.

2:11:47

Like there has to be something that makes sense.

2:11:48

makes sense. Like we did something back in the day with party round where we did it was called helpful VCs where we made crypto punks for every VC and then we put them on a website >> and we said you have four hours to retweet these otherwise we're going to auction them off and it was like it's

2:12:03

for like charity right we weren't like trying to make money and that was like one of the ways that we broke through and like shortly after um >> uh >> I think there might be something to do with crypto in like stable coins where there's not as much gambling involved but there might be something with crypto I don't know. >> Yeah. Yeah. You have to remove you have >> Yeah. Yeah.

2:12:16

You have to remove you have to make it not like a speculation, you know, >> based on >> because as soon as you leave the like like the the people who get the joke are and then the people who don't get the joke show up and then and then they're losing their life savings, that doesn't feel good.

2:12:30

Uh whereas whereas like if somebody, you know, is is bidding on that handbag um like they they probably know what they're doing.

2:12:37

They understand that it's not and they understand that.

2:12:40

>> Look, I think I think that crypto route for us is um we've often plotted our own death and I think there's like a chaotic evil way to go out where you just pull that rug. >> Yeah.

2:12:53

>> I shouldn't even because now everybody's going to know.

2:12:56

>> Everybody's going to know.

2:12:56

Well, I think um switching gears to to another uh topic that I'm I'm sure you guys uh so one, I don't think AI is going to take your guys' jobs anytime soon.

2:13:06

Uh despite it being able to maybe piece together, you know, random ideas, uh it's so much about >> uh taste and things like that, but uh how excited are you to leverage that technology for for some of um for some of these various plays?

2:13:27

Honestly, TBD and I think I think it's I do I will say like unlike the whole like crypto craziness of the last couple years, um we we didn't get into that cuz we didn't feel like it was here to stay yet.

2:13:40

That that's like the truth.

2:13:40

Like when you see the herd run so fast, it usually means they're running towards the end of a cliff.

2:13:47

>> There's a little bit of a similar sentiment right now.

2:13:49

That said, I it is rooted in something real.

2:13:52

Um, but we're not in a rush to like use it necessarily, right?

2:13:58

If it's going to be here forever, we've got time.

2:14:01

Like it's mischief doesn't win as a first mover.

2:14:05

If you look at all of our work, it was never about being the first to react to something or the first to like make use of a format. >> Um, so we'll see. We just have to see.

2:14:16

>> How do you think about uh are there any projects like time is an interesting variable that you can play with, right?

2:14:23

Like part of the thing with the the the the sync stunt that you guys did is is just the narrative and the story of like you guys were conducting this in secret for a long time.

2:14:32

It was like a very and and that's part of what makes it so entertaining is kind of like the the lore of what went into it.

2:14:38

But but have you thought about even longer term plays like things that you could do over a decade or or or 50 years that in hindsight would just be absolutely hilarious?

2:14:49

And uh but but again take this sort of incredible planning. >> Yeah. Yeah.

2:14:55

Honestly, we are at that stage of just like our lives and as a company now to be thinking like that, right?

2:15:03

Cuz like if you think about it, we >> we're in >> Yeah.

2:15:08

If you if if you had, you know, I know you guys have have a few investors.

2:15:12

If you told your investors, first thing we're working on is a 50-year stunt and you're they're telling you like, "All right, buddy."

2:15:17

Like, "Why don't you put some points on the board first?" >> Yeah. Yeah. Yeah.

2:15:22

>> But you guys, I feel like have earned it now.

2:15:23

Like I I you know, something like a decade. >> No. Yeah.

2:15:27

Cuz like we've we spent basically like we've only been around for the pandemic and post like a little bit post pandemic, right?

2:15:35

bit post pandemic, right? And all of that was just like you don't know what's coming like live day by day like just survive like banks are shutting down okay get through to till tomorrow right >> but now we've been around for a second >> we're more well known we have our

2:15:48

existing infrastructure and now we really are asking the questions of like okay what does that legacy look like in 10 years or should we uh should we disappear for 20 years and then come back that would that would actually be the coolest thing we could do is uh go dark for two decades and then come back with a >> banger. It's

2:16:08

It's >> worked for It's worked for other It's worked for other artists.

2:16:10

Um I I I uh could you do something?

2:16:14

So So we both loved uh Nathan Fielder's latest sort of like aviation stunt television show.

2:16:19

I can see >> I've heard a lot about it. I haven't watched it.

2:16:24

>> I mean you you should.

2:16:24

It's one of the few people in the world that I think is executing on your guys' level in the sense of >> he's next level. Yeah.

2:16:31

Like one sentence, you know, pitch.

2:16:34

That's hilarious, but then the more you get into it, like the funnier it is.

2:16:39

>> But could you do something like that for like the cost of housing or something like that?

2:16:43

Like sort of a decade long stunt like Mischief like Mischief figures out a way to like reduce, you know, cut the cost of housing in the United States.

2:16:52

I mean, that was the craziest. Yeah.

2:16:54

>> Like, we um if there are any uh like pricey neighborhoods that would love to commission us as an artist to install a sculpture, we have a really good uh trailer park trailer that we would love to erect in a public square that will drive.

2:17:14

>> I can just see like the like the right RV.

2:17:16

Like there's this RV outside of our gym. >> Yeah. Yeah. Yeah.

2:17:19

kind of yellow from the sun.

2:17:21

Really, the windows are open, the aluminum on the outside. Yeah, Winnebago.

2:17:27

>> It could be a stunt for Winnebago.

2:17:29

>> What do you um how do you feel about the state of uh social media like broadly?

2:17:36

>> Uh it it feels like we've gotten to a point where in many ways the platforms look so similar, but they also have their own character.

2:17:44

Some of them are becoming Lindy.

2:17:46

even thinking about how durable like Twitter and Axe have been despite all this change, >> but what what's your and and do you even do you do you feel like you need to use social media or are you better off not using it and then just releasing you know stuff on the platforms? >> Yeah. Yeah.

2:18:04

It's um definitely a lovehate relationship with the with the platforms, right?

2:18:08

Because like platforms, if you if you really think about it, they started as a way to share things that you found were interesting and now they are mostly ways to share uh photos and videos of you talking about things that you find interesting.

2:18:28

>> It's kind of become like less novelty based and more vanity based, >> which is fine.

2:18:33

That's like a fundamental human instinct, so I get it.

2:18:35

Um, I would continue to love to find a way to not use the platforms to disseminate information.

2:18:44

And that's kind of how we started.

2:18:46

Our first audience portal was Venmo.

2:18:48

Um, >> we had a huge audience on Venmo.

2:18:50

And we like had this huge phone bank, uh, like a Chinese phone click farm.

2:18:55

And we every time we had a new project, we would Venmo our audience a penny and they would get that notification because that's the most powerful notification layer in your >> ching. The chuch-ing, right? >> Yeah.

2:19:10

>> I am now banned from Venmo for life.

2:19:15

>> Well, PayPal, >> let me know. >> PayPal.

2:19:19

We we'll have the CEO of PayPal on.

2:19:22

>> I think the last company that came on works with Venmo.

2:19:24

We'll we'll uh we'll work on this.

2:19:26

>> We should get this first.

2:19:26

This is this is important.

2:19:28

Um uh we were we were reading an article in the New Yorker yesterday about this concept of IRL brain rot, the the the examples were the viral >> laboo laboo Dubai chocolate, the the viral Italian sandwich.

2:19:43

Uh we saw a few of these.

2:19:46

Joe Weisenthal at Bloomberg was also writing about the the these kind of like viral as being something that you stamp on your product as a product feature almost and I was wondering if uh if you're worried about the the the spillover from uh internet slop culture into the real world just kind of how you're tustling with this idea of brain rot coming into the real world.

2:20:10

I don't even know if you read the article or you have thoughts there but >> I think I saw it on Instagram. Yeah. >> Yeah. Of course. Yeah. Yeah.

2:20:15

You probably did a frontfacing video talking about it or you saw someone do a frontfacing video talking about it. >> Reaction video. >> Yeah.

2:20:25

>> I mean, look, it would I I >> I don't know if it's a good thing or a bad thing, but like what Mischief has put out historically is kind of in that brain rot category. >> Yeah.

2:20:39

I mean, the boots the boots, right?

2:20:41

It's like absurd absurd aesthetic and and maybe maybe even I could be so conceited as to say we were part of creating the responsible.

2:20:50

>> No, you guys definitely that that it was an era of minimalism and it was like the the Everlane era and the boots the boots are like, you know, an incredible reaction to that of being like >> absurd absurd aesthetic just like popping imagery off of your screen.

2:21:05

But then the ch the challenge with this this trend, right, is if you go back to minimalism, it's it actually becomes I mean maybe maybe as as like for me, right, I'm I I became part of the mischief audience and mischief customer in my early 20s. I'm in my late 20s now.

2:21:26

Eventually, maybe I'd want some ultra minimalist uh you know, mischief item for my home.

2:21:32

But but but but at the same time like once you get on the on the on the maximalist track like you're kind of on this maximalism treadmill where you just got to keep like going crazier and crazier and and you know maybe maybe in the internet era there's no turning back.

2:21:49

>> It it's totally possible, right?

2:21:49

Like we might we might have built a machine so effective that it's now our master and like how do you break free from that, right?

2:21:58

But um I'm not too worried.

2:21:58

Like we don't we actually don't think so hard about it.

2:22:03

We just kind of make the things that we like and then uh if it ever stops working then >> do something else. >> Yeah. Figure it out.

2:22:10

>> Well, I I have an idea to put out in the ether.

2:22:13

I think next time a an A-list celebrity is going through a massive comm's crisis, PR crisis, or a scandal, they should hire you guys to create something more more viral than than the scandal.

2:22:25

just need to they just need to do another crisis to compete with that.

2:22:29

All they got to do is go to Twitter and be like, "Send a Bitcoin to this address and I'll send you two Bitcoin."

2:22:33

And then they actually do it. >> They actually genius.

2:22:38

Yeah, that is a get that is a one time get out of jail free card. You heard it here first. >> Just payoffs. >> If you Yeah.

2:22:44

If you're a celebrity and you're in trouble, you're you're you're caught at a Coldplay concert, that's the only way you can reverse the tide. >> You will be good. Go for it. >> Yeah. >> Awesome. This is great. Yeah, thank you.

2:22:56

>> It's really great great to get the update and uh just hear your so much clarity on on all these different areas that we talk about all the time.

2:23:05

>> Yeah, we could talk for hours.

2:23:05

Yeah, we'd love to have you back uh whenever whenever it makes sense.

2:23:10

>> Yeah, we gave Gabe and I kicked around an idea that we won't share now.

2:23:11

We'll we'll we'll leverage that for for your next appearance.

2:23:15

I >> can't I can't wait for it. >> 100%. >> Let's do it. >> Awesome. >> Awesome.

2:23:22

I'll talk >> Great to catch up. See you. >> Cheers. >> Bye. >> Absolute legend.

2:23:28

>> One of the best to ever.

2:23:29

>> What if we found out that that Dubai Chocolate was was uh was was a mischief stunt? >> It's It's possible. It's possible. >> Anything's possible.

2:23:38

>> Well, if you want uh to do a stunt, you want to promote a stunt, you want to have some fun with billboards, head over to adquick. com.

2:23:44

Out ofome advertising made easy and measurable.

2:23:46

Say goodbye to the headaches of out of home advertising.

2:23:49

Only adqu combines technology out of home expertise and data to enable efficient seamless ad buying across the globe.

2:23:54

Uh you should also get a hitter. Go over to bezel. Get bezel. com.

2:23:59

Shop over 26,000 luxury watches.

2:24:02

Fully authenticated inhouse by bezel's team of experts.

2:24:04

What you got for me on the timeline? Jordy Hayes. >> Uh wow.

2:24:09

Uh our our >> timelines in turmoil.

2:24:11

Are you seeing this Franis Chole post or you want to go?

2:24:15

>> I was saying our our next uh our next guest has a 133 ETA.

2:24:17

Okay, >> coming into the studio, David Senro.

2:24:20

He says, "A fire broke out on my route to the studio." Little >> Oh, no.

2:24:24

We'll have to read more posts >> now.

2:24:26

He said, "Little do they know, nothing nothing will stop extreme podcasting."

2:24:31

>> We'll see you in a little bit, David. >> In person.

2:24:32

Yeah, let's dive into this post.

2:24:35

You want to read through it?

2:24:37

>> So, uh there is a debate um France Chole and uh someone named Oend on the timeline.

2:24:44

Uh so there's this debate around how will AI impact worker productivity.

2:24:52

It's very clear that when you use AI it feels like you're getting a speed up.

2:24:57

When you use uh a a coding agent it feels like you're getting a speed up.

2:25:02

Meter did a study and found that maybe some AI tools are making you less performant.

2:25:08

But uh so there's this debate uh and for a long time there was this idea that AI will supercharge workers and labor and anyone who's using AI will be uh more uh more productive than someone who's not and and the advice that was given um by pretty much everyone in the industry was like you need to learn how to use AI effectively because if you don't you'll be left behind.

2:25:32

And so um Oyan shares that LLM adoption rose to 49 point 45.

2:25:34

9% among US workers 18 plus as of June July 2025 according to a Stanford World Bank survey.

2:25:45

Inference demand will continue to surge not just as more and more user uh more usage per user but as newer more advanced genai models require far more compute.

2:25:56

And so um obviously no one was using LLM just a few years ago.

2:26:00

Now almost half of the American workforce over age 18 are using AI tools in their daily work.

2:26:08

Um Franis Chalet of course is asking the question about when is this going to show up in the economic statistics.

2:26:15

Uh he says LLM adoption among US workers is closing in on 50%.

2:26:21

Meanwhile labor productivity growth is lower than in 2020.

2:26:23

Many counterarguments can be made here like they don't know yet how to be productive with it.

2:26:30

They've only been using it for one to two years, 50 50% is still too low to see impact models next year will be unbelievably better, etc.

2:26:37

But I think we now have enough evidence to say that the 2023 talking point that LLMs will make workers 10 times more productive.

2:26:48

Some folks even quoted a 100x is probably not accurate.

2:26:51

And so he gets in a little bit of a turmoil here, a little dust up with Gnome Brown polomial over at OpenAI.

2:26:58

Francois continues and explains a little bit more of his uh his reasoning here.

2:27:04

He says, "By the way, I don't know if people realize this, but the 2020 work from home switch coincided with a major productivity boom and the late 2021 and 2022 back to office reversal coincided with a noticeable productivity drop."

2:27:18

It's right there in the in the statistics. narrative violation.

2:27:22

Productivity growth is now back to pre2020 levels.

2:27:24

So for a lot of jobs in the American economy, working from home was actually a benefit. I'm not a huge fan.

2:27:30

I I I felt like maybe I was more productive, but I certainly didn't enjoy it.

2:27:33

Uh and so we've been having person.

2:27:37

>> It's it's interesting.

2:27:40

>> You'd have to dive in. Okay. >> Depends on work. >> Well, yeah.

2:27:42

And you can also be more productive but on the wrong strategy.

2:27:46

Like I I found it >> working from home during that era >> and having a remote team I found it very difficult to >> um like change strategy quickly even on small things easy to lock in but not you were zoom I felt less yeah adaptable right you'd sometimes waste three weeks working on something where maybe if you were in the same room you would have like not made a different decision.

2:28:09

So Nome Brown over at OpenAI says I don't Yeah. Yeah. Get the gunshots ready.

2:28:14

I don't recall a lot of people in 2023 saying that within a few years worker productivity would 10x due to LLMs.

2:28:20

I do recall a lot of people in 2023 saying LLMs can't reason though.

2:28:26

Shots fired on the timeline. That is a fantastic.

2:28:31

>> Ransis says back in 2023 AGI was one to two years away in the form of GPT5 no less.

2:28:37

Productivity was about to create increase 10 to 100x and developers were about to go extinct.

2:28:42

People who question the narrative were a tiny minority. We were proven right.

2:28:48

>> He was the John Brown standing up.

2:28:48

I believe that AI will not 10x productivity.

2:28:52

Tyler, do you have any takes on this? >> Uh yes.

2:28:56

>> Will AI be able to get Tyler a shirt?

2:29:01

>> An AI agent might but he might be able to do it faster himself. Who knows?

2:29:04

>> Still unclear on that.

2:29:04

But I I think um the idea that people don't know like aren't using them efficiently yet makes sense, right?

2:29:11

Cuz so if you look at that meter study >> Yeah.

2:29:14

>> Um it was about how like open source developers are actually slower when they use AI tools, but like if you actually read it, it was >> open source developers who it was their first time ever using AI tools. >> Oh, interesting.

2:29:25

>> So it's like obviously like >> small, you know, some learning curve to using >> like cloud code or whatever like it's you're not going to instantly become way better. >> Yep.

2:29:34

So I think it's it's kind of a similar thing there. >> Yeah.

2:29:36

I mean we heard that there's a yeah there's an interesting phenomenon where some companies are getting more value out of having um uh having designers go from zero x engineers to 1x engineers and they're not seeing as much of like the 10x engineer becomes a 100x engineer um because the problems are still like intractable in some ways.

2:29:54

Uh nome brown says I was on the AI job market in 2023.

2:29:58

I spoke with a lot of researchers uh a lot of frontier labs.

2:30:03

Some thought scaling pre-training was sufficient.

2:30:04

Some thought additional paradigms were needed.

2:30:06

But regardless, almost none of them thought AGI was one to two years away.

2:30:10

Yeah, that was something that was maybe more parited in the uh uh on the podcast circuit and from those who were raising the mega rounds who needed to kind of justify underwriting AGI at uh you know a thousandx revenue multiple. Um it's interesting. Um I don't know.

2:30:26

It it it is interesting that uh to look at the economic statistics to keep going back to um what's happening in the labor market, what's happening with unemployment.

2:30:35

Like it would be it would be weird to have a unemployment tick up without productivity tick up because that would just mean lower GDP and that seems like an like an impossible scenario.

2:30:48

I I'm somewhat receptive to the idea that uh that maybe AI is substitutive for a different type of work and and maybe it's not driving major productivity booms just yet.

2:30:58

But uh I I I would I have trouble believing both narratives that that AI is driving massive unemployment and also it's not driving increased productivity because companies would just be doing less then and we're seeing earnings growth.

2:31:13

We're seeing growth in the economy.

2:31:14

Uh we're seeing high GDP growth.

2:31:17

So um one of those has to be has to be right.

2:31:20

They can't both be >> Well, breaking news.

2:31:22

Uh President Trump was on Newsmax, not our not our internal software that we used to create the show, but uh the TV show. Yep.

2:31:34

>> President Trump uh tells Todd Starns he's going out on patrol tonight with DC law enforcement and the military.

2:31:41

>> They're really taking DC seriously, aren't they? >> Yeah. Yeah.

2:31:43

Um, if you're in DC, have a keep an eye out.

2:31:46

Nate Elias says, "Going to have to block Elon Musk soon for all the AI goon spam.

2:31:52

Feels like I can't open X in public."

2:31:56

I think a lot of people are feeling that way. >> Skill issue.

2:32:01

I've still fine-tuned my algorithm so much towards tech and AI content that I don't see any of the Elon.

2:32:07

You don't >> I don't see it. >> I get it.

2:32:09

I get it because people that I follow reply to him and tell him to stop this. >> Okay. Okay.

2:32:13

Yeah, I follow Elon, but I still don't see it.

2:32:15

I see I see high yield Harry saying Tom Brady was so ready back in the day to be a finance bro. And they share Thomas E. Brady Jr.' s resume.

2:32:24

Uh he was at University of Michigan.

2:32:28

Then he went to Meil Lynch, the University of Michigan golf, >> summer intern, summer >> Polo Fields Golf and Country Club.

2:32:33

Uh he was >> sales rep says this guy could have had a successful career working at Meil Lynch instead of pursuing football.

2:32:41

Now no one on Wall Street will remember his name.

2:32:47

>> Oh funny >> Andrew Reed uh Jan had said what the heck is a series K.

2:32:52

>> That was data bricks >> data bricksong gong of the day.

2:32:56

>> I know quite a few people in SF with a series K problem. >> Yes.

2:32:59

Uh stay away from the series K.

2:33:03

Enjoy the series C, the caffeine, the yerba mate, the >> next time you have an urge to do a hard drugs, >> why not try a fridge cigarette? >> Yes.

2:33:14

>> Run says the demand for anti- AI takes is enormous and we'll take anything and run with it.

2:33:18

Meta consolidating and doubling down on MSL is being misre misrepresented as bearish on AI for example, something to keep in mind.

2:33:27

Yeah, the the the they they conducted a reorg because they added a bunch of people to the team. >> Yeah, it makes sense.

2:33:33

The >> and had a few weeks together and kind of updated.

2:33:36

Um certainly and and they it got positioned as like like layoffs almost kind of what I was reading into.

2:33:43

>> Yeah, it was very odd.

2:33:43

Um >> uh for obvious reasons it's not that popular when we're like this technology will bring about a turning of the age and reshape the thought the thought world of man and even a miraculous year in AI progress can be spun into a disappointment.

2:33:56

Uh careful observers will note that revenue growth of the consumer AI industry has outstripped the most bullish expectations.

2:34:02

We maxed out the metacalcalculus predictions and nobody is maya culping on this.

2:34:07

You're not going to get an apology from maroon. No way.

2:34:12

He's absolutely printing in consumer AI adoption. People love it.

2:34:19

Anyway, what else is going on?

2:34:19

Uh Elon Musk, we were talking about the the potential spam.

2:34:26

He is unpacking his ideas around the birth rate.

2:34:29

People saw this as inong in congruent.

2:34:31

How can you possibly be worried about the birth rate when you are creating romantic companions that by definition cannot reproduce with their companions?

2:34:40

Elon Musk says >> Theoon Theo was was you know kind of getting into this with his interview with uh >> a little bit with Sam.

2:34:49

Yeah, it was it was a very funny interaction where he asks Theo asks Sam Alman, "What do you think about test tube babies?

2:34:56

The idea of having children outside of the womb >> that you could visit on the weekends, he specifically said. >> Yes. Yes. Yes.

2:35:02

And and so Sam is kind of like that it sounds maybe like that's futuristic and that might happen and there are some scientific reasons but it feels unnatural to me.

2:35:10

I'm not that into it says Sam.

2:35:12

And Theo just goes oh like I thought you'd be into that.

2:35:15

It's like very revealing that that Theo clearly sees Sam as like this like techno >> dystopian. >> Yeah.

2:35:22

Or just like this crazy crazy tech futurist guy like from the future.

2:35:24

Uh and so they kind of like they kind of like work through it on the conversation.

2:35:29

But it's a very funny clip because uh they're both like playing the game of like who are we like what what what how how do we perceive each other?

2:35:36

Um >> but uh Elon says AI is obviously going to oneshot the human lyic system.

2:35:39

That said, I predict counterintuitively that it will increase the birth rate. Mark my words.

2:35:46

Also, we're going to program it that way.

2:35:49

>> But what I find interesting is what Nikita Beer says here.

2:35:51

He says, "Grock companions duolingo for dating."

2:35:53

So, the idea that you could go to a Grock Companion, learn amazing pickup lines, like, are you a Nvidia GB200?

2:36:02

Cuz I'm getting hot hanging out next to you.

2:36:07

And uh and and then take that that pickup line on the road and meet someone real.

2:36:15

Uh the more interesting thing, I think, is is if two people are talking to companions and they're a perfect match, just introduce themselves.

2:36:22

introduce them to each other and say, "Hey, we are foregoing the the the future revenue here and we are introducing you in IRL." Um, but we'll see.

2:36:32

Uh, the proof is in the pudding.

2:36:34

He's got to put his his AI where his mouth is and actually deliver a product that increases.

2:36:41

>> Hopefully, he doesn't put his AI where his mouth is. >> Who knows? >> A literal sense. >> Who knows?

2:36:46

>> Uh, China is pushing for a total ban on the use of foreign chips for inference.

2:36:50

>> They're not a fan of Nvidia. >> Financial Times.

2:36:52

Some Beijing policy makers are pushing to ban foreign chips altogether for inference, which accounts for most AI demand, according to a person recently summoned for a meeting with them.

2:37:01

That is unlikely to happen soon due to a shortage of domestic chip supplies, which Beijing hopes.

2:37:09

>> Who do you think's going to sort it out? Who you think? The juggler. Jensen the juggler.

2:37:13

He's able to juggle the interests of America, the interests of China, the interests of Taiwan, the interest of Nvidia.

2:37:18

Jensen the juggler will get it done.

2:37:22

>> Hopefully he doesn't drop one.

2:37:22

Uh but we will talk to Aaron Gin if there is a dropping of a juggle ball from Jensen the juggler.

2:37:30

>> Google declares the green verse blue bubbles debate silly and tired and signal says precisely what a green bubble would say.

2:37:37

I will say my uh little brother uses an Android.

2:37:39

It's gotten bet quite a lot better to text him.

2:37:43

We can now effectively use FaceTime through this weird kind of hangout >> integration thing.

2:37:48

Certainly better, but um >> it's definitely uh it's definitely still a real thing.

2:37:55

>> You got something, Tyler?

2:37:56

>> Uh yeah, this is different though.

2:37:57

>> No, out of the White House, new executive order uh they're establishing a national design studio.

2:38:02

>> Okay, >> so giga bullish for Figma, obviously. >> Let's go.

2:38:07

>> Universal basic, >> but yeah, like a new a new design language announcing America by design. >> Okay.

2:38:12

National Initiative to improve experience for Americans >> like UI UX or >> Yeah.

2:38:17

new life into into the designs of sites where people interface with their government. >> I'm getting Stop. Stop.

2:38:24

>> Chad is saying you look like Greg from Succession.

2:38:27

>> Uh that's the first time I've ever gotten that. That's weird.

2:38:28

Never gotten that before.

2:38:29

Anyway, >> I've never I've never >> You've never said that. I've gotten that a ton. Of course.

2:38:33

I I actually live in the same building as him. >> Okay.

2:38:36

What if what if Greg from Succession was a Sigma?

2:38:37

That's kind of the question that that >> Well, the actor who plays Greg from Succession is in fact a sigma sort of lone wolf.

2:38:45

He had his own apartment.

2:38:48

>> Uh Dar uh says, "Who has a house like this in the bay?

2:38:53

I would like to spend some time together, preferably in your house."

2:38:57

>> That is a cool That is a cool design.

2:38:59

Like the soft wood tones, the the uh the brown the >> It feels very comfortable. >> Conversation.

2:39:05

>> It actually doesn't feel that comfortable. Like look at that couch.

2:39:09

That couch does not particularly look like loungeable.

2:39:11

Um I'm much more I'm a I'm I'm a fan.

2:39:15

My ideal my ideal living situation is uh a a modern glass box on the outside but then log cabin on the inside.

2:39:24

I want like the moose head on the wall of the glass and then I want over stuffed chairs. Brutalist outside. >> Exactly. Exactly.

2:39:32

the fortress on the outside super cozy maxing like thick rugs, warm hearths.

2:39:38

These are the things that I want inside.

2:39:41

I I I always hate going inside those glass box houses and being like, "Okay, like everything's uncomfortable.

2:39:47

There's so many corners."

2:39:49

>> Yeah, I can just bang into everything.

2:39:51

>> Look at this glass point.

2:39:53

>> Yeah, I want like a wooden coffee table that's just so rounded.

2:39:55

Like there's just no sharp edges whatsoever.

2:39:56

Uh that's my that's my ideal thing, but this is a little minimal.

2:40:01

But Gabriel says, "What does he know?" He's quoting our post.

2:40:06

>> We broke the story really after uh Ben Thompson emailed his newsletter uh subscribers on Strateer.

2:40:11

Uh we broke the news that after 21 I mean 22 years in Taiwan.

2:40:17

Uh I think >> Did you say he's moving again?

2:40:20

>> I think he's Oh, he I don't know if he said, but he's been he spends time in the Midwest.

2:40:23

So I think Wisconsin, I think Michigan, that general area.

2:40:26

Um, did you see Dylan Bruscado's reply to our uh to our breaking news that Ben Thompson has officially relocated to the United States? Welcome home, brother.

2:40:36

He he did the NBA commissioner me said, "Time to learn English, buddy."

2:40:43

Of course, Ben Thompson does know English.

2:40:44

But a lot of people were speculating this is about Taiwan tensions. I don't think it is.

2:40:47

And Ben Thompson said that exactly in his email.

2:40:51

He said, "Yes, Taiwan is complicated, but that's not why I'm doing this.

2:40:53

it's that uh my kids are at an age where we want them to go to school in the United States and we've been contemplating this for a long time and we're very happy and we always spend he already spends summer in the United States and spends time so going back and forth.

2:41:04

So, but he it was it was not uh it was not like uh the last the last uh helicopter.

2:41:09

>> One thing is for sure if there is a ever conflict in Taiwan, Gabe Gabe will quote repost his uh this victory lap.

2:41:16

Jira tickets just said uh my boss just told me quote unquote don't make mistakes like I'm a clanker >> that's brutal.

2:41:29

>> Yeah mostly I clanker is still getting more juice.

2:41:32

I'm still laughing when I hear it.

2:41:34

It is not completely dire.

2:41:34

It's >> actually gotten better.

2:41:37

>> I think it's gotten better. It's building some lore. I'm having fun with it.

2:41:40

Plus it is turning into a bit of a culture war issue.

2:41:41

Mike Salano was talking about this where there were a couple posts that went out that were saying like you should not use this. It's a slur. It's bad.

2:41:47

like don't don't reuse the word clanker.

2:41:50

And I like I like that we're already being language police on it.

2:41:53

So that that gives it a little extra juice.

2:41:55

Like it's it makes it more fun to use. >> Yeah.

2:41:59

>> Anyway, um should we go to this post from Bone GPT? >> Back on the show. >> Back on the show.

2:42:04

Thank you for posting our reaction to your post a few days ago.

2:42:09

Bone says, "Apple has no chance of catching up to Google with AI features that will become increasingly useful.

2:42:16

OpenAI thinks they can do hardware from scratch because they bought Johnny IVive, which is a level of delusion that's hard to describe. Shots fired.

2:42:21

Uh Meta is playing catch-up desperately distorting the whole market with their hire so they can ship stepmom. Google one shots fired. >> I don't know.

2:42:35

>> This is what I was kind of getting at earlier.

2:42:36

They they look really great during during the AI wars right now.

2:42:38

during during the AI wars right now. I think it's like I think it's like you know if this is if this is mobile if this is cloud like yeah AWS was the first mover in cloud Azure GCB these are great businesses you know uh did Apple kind of win mobile and get the vast

2:42:53

majority of device revenues yes was it still extremely valuable for for Google to develop Android uh did every company eventually need to have a mobile strategy and a mobile app and a mobile design like yes and so um it's it it's a little it's a little uh dramatized but that's why we love your posts phone. Keep things spicy on the timeline. Keep

2:43:11

Keep things spicy on the timeline.

2:43:11

Keep us hitting that soundboard.

2:43:16

Uh anyway, uh timeline and turmoil again.

2:43:19

Malo Bergone from the machine intelligence research institute.

2:43:23

Merie is getting in a fight with Rune on the timeline.

2:43:29

Uh Rune says Merie is an incredible name, maybe the best in AI.

2:43:34

It's a shame they don't research machine intelligence. And Malo fires back.

2:43:36

Open AI is a pretty good name.

2:43:38

It's a shame they don't dot dot dot. He said, "Yes, I know.

2:43:43

They published a couple open models recently and he has to kind of walk it back because he got hit with a community note."

2:43:48

Um, but I think uh I think Rune also got hit with a community note.

2:43:52

Maybe Merie does research machine intelligence.

2:43:55

>> Oh, I think Rune is getting community note.

2:43:56

>> Yeah, because I think machine re intelligence research institute does in fact research machine intelligence.

2:44:02

Maybe just maybe they just don't ship uh models or products.

2:44:04

Maybe they uh uh ship other types of research.

2:44:07

Anyway, >> anyways, opening eyes CFO Sarah Frier, who we've had on the show. Yes. Absolute dog.

2:44:16

>> One of the Bloomberg TV doing an interview.

2:44:20

>> She's a real CFO CFO, if I can say so myself. >> Totally. >> Yeah, >> totally.

2:44:24

Anyway, >> if you know, you know. >> Yes.

2:44:26

>> Uh, but uh she was asked about um selling infra services in the future.

2:44:31

She said, "I do think about it as a business."

2:44:33

>> This is why Ben Thompson moved back to America.

2:44:34

He's like, "I gotta end this right now."

2:44:37

Now Ben Thompson famously does not like the fact that OpenAI is uh has an API.

2:44:42

He thinks that they should focus entirely at least for now on uh on saving every GPU cycle for the consumer product that is absolutely on fire.

2:44:50

They should become they should embrace being the accidental consumer company and they should go as deep as possible before >> where they have an incredible edge. Yes. To consumer. Yes.

2:45:00

>> Where do they not have incredible edge in the just vending intelligence?

2:45:02

I mean they're doing great and it is a good business and so from a CF CFO's perspective and from Sam Alman's perspective it certainly makes sense like yeah like why not also be in that game. It is value creative.

2:45:14

It it is profit generating.

2:45:16

Um and and yet uh Ben Thompson thinks now is the time to focus uh focus focus focus.

2:45:21

At the same time I kind of take the other side of this a little bit and say OpenAI is in a fortunate situation.

2:45:28

They have a lot of data, a ton of customers.

2:45:30

they're going to have a very great consumer business that will probably throw off a ton of cash, especially once they start monetizing it more heavily.

2:45:36

And what that means is uh is they they might be the next Google in the best sense possible in this and what I mean by that is that they might be able to go and do 20% projects.

2:45:51

They might be able to go spend a couple billion dollars trying to build a new device and maybe it doesn't work.

2:45:57

But when you add up all the Google 20% time projects and Google's side projects, it's easy to laugh at all the things that they've shut down like 10 different chat bots >> businesses that have a 100 million of revenue.

2:46:09

>> Yeah, they launched Wave and they've launched uh Google Glass and they've missed on so many things, but when they hit it's great, you know, they they hit on Gmail, they hit on Docs, they hit on cloud, they hit on Whimo.

2:46:18

And so I love that you get sort of this like h humanity or America gets like somewhat of a subsidized R&D org at Google.

2:46:26

Uh and it all pencils out and it all makes sense.

2:46:32

And so uh maybe the infrastructure services uh group is not that at OpenAI, but I'm excited for these other uh these other like less focused efforts from OpenAI and I think we'll see some cool stuff.

2:46:44

Uh anyway, >> this is a good post from Stever.

2:46:46

If we can pull it up, >> read it out.

2:46:51

>> Say, "You're finally awake.

2:46:51

That fall looked real bad.

2:46:54

>> We l B2B SAS, what are you talking about?

2:46:57

You're an electrical engineer.

2:47:00

Let's go build some chips." >> I love it. >> So good.

2:47:03

Uh this post uh really going to be relatable for a lot of you. >> Yes.

2:47:08

>> Nine figure hell is real.

2:47:08

Sure, you don't have to work and have some assets, but you can't really afford a decent yacht, let alone a private army or nuclear submarine. >> It's real. It's real.

2:47:19

Get back in the wage cage.

2:47:21

>> Yeah, get back in the cage, buddy.

2:47:23

>> Get ready to send some emails, buddy. >> Captain Nemo.

2:47:25

Uh oh, they're doing an OpenAI movie >> and they're basing it on a book.

2:47:29

I think there might be two competing scripts, but uh they are finalizing the cast.

2:47:33

I believe Andrew Garfield, who is also in the social network, it will be playing Sam.

2:47:38

But Will Brown, friend of the show, has a recast.

2:47:41

Uh he has Can you name these actors?

2:47:49

>> Uh Jason, you got Jason Stathithm.

2:47:49

And who will he be playing? >> Uh he'll be two.

2:47:54

Is that >> the the joke here is that he's playing both Greg and Ilia.

2:47:58

When I saw Jason Staithm as Ilia, I was laughing out loud. I was there.

2:48:04

>> And then uh I don't The guy on the left, he was in uh a bunch of stuff. >> What stuff? What have you seen?

2:48:12

>> I don't even know his name.

2:48:14

>> Uh he he I've seen him a bunch though. Seen him around. >> Yeah. Yeah.

2:48:17

It uh he's a Nightcrawler. Toby.

2:48:20

Wait, actually now >> is it Toby McGuire? I don't know. I'm all over the place. Who >> Who's Armas? >> Oh yeah. Honor Day Armas. >> Honest.

2:48:30

But who's who's the guy in the top left? Uh Jake Gyllenhaal. >> Jake Gyllenhaal. That's right. Not Tony McGuire. Tommy >> McGar. That's a famous guy.

2:48:36

Everybody knows that guy.

2:48:39

>> I got so tied up in >> Anyway, so the real cast of the movie Artificial is going to be Andrew Garfield, Monica Barbaro >> as Mir Maratti.

2:48:48

>> And then a guy named uh Ike Baron Holtz is rumored to be Elon. >> Yeah.

2:48:53

Who's playing Dylan Patel? >> Who's playing Rune? >> Who's playing Rune?

2:48:58

>> Rune is the real question.

2:48:59

>> Yeah, Rune is the real question. >> Who should play Rune? Who should play Rune?

2:49:02

Wait, so I wonder if Mark Zuckerberg is gonna like make an appearance in the film and they're going to have to cast him again.

2:49:06

Should they have uh the guy who played him in the social network reprise the role?

2:49:12

>> No, but he like hates Zuck now. >> Wait, who? >> Jesse Eisenberg.

2:49:14

He has beef with like in general. >> Why? >> I don't know. >> Oh, weird. Okay. Why? Uh uh very odd.

2:49:22

Anyway, uh should be fun.

2:49:22

Hopefully we can get a cameo. We can sneak in.

2:49:24

If you know someone, introduce us. This is >> chat.

2:49:28

Thank you for helping us out there.

2:49:29

Yeah, Junk Jelly Hill from Gold Rock. Jake, thanks, guys. >> It's good.

2:49:36

>> Uh, Kylie says, "I can't believe this is real." I can't.

2:49:39

>> I can't believe this is real.

2:49:39

I saw this and I was like, "This is not real."

2:49:43

>> President Trump dialed into Fox and Friends on Tuesday morning.

2:49:45

We're a couple days late and revealed his newest and truest motivation for brokering an end to the war in Ukraine.

2:49:49

He's worried he might not get into heaven after he dies.

2:49:53

"I want to try and get to heaven if possible," he explained.

2:49:55

"I'm here and I'm not doing well.

2:49:57

I'm really at the bottom of the totem pole, but if I can get to heaven, this will be one of the reasons.

2:50:02

So, >> I like whatever whatever your motivation is uh for peace, uh do it. Do it. >> It is solid.

2:50:10

>> Um >> uh so there was a debate. Break it down. >> Yes. Break this debate down.

2:50:14

>> Craig Weiss says, "Having kids before 30 is how you stay generationally poor."

2:50:20

>> A real a real takes there. >> Yeah.

2:50:23

So, surely perfectly put the timeline in turmoil.

2:50:26

Um the dude Richie the rich highlighted uh Warren Buffett uh net worth 154 billion. >> He had a kid at 22. >> Child at 22. Michael Dell 23. >> Let's go. >> Bernard Arno 26. Some absolute dogs. >> Jordy Hayes. >> Hes 26.

2:50:46

>> I just missed it by a year or two.

2:50:46

I think Davidson remind >> I will say I will say uh >> you can come on whenever.

2:50:51

Uh, kids are much more I I I uh when I started making money, I got into cars. >> Yep.

2:50:59

>> And when I started having kids, I realized that um they are a much more uh quite, you know, I used to think uh there there was kind of a meme growing up.

2:51:09

>> They're pretty close to cars in terms of >> people would say like the average American kid by the time they they're 18 is like a quarter of a million dollars. >> Yep.

2:51:17

But it feels like uh children feel like the carrying cost of like 10 supercars at times between school and >> Well, we have our first in-person guest of the show.

2:51:27

We got David Sen from the TVPN Ultra Dome live. >> See you. >> How you doing, man? >> Good to see you, man.

2:51:35

>> Finally in the Ultra Dome. >> Finally in person.

2:51:38

>> Been How was the drive? >> Five times.

2:51:41

>> It's a testament to how much I love you that you got me out of Malibu.

2:51:43

It >> was an hour and a half. >> Yeah, it's rough. >> I know.

2:51:47

I did speeder and I do it every day.

2:51:49

>> No, but it takes you an hour and a half at like Yeah, you go at like 5:00 in the morning. 5:30 in the morning.

2:51:52

This is a different story.

2:51:54

>> Anyway, >> this is I love this place though.

2:51:54

You guys have an army back here. >> We do have an army.

2:51:58

>> You have the crew, the interns.

2:51:58

Then I met the friends of the interns. >> Oh, yeah. We got everybody. Oh, yeah. We have intern friends. Lads, lads, lads. >> Lads studio. >> What is this? >> That's a chat.

2:52:09

So, if if they if they saying how handsome you look, you you'll see it.

2:52:14

People are talking about uh yeah we're the chat is still talking about uh this debate over whether you can become generationally wealthy while having a child before age 30. >> What do you think? >> Obviously. >> Yeah.

2:52:28

And they gave Michael Warren about Michael Shout out to Michael Dell Dell Technologies.

2:52:35

>> Don't say don't say too much because because that'll dox us.

2:52:40

>> No, actually that that might be an interesting place to start. flew to Austin.

2:52:43

I spent five hours with Michael Dell. Wow.

2:52:45

Like two weeks, two or three weeks ago. >> Yeah.

2:52:47

What was What was that like?

2:52:48

What'd you learn from him? >> You know what?

2:52:50

>> What What surprised you because you've already you've already done uh a full episode on him. You've dug in.

2:52:53

You probably spent days on his story.

2:52:55

You know most of the facts. >> It's not only that.

2:52:58

Like I listen to >> So I highly recommend every entrepreneur listen to his audio book because he actually narrates it.

2:53:04

And so I listened to it three Can you hear me?

2:53:05

I listened to it three times.

2:53:10

Uh, I listened to it three times and then read the book and then made the episode. >> Wow.

2:53:14

>> Um, there was like halfway through our conversation I said some I said something like I'm having a hard time rec reconciling like this paradox in my mind about between like your immen immense accomplishments and just like how normal you are.

2:53:27

And obviously he's a superlative and extreme person but he's unbelievably measured and controlled and kind and generous.

2:53:34

It's just like one of my favorite people I've ever met.

2:53:37

We got to get him on the Theo Vaughn podcast and see how he does there.

2:53:40

Uh I mean everyone in tech has been making the rounds and there's been multiple times when tech people have gone on Theo and they've had like kind of awkward starts like it took it took Zuck a minute to warm up.

2:53:52

There was like the caffeine thing.

2:53:54

Oh, you know I don't I don't >> I think this is the mistake that people in tech make and you're seeing it with a bunch of other podcasts.

2:53:59

They should definitely come here before anywhere else. This is not a joke.

2:54:02

It's like they see a number on a screen and they don't think about what that number represents and who that person is.

2:54:07

And like you can just read some of the comments on some of these like mainstream podcasts.

2:54:10

It's like this guy's 6 years old.

2:54:12

He's in his [ __ ] underwear in Missouri.

2:54:14

He's got like unemployed.

2:54:16

Like he's not the person you're just optimizing for the number when you should optimize for the quality of the people in the audience.

2:54:22

>> Yeah, that's for sure. >> It's insane to me.

2:54:24

>> Uh well, you know who's optimized for the quality of the audience? Colossus Magazine.

2:54:28

There's a new uh I love how you printed out while we wait for the print edition.

2:54:32

don't have the print edition yet, but you can still go subscribe.

2:54:35

Patrick Oanaughy has printed uh a a warehouse full this time.

2:54:38

Um have you had a chance to read through the Joe Lamont profile?

2:54:44

>> I I get I got Patrick's, you know, obviously like my brother.

2:54:46

I talked to him like every day.

2:54:48

There's actually an interesting um something we can talk about that that I called him about yesterday.

2:54:54

>> Um I actually think moving forward and I told him this yesterday and I told him again today.

2:54:57

I was like, if you give me more than a day to read it, we if on especially in these long form profiles. >> Totally.

2:55:04

>> I should do an that is good enough if you read the whole thing.

2:55:06

It takes like an hour and a half to listen to. So I read it.

2:55:08

>> I was thinking about printing the whole thing and I was like there's no way I can read this before the show. It's like 100 pages.

2:55:14

>> I read it and then I threw it into 11 reader and I listen to it on the way out. >> Oh, that's good.

2:55:18

>> And so it's like an hour and a half if you listen at 1x. Okay.

2:55:20

>> Um but I think that source material is so good.

2:55:23

>> I could definitely do an episode of Founders and I was like just tell me in advance.

2:55:27

2 weeks and then we can drop it.

2:55:31

But one of my favorite little stories from that is how so um Joe had sort of a non-traditional you know entrance into uh >> startups and business.

2:55:42

He was a Stanford undergrad.

2:55:47

>> So no more more traditional in that sense.

2:55:49

But but at the time that he was building um what was the software called?

2:55:53

Sales builder or something like that.

2:55:54

that. at the initial >> but his name the names for his >> they're so great enterprise software >> it's like building a microphone like it's called microphone >> and I mean you look at some of the companies here like new view GCE still secure Ignite or object store accept prologic raven flow it's just like

2:56:13

>> so there's two there's two things that stood out from the story one is there there wasn't a meme yet of of VCs being like I'm going to back the Stamford dropout they wanted to back executives that had experience and crazy access to talent and there were some early kind of garage success stories. So, he was kind

2:56:30

So, he was kind of following that.

2:56:32

Um, but the other thing is is bigger companies, Fortune 500 companies refuse to buy startups.

2:56:40

>> But then his pricing, did you see what he did with his like it's 100 grand and they're like, "No, we we don't buy from startups."

2:56:45

They come back, okay, we need it. No one else can do this. It's 300.

2:56:48

And it goes from like now it's a million, now it's 7. 5 million. Now it's $25 million.

2:56:51

And everybody kept saying yes cuz you're the best. He did.

2:56:53

But but the other thing is the why now for the business, >> the initial why now and why it worked, not like the business catalyst was that they started sending this was at a time where they would just send you a letter and said you're preapproved for this credit card.

2:57:07

You just have to fill this out. >> Credit card roulette. >> Yeah. He Yeah. Yeah.

2:57:10

So he would he signed up he he racked out half racked up half a million dollars of credit card debt >> and was was finally before he got his first customer still no VC would fund him at all.

2:57:21

He fin like he was basically like evaluating like bankruptcy like he had played it out in his head like it's going to be like seven years.

2:57:27

It's going to be really annoying but it doesn't matter if you have his calculus was like you you know how much debt you have going into bankruptcy actually like it doesn't going from a 100,000 to 500k.

2:57:40

It's not that much worse to go through.

2:57:42

>> You might as well do it do it the real way.

2:57:44

>> So he just went full send. >> Absolutely.

2:57:47

and then and then basically made it all back from like his basically first couple customers.

2:57:52

>> So I we have a mutual friend that we can't name >> uh that has been close and is basically Joe's been like a a mentor of his for a long time and actually like a funer of a lot of projects that he does behind the scenes.

2:58:02

So I've heard a lot about this guy. >> Yeah.

2:58:04

>> And what I like about people is just like the people that chart their own path and and and intentionally go in the opposite direction.

2:58:13

So everybody heard Buffett and Munger and they're like you should buy phenomenal businesses even if you have to pay more.

2:58:18

And so then everybody in the VM VMS like rollups they're like I'm going to buy the great businesses.

2:58:23

And what he realized and I heard the story like probably two years ago it's like well okay so I have a lot more competition for the everybody wants the great ones.

2:58:29

What about these shitty ones over here?

2:58:30

And he turned it into a math problem.

2:58:32

He's like okay this is your ARR.

2:58:34

This is your rate of attrition.

2:58:35

Like I know exactly how much I'm going to get out of this in the next six years.

2:58:38

They talk about it in the you know they they use different words that they talk about in the article.

2:58:41

He's like, "Okay, so I can easily buy this thing for two because I'm going to rip I'm going to pull out, you know, 15 over the next 5 years."

2:58:47

He just does it over and over and over again.

2:58:48

And then I heard, I don't know if it was in if they said it, but like, you know, he bought out his uh shareholders. He 100% of the company.

2:58:55

And what I heard is like he's been taking billions. >> Yep.

2:58:59

>> A year in dividends, his [ __ ] paycheck at the end of the year.

2:59:00

Like, how much money did you make this year? I don't know, like $1. 5 billion. >> Crazy. >> Yeah.

2:59:06

>> And no one knew who he was. >> Yeah. He's built a machine.

2:59:08

A machine behind the scenes. Yeah.

2:59:10

The other thing is going dark going dark for 25 years and basically until this.

2:59:16

>> Well, this is this, you know, this is like one of my fetishes is like these very silent family-owned businesses where they own 100% of them.

2:59:23

You'll never hear about them.

2:59:24

I've been lucky enough to have dinner with some of these people and they're just again >> Yeah.

2:59:29

I know you you you've talked to people that have written books about their family business. >> Yes.

2:59:34

>> And and haven't released them and you're like, "Please let me let me" And they're just like, "No." >> Okay.

2:59:39

So, I'll just tell the story.

2:59:39

Um, so there's this guy, he's 76 years old.

2:59:43

Okay, this is like one of my favorite dinners I've ever had and uh I had to be really careful here.

2:59:48

So he started working in the family business when he was six.

2:59:51

Okay, he's the second generation this family dynasty.

2:59:53

They own the original source of their wealth was one of the largest uh privately held shipping companies in the world.

2:59:59

And when I mean privately held owned by his dad and now owned by him and his brother like it's just like there is no shareholders, no board of directors. It's just the family.

3:00:06

I've now become close with the third generation and I get insight on how they run their the family is a business and the business is a family and the way they go about things is just really really intelligent and so we have this phenomenal dinner.

3:00:19

You look up his reported net worth and it's like it's still in the tens of billions and it's like under reportported by a factor of like four cuz then he tells you about like oh yeah I put money into this and I never sold and it's worth this.

3:00:29

There's just crazy stories and he's also like super charismatic, one of the best rock and tours I've ever like been around and just super and he's still on it 24/7 like works all the time.

3:00:40

>> And uh so there's two things that were funny.

3:00:42

The this has happened a few times where families will commission real writers, right?

3:00:47

They they commission actual biography and I have a copy of a separate family. It looks like the book.

3:00:54

It looks like a book you'd buy in a bookstore.

3:00:55

The only difference is there's no ISBN number on the back.

3:00:56

Oh, >> so like some of the books I have, no one else in the world has these books besides like the family and the author and everything.

3:01:04

>> Open AI would give you a lot of money for that.

3:01:07

>> So, so there's two two funny things that that uh that he told me.

3:01:10

Uh one that they commissioned one for his father and then one for him.

3:01:14

And I was like, "Dude, give me those books.

3:01:15

Like, I would do such a good job on them."

3:01:17

Um and he without hesitation, he's like, "Absolutely not."

3:01:21

Like that's out of the question. I go, "Why not?"

3:01:23

And he says something like, "I have no desire to educate my competitors." Mhm.

3:01:27

>> And just like real fast and then um he >> Leave it to me in the will. >> No.

3:01:32

And then no I I think so there's another founder >> the next generation.

3:01:36

>> No but there's another founder who's 78 and I spent four days with her her the CEO and the top 40 executives.

3:01:41

Uh they were going through um how do they they wanted advice on like how to transition from a founder company.

3:01:47

She's one of my favorite founders I've ever come across in my life and like to the next generation.

3:01:52

And so after talking to all the executives spending time with them they're like okay so what do you think?

3:01:56

And after 4 days I go, when she's dead, you're [ __ ] This is like there's not like you're not replacing her.

3:01:59

She's a singular individual.

3:02:01

You should like either sell the company or go do something else cuz like there's no transition here.

3:02:05

It's like what do you do when Michael Jordan retires?

3:02:07

Like I don't know, start rebuilding.

3:02:09

Um but he said something funny because they're obviously in shipping.

3:02:12

They now have real estate and they have finance and everything else.

3:02:16

And he's got also one of the biggest boats in the world.

3:02:18

And he had mentioned no but no he mentioned earlier though.

3:02:22

earlier though. So early in the conversation he's like yeah uh when you read a lot of these biographies he had read all the bio same biographies I had read like Daniel Lewig Aristotle Nasis all these like shipping guys he knew all of them they would also run their company headquarters from their private like giant boats right giant yachts

3:02:38

>> and so he he had mentioned earlier like yeah I spend you know part of the year I live part of the year on uh he said I thought he said boat he said ship so I was like yeah so like when like what month he's going to tell us I go so what what uh like when are you going to go back on your Oh, like how long you been in your boat for? He goes, "What are you

3:02:53

He goes, "What are you talking about?"

3:02:54

I go, "You said like earlier you live on your boat."

3:02:56

He goes, "I don't have a boat."

3:02:58

I'm like, "Is this some kind of joke?"

3:03:00

He goes, "I have a ship."

3:03:02

I go, "What's the difference between a boat and a ship?"

3:03:04

He goes, "You can put a boat on a ship.

3:03:06

You can't put a ship on a boat."

3:03:11

>> So, my favorite conversations are these kind of people.

3:03:13

And I I like obsessed with like older entrepreneurs cuz they're just so much smarter than like a 25-year-old [ __ ] startup founders.

3:03:18

It's like not even They're not even It's like It might be an alien. Yeah.

3:03:20

Yeah, it's just like these people are just they just they have so much more experience.

3:03:24

They have so much more knowledge.

3:03:25

Like I just can talk to them for hours and hours and hours. >> Makes sense.

3:03:27

Uh have you dug into the Trilogy software mafia?

3:03:29

Are you familiar with this? >> No.

3:03:32

>> So uh this was some secondary research I did out of Trilogy.

3:03:35

This is of course the the the the subject of that Colossus profile um on Joe Lamont.

3:03:40

Um the in the9s Trilogy was on such a tear.

3:03:45

They were recruiting out of MIT, Harvard, and a bunch of folks went through Trilogy and then went on to do cool things.

3:03:54

Eddie Ellie Seidman, the CEO of Tinder, was a Trilogy alum.

3:03:56

Cyrus and Nick Ganju, founders of Zachdoc.

3:03:59

Henry Ward, the CEO and founder of Carta. Crazy.

3:04:04

>> John Lily, former CEO of Mosilla, and now he's an investor at Greylock.

3:04:07

Um and Andy Palmer, the CEO of Tammer, former former CEO of Vertica, which was acquired major major angel investor, invested in Carta in part due to his connection.

3:04:18

Um there are 87 Trilogy mafia companies now on Crunchb.

3:04:22

So it's just like this very interesting I we we've heard that that Joe like keeps his team very tight, but clearly selects for extremely entrepreneurial people.

3:04:33

They don't often get poached to like other firms that are doing the same thing.

3:04:36

If you leave, you're probably going to start a company.

3:04:39

>> I I wouldn't want to make money the way he did, you know? But put that aside.

3:04:43

Like what I do admire like I'm like the most pro-American person.

3:04:46

Like obviously the son of Cuban immigrants.

3:04:47

Like I grew up, you know, meeting people that literally came to his country, fled communism on a raft. Yeah.

3:04:54

>> Like so I don't like outsourcing anything.

3:04:56

I've had tons of people try to advertise on the podcast.

3:04:57

They're like, I have this outsourcing thing.

3:04:58

I was like, I want to find a way to pay people more money, not less.

3:05:01

Like I want to work with the best people in the world, and those people don't work for $8 an hour. >> Yep.

3:05:06

But I do what the thing I respect about Joe is like he he doesn't give a [ __ ] what other people think and he's willing to have his own thinking and then even if everybody's going in one direction.

3:05:14

I was listening to your mischief uh on the way over here.

3:05:15

He had this great line where he's like if everybody's running fast to something they're probably running off a cliff. I was like that's a bar. That's a good bar.

3:05:23

And so I like that Joe's like, "Hey, you guys are all doing this.

3:05:25

So that means by default there's an there's an opportunity."

3:05:28

I called Patrick yesterday.

3:05:29

I was like, "It's kind of funny.

3:05:31

You know, there's not any other people more obsessed with podcasts than me, you, Patrick.

3:05:34

me, you, Patrick. We talk about this this stuff all the time and I was like everybody even now is still starting podcast and Patrick you know he has the highest value uh investing audience in the world and what I love about what he's doing is he's like oh yeah you guys are all going to go to the podcast I'm

3:05:49

going to write what is that 15,000word in-depth profiles but they're so much better >> like the Neil Meta one um who's the paradigm guy I read Matt yeah this one >> yep >> like and they're going to keep doing it and he he'll tell you in the group chat like some of the ones they're working on Wow, they're crushing. I just love this. I just love this.

3:06:07

>> It's a very special product.

3:06:07

And and I went through when we were doing the show and we didn't have guests or anything and we needed a lot of material.

3:06:11

I went through and looked at old New Yorker deep dives profiles. They're so good.

3:06:18

Old Bloomberg and old Forbes pieces and they're just not doing it at that level anymore.

3:06:23

There are some profiles, but the one profile that was written in prestigious magazine about Zuck is all about like did he have the right team in place?

3:06:32

like it's like a diversity like referendum on diversity and stuff like brotopia type take which is just not really the story like that's a very minor piece of the Zuck story in my opinion.

3:06:43

>> I'm glad you said the New Yorker part because every single time I do a deep research report on every single person I'm working on just to get an overview and it's always like you I I want to have written like a New Yorker piece.

3:06:51

One of my favorite you >> fantastic writers Jackson we read one on Mary >> Mer.

3:06:55

Yeah, the Mary Mer fantastic. >> I read that too.

3:06:57

So um Jackson D you guys should have him on there.

3:06:59

He's doing for the record. >> We have invited him.

3:07:04

>> He needs to call in today.

3:07:04

I don't know what he's doing.

3:07:07

>> Nick, send him an invite.

3:07:09

>> Have him join right now.

3:07:11