The Cerebral Valley AI Summit is right around the corner! To help you navigate the fast-evolving AI landscape ahead of the event, Newcomer Podcast is launching a special four-part series — co-hosted by James Wilsterman and Max Child of Volley. Get insider insights, expert analysis, and fresh perspectives on the trends shaping the future of artificial intelligence.
In this first episode, James, Max, and host Eric Newcomer dive into what it really means to be an AI agent — and explore how agentic AI could reshape the future of work and everyday life. From picking wedding outfits to writing code, they share personal experiences of agents in action and reflect on where this technology is headed next.
So — where is AI headed? In the second half of the episode, the trio revisits market predictions made by AI last November and puts them to the test using fresh data pulled by Deep Research. After a spirited round of forecasting, they return to their 2024 AI Fantasy Drafts to see whose lineup is raising, exiting, and, ultimately, leading in the race for AI dominance.
Our next episode focuses on AI's impact in voice and video, and may include a few more surprise games...
The 2025 Cerebral Valley AI Summit will be held in London on June 25th
Timestamps:
00:39 - Intro & the scaling wall reversal
06:13 — How we use Claude and Deep Research
08:45 — Agents are here for the web search
14:44 — Coding agents as the breakout tool
24:24 — Update on last year's AI predictions
36:51 — AI Fantasy Draft
00:00:00
This episode is brought to you by Forethought.
00:00:03
Most companies build their customer experience in pieces.
00:00:06
Sales in one system, support in another, and onboarding
00:00:09
somewhere else. Forethought brings it all
00:00:11
together. Forethought is an AI system
00:00:14
made-up of advanced agents that handle sales, onboarding,
00:00:17
support, and retention. Each team manages its own
00:00:21
agents. The customer sees one unified
00:00:23
experience. Forethought powers over a
00:00:26
billion interactions every month for brands like Scale AI,
00:00:29
Cohere, Air Table, and Upwork. Learn more at Forethought dot
00:00:34
AI. Hey, it's Eric newcomer.
00:00:41
Welcome to the Cerebral Valley podcast.
00:00:44
Our occasional detour from the newcomer podcast with me are are
00:00:50
now three time hosts Max Child and James Wilsterman, the Co
00:00:55
founders of Volley and the Co hosts of the Struble Valley AI
00:00:58
Summit. Welcome to the podcast, guys.
00:01:00
Thank you, glad to be here. Eric, happy to be back, excited
00:01:04
about the conference coming up in London.
00:01:06
Yeah. So we, you know, always sort of
00:01:08
jump into the Struble Valley podcast ahead of our Tribal
00:01:11
Valley AI summits. And we've got one June 25th in
00:01:15
London. People are like, oh, isn't it
00:01:17
hard to do an international conference?
00:01:19
And like, like everything in startups, it's easiest to do
00:01:22
hard things when you underestimate how difficult it
00:01:25
is until you're like, oh, yeah, we're just doing it.
00:01:27
Someone is figuring out taxes. We have an event team there.
00:01:31
I saw we have the ticket prices in pounds on the website so I.
00:01:34
Think exactly. Like someone was on top of that.
00:01:37
So a couple fun things for this episode for the long time
00:01:40
listeners. At the end of the podcast, we
00:01:43
will return to the startup draft, perhaps my favorite part
00:01:48
of the show. We have overtime accumulated
00:01:51
startups in our imaginary portfolio and get to score keep
00:01:55
how we're doing working backwards.
00:01:58
Before that, we're going to dig into some of our predictions
00:02:01
from the last series. This is sort of a mid year
00:02:04
snapshot, right guys, in terms of the predictions that we made?
00:02:07
Yeah. We made about 10 predictions
00:02:10
last year in November with the idea that they would come to
00:02:14
fruition within a year. So this is the mid year check
00:02:17
in, no official scorekeeping needed, but we will obviously be
00:02:22
competitive on our mid year check insurance as well.
00:02:25
In the latter half of the show, we'll be doing our, our games.
00:02:28
But I just wanted to start off, you know, we've been doing this.
00:02:32
We launched the first Cerebral Valley in March 2023.
00:02:36
ChatGPT had just come out and everyone was getting their heads
00:02:40
around models Max and James, if you want to sort of walk me
00:02:42
through, how do you think of sort of the thematic evolution
00:02:46
of AI in that period? And where?
00:02:48
Where are we now as we're programming for Cerebral Valley
00:02:51
London? Last time we chatted we it was
00:02:55
the end of last year, which feels like a long time ago in
00:02:57
the AI world. But we were all having this
00:02:59
discussion of if we were hitting a scaling wall and if the models
00:03:02
would stop getting better. And I would say it feels like
00:03:05
every two weeks since then, something amazing has happened
00:03:08
in AI. Is pretty trivial.
00:03:11
Has been a great leap forward. I actually think the theme of
00:03:13
hitting the scaling wall aged really poorly, which we somewhat
00:03:17
predicted at the time. But you know, maybe we didn't
00:03:19
know how wrong we were. I mean, just some highlights.
00:03:21
We had, you know, GBT 01, the sort of first thinking model.
00:03:25
We had O3, which has brought thinking models into the tool
00:03:29
world. We've had Gemini really take
00:03:31
great leaps forward and kind of become a state-of-the-art model
00:03:33
system. We just had Claude 4-3 weeks
00:03:36
ago, which a lot of people consider the greatest model
00:03:37
built so far, maybe the best coding model ever made.
00:03:41
We've had incredible Leafs in image generation from mid
00:03:44
journey, Gemini Flux, a bunch of other folks, and then of course
00:03:49
we're going to this episode 2. Video generation has also been
00:03:52
unbelievable. VO3 in particular, which is
00:03:55
Googles new video model I think is that the first sort of truly
00:03:59
realistic seeming video generation model in my opinion
00:04:02
and and kind of a great leap forward there.
00:04:04
You know, Alexander Wang at our conference in November was
00:04:08
probably the most prominent person arguing for a scaling
00:04:12
wall or a potential issue. I mean, he was talking his own
00:04:16
book. You know, he's in the post
00:04:17
training business. I do think there's an argument
00:04:20
that a lot of the progress, some of the progress, some of the
00:04:23
progress we've seen has been about post training.
00:04:26
The models are a certain level of smart, but then they sort of
00:04:30
talk to each other, they're corrected in certain ways and
00:04:33
that's where they get more intelligence.
00:04:35
That seems to be the improvement we're seeing with O3 and the
00:04:40
Chain of Thought models. Yeah.
00:04:41
I mean, I think, I think Alexander Wang, I give him a lot
00:04:44
of credit for what he was saying at our conference in November
00:04:47
because he was arguing that maybe we would see a scaling
00:04:51
wall to some degree in pre training, but we wouldn't see
00:04:55
performance levelling at all. And I think that's exactly
00:04:59
what's happened. And to your point, like a lot of
00:05:00
the gains from thinking models have come from new post training
00:05:05
techniques, reinforcement learning and new a whole new
00:05:08
scaling paradigm I guess. Yeah, I mean, in the end, the
00:05:12
models have gotten really, really effing good.
00:05:14
And whether it's pre training or post training, they go good.
00:05:16
Yeah. Yeah.
00:05:17
I feel like the vibe was a little more like maybe this is
00:05:20
the end of AI progress, not just like pre training is over.
00:05:22
And to your point, there was a distinction drawn by some of the
00:05:25
people at the conference, but I think there was a maybe a more
00:05:27
negative tenor going forward from a lot of folks on stage.
00:05:30
And instead, I think it's been maybe the craziest 6 months in
00:05:33
the history of AI. I don't know.
00:05:34
It's felt like that to me. The interesting thing is, I
00:05:37
guess like in January, we had that DeepSeek moment, right?
00:05:40
We haven't talked about that. I forgot.
00:05:42
About DeepSeek. Too.
00:05:42
Yeah, yeah. Wow.
00:05:44
So to some degree that was maybe echoing some of these scaling
00:05:48
wall concerns because if you can get such high performance out
00:05:54
of, you know, open source models that are effectively competing
00:05:58
with the frontier U.S. companies, like maybe there is
00:06:01
an argument there. But then I guess, you know,
00:06:05
that's kind of died down a bit as we've seen both anthropic and
00:06:09
open AI come out with superior models to some degree.
00:06:12
What models are you guys using? Applaud for Gemini 2/5 for
00:06:16
coding and then for my own personal sort of research and
00:06:19
note taking and stuff. Probably O3.
00:06:21
That's the same for me. I've been using a lot of deep
00:06:23
research through ChatGPT. I think that's my favorite
00:06:27
product maybe of the last few months.
00:06:29
I don't have time. I use O3 a lot.
00:06:32
I use deep research some. I don't want to blow anyone up
00:06:34
here, but I I had deep research, right?
00:06:37
A whole like political consulting memo for somebody I
00:06:40
was trying to get to run for office.
00:06:42
You know, it's like it's amazing.
00:06:44
I mean, there is a world where I would have gone and paid a
00:06:47
consultant to be like draw out for me, like when races will be
00:06:50
available and when, when they could run.
00:06:51
And it's like you just like, oh, in the morning you're having a
00:06:54
manic fit and you're like, oh, let's, let's see what Chachi BT
00:06:57
can do. It's it's insane.
00:06:59
It's crazy. This is a very nerdy, lame use
00:07:02
case, but I like to buy cheap wine that is still good.
00:07:06
And there was a secret wine from a local provider and they said
00:07:10
it's this anonymous secret wine that we're selling for 1/5 of
00:07:14
market price. But you know, if you bought it
00:07:16
at market price, it would be a 250, three, $100 wine.
00:07:19
And they were like, it comes from these amazing vineyards and
00:07:22
the West side of Napa and the foothills, blah, blah, blah,
00:07:24
blah, blah. And I just pasted the
00:07:26
description into deep research and I was like, figure out what
00:07:28
the secret wine is like 20 and 20 minutes later comes back.
00:07:32
It was like, obviously this is like this BV Latour 2022 cab or
00:07:37
whatever, and I was like, what? I've been super impressed with
00:07:40
ChatGPT multimodal for shopping. The three of us are all going to
00:07:45
the same wedding in France in a month or so, and I don't know if
00:07:49
you guys have looked at the required attire, but they're
00:07:53
half of. It Oh my God, my wife is very
00:07:55
concerned. Yeah, this hot tip is paste that
00:07:58
whole thing in a ChatGPT, ask it to shop for you.
00:08:00
It'll be great. My wife and I.
00:08:03
This was a good idea. I mean, it's not perfect, I
00:08:05
won't lie. Like how like maybe one out of
00:08:07
10 suggestions are like completely off base, but the
00:08:11
shopping integration, it just, it's kind of showing where
00:08:14
things are headed like, you know?
00:08:15
What do you mean integration? Well, because it's actually like
00:08:19
searching the web, you know, doing an agentic workflow of, of
00:08:23
looking for these items and then it's pulling that information
00:08:26
back into Chachi BT in line in the chat.
00:08:29
There's links that link out, right?
00:08:31
Eventually I'm sure you'll be able to just like add that to
00:08:34
your cart within Chachi BT It's a full shopping experience.
00:08:37
It's not just researching. Yeah, interesting.
00:08:40
He used the word. He used the word agentic.
00:08:43
Yeah. OK.
00:08:44
I mean, yeah, we're going to talk a lot, you know, over the
00:08:46
next couple episodes in terms of how AI can get put to use.
00:08:51
The topic I really wanted to get into this week before reviewing
00:08:55
our games and our scorekeeping is agents.
00:08:58
Like I feel like agents have been at once sort of the
00:09:02
buzziest thing in the backdrop of a couple events, but and I
00:09:06
never quite here. And so I guess the first direct
00:09:10
question I want to ask is, is deep research an agent?
00:09:15
Is O3 an agent? Like what, what?
00:09:17
What is an agent these days if it's just delivering you a
00:09:21
report? I buy the agent definition that
00:09:25
an agent is, you know, an AI tool that can actually do stuff
00:09:29
for you, right? That can go through some sort of
00:09:31
series of steps involving actions and you know, quote UN
00:09:36
quote tool use is sort of one of the popular phrases these days
00:09:39
of, you know, using different tools.
00:09:42
Currently the only tools these agents can use really are
00:09:45
essentially like web search and you know, maybe pulling shopping
00:09:49
links and showing pictures to James's discussion.
00:09:53
But I still fundamentally think like there is a big difference
00:09:55
between, you know, ChatGPT of 6 to 12 months ago where right,
00:10:00
you know, you ask it a question, it gives you an answer.
00:10:01
Essentially it's you know it's a text in text out engine, right?
00:10:05
You put text in one side, it gives you text out the other
00:10:07
side, but it doesn't go do stuff that's.
00:10:09
Part of that kind of web it's. Like, yeah, yeah, yeah, exactly
00:10:12
right. Even the deep research that we
00:10:14
were talking about, I mean, I used to be a management
00:10:16
consultant for two years, 2 horrible years, and it can do
00:10:21
much better versions of what I did as a management consultant,
00:10:26
you know, in a matter of minutes, it could do 5 days of
00:10:29
world class management consultant level research on a
00:10:32
topic, right? And I guess if you don't think
00:10:35
that's agentic, I think you're a little crazy, like you're too
00:10:40
high, right? It's clearly running around.
00:10:42
I I've seen some people use like time is it's like if it takes
00:10:46
time, how long it takes, you know, if it's all going and
00:10:48
doing things and interacting with the world and coming back,
00:10:51
I want to throw down a gauntlet to me.
00:10:55
And this could come soon. We'll we'll be in the world of
00:10:58
agents once people are letting them run wild with their own
00:11:02
credit cards once, once agents are spending money without a
00:11:06
human check in, that's when we've got sort of real agents.
00:11:10
What do you think? So does it.
00:11:11
Not count in your mind if like my agent finds me a dope pair of
00:11:15
shoes and, you know, text me, hey, can I buy this?
00:11:18
And I'm like, yeah, go for it. No, it needs to transact.
00:11:20
It needs. It needs to do without you.
00:11:22
Oh, truly? It'll be like this.
00:11:25
My great. Thing that's like that's an
00:11:28
agent. I mean, I understand like you
00:11:30
know, you, you book flights and an agent would have like asked
00:11:33
you before that. Definition I I feel like if it
00:11:36
does a restaurant reservation, maybe no money changes hands.
00:11:38
That's still an agent. I mean, I I just think this.
00:11:40
Is it's definitely an agent. I I'm just saying that's like a
00:11:43
great employee solves the problem, right?
00:11:46
It's not like, oh, they come back to you and want all this
00:11:48
feedback. It's like do the thing like pull
00:11:50
the trigger like once we can trust to do that.
00:11:53
I'm just saying that would be a landmark moment that I don't
00:11:56
think is so far away. Or do you think that is far
00:11:58
away? No, I, I honestly think that if,
00:12:01
if you had the capability to do that already in ChatGPT, like
00:12:05
James did his clothing research, probably some people would be
00:12:07
like, yeah, fine, go do it. Like don't buy too many things
00:12:09
without asking me. But like, I think people would
00:12:11
already be down. I honestly think the company's
00:12:13
reluctance is probably just mostly like they don't want it
00:12:16
to go off the rails and spend, you know, thousands of dollars
00:12:18
of people's money or whatever. But it's it's possible already
00:12:21
in my opinion. I don't know if it would work
00:12:23
yet. Like, and I think a lot of this
00:12:25
ties into context how much Chachi BT knows about me.
00:12:29
I mean it it's obviously, you know, starting to build that
00:12:32
memory, but I don't think it knows enough without me
00:12:34
prompting it or, you know, having very targeted list of
00:12:38
shopping ideas for this wedding to go just start buying me
00:12:41
stuff. I that'd be super interesting
00:12:43
once we get there, but I don't think it's ready for that yet.
00:12:47
Max just was sort of getting at this.
00:12:49
I mean, stop limits I think are key, right?
00:12:51
I mean, there's a degree to which algos, you know, I was
00:12:54
talking to a former banker about this the other day.
00:12:55
Like, you know, it's not crazy that we would let a machine, a
00:12:59
computer make payment decisions on its own.
00:13:02
Traders do it all the time. You just create some limits and
00:13:05
checks and hopefully have people hovering.
00:13:07
But like the algorithms have to move before a person could
00:13:10
react. And so that's happening there.
00:13:13
And so you can see with LLMS, it's like, OK, you build up a
00:13:17
trust up to the, you know, $500 limit.
00:13:19
And you're like, you can, yeah, it'll be interesting.
00:13:22
I mean, how many chargebacks are our chargebacks can go up with
00:13:25
everybody releasing power to agents?
00:13:27
And you're like, oh man, we need to do more refunds when the
00:13:29
agents do something crazy. I mean, I think that kind of
00:13:33
gets into some of this like quote UN quote agentic web
00:13:36
discussion, which is like, can we redesign the web in a way
00:13:38
that enables more of this behavior?
00:13:39
Because I do actually think if you're a clothing retailer,
00:13:43
there probably is some kind of business model in which you let
00:13:47
people buy way too many clothes and then they send most of them
00:13:50
back or they cancel most of them before they ship or whatever.
00:13:52
Right? Like, which obviously is along
00:13:54
the lines of like a Stitch Fix or Trunk Club or some of these
00:13:57
other companies which, you know, weren't wildly successful
00:13:59
because I think the return fees are pretty punitive or, you
00:14:03
know, and you end up with a lot of fraud and, you know, damaged
00:14:05
clothes and stuff like that. But I do think there's an
00:14:07
interesting question of like, could you build some sort of
00:14:10
consumer commerce website where people's agents can like buy way
00:14:13
too much stuff, like basically and then cancel it or return it
00:14:17
or limit it or some in some way because people buying too much
00:14:20
stuff by accident. And it's it's probably good if
00:14:22
you're selling things like even if you have to find a way to let
00:14:25
them cancel it or return it. My first reaction to what you
00:14:27
were saying is, oh, this is like a workaround to allow agents to
00:14:31
spend when we're really going to unwind it, but it makes the
00:14:34
default spending versus not. It's like, oh, the sugar is
00:14:38
good, which is which is really good for if.
00:14:40
You're selling stuff. Yeah, yeah, yeah, yeah.
00:14:42
James, what's the agent use case you're most excited about?
00:14:45
I can't ignore coding. We haven't talked about coding,
00:14:47
which I think is like the most actually valuable the.
00:14:51
Real one. That's why we want to talk about
00:14:52
it's boring. It's happening.
00:14:54
And that's. Something actually works?
00:14:56
Oh yeah, yeah, yeah, OK. AI podcasts are only about the
00:15:00
future, you know. Yeah, I think, well, I just
00:15:04
think that there's a lot to unpack about the future of the
00:15:07
of coding. I mean, I am a CTOI code when I
00:15:11
can and this agentic world has dramatically changed what I can
00:15:16
do in terms of prototyping and participating in the coding at
00:15:21
Volley and learning faster, right?
00:15:24
I mean, you just learn so much faster about different
00:15:27
technologies and tech stacks. And yeah, I think it's like if
00:15:31
you're not an engineer, you maybe understand this a little
00:15:33
bit or you've played around with things that are a little bit
00:15:36
more. No code like lovable.
00:15:37
Or figma or. Figma right?
00:15:39
I mean I. Think both Lovable and Figma are
00:15:41
speaking our event and they are now competitors in the the no
00:15:45
code world. Yeah, both non coders and
00:15:48
coders, if you've dabbled with any of this, you are, you know,
00:15:52
receiving the future of like what could happen to all types
00:15:55
of computer work, white collar work, if you want to call it,
00:15:59
but is happening first in in the engineering space and it's it's
00:16:04
pretty remarkable. I think Nat Friedman had a
00:16:06
really good analogy that I can't get out of my head about agent
00:16:09
tick coding. And he talks about the idea that
00:16:11
you have, you know, you have this room full of interns who
00:16:16
are all like junior engineers basically, right?
00:16:18
And you assign each of them a task and they go off and try to
00:16:22
do it. And then when they get stuck,
00:16:24
they raise their hand and they say, hey, I need help here,
00:16:27
come, come help me. I'm, I'm stuck with this bug or
00:16:29
or this this, you know, issue the, the app's not working,
00:16:31
whatever. And he podcasted, I think about
00:16:35
six months ago and he was like, right now basically these
00:16:37
interns raise their hands like every 5 minutes in like human
00:16:40
time. Like so they do like 5 minutes
00:16:42
of work and they raise their hand and you've got to go help
00:16:43
them. And then they raise their hand 5
00:16:44
minutes later and you know, over and over and over again, he's
00:16:46
like, so you can't really have met that many of these, you
00:16:48
know, imaginary interns going because you can just become, be
00:16:50
running around fixing their problems all the time.
00:16:52
You know, maybe only one really like because you're just
00:16:54
constantly having to give them feedback.
00:16:56
I do think like the frontier for like how much quote UN quote
00:17:00
human work these agentic coding tools can do now without you
00:17:04
having to run over and help them when they raise their hand is
00:17:06
like probably somewhere in the like 15 to 30 minute range now
00:17:10
where like, you know, they obviously come back to you very
00:17:13
quickly. Like they iterate through their
00:17:14
work very fast because they can type at a superhuman speed,
00:17:17
right? They can put out hundreds of
00:17:18
lines of code. But the sort of amount of human
00:17:21
level work I would say in my testing that they get stuck is
00:17:24
probably like, you know, some of that 15 to 30 minute range.
00:17:26
I'm confused, like are you saying 15 to 30 minutes of like
00:17:29
what it would take for a human intern?
00:17:31
What it would take like a good human engineer like, you know,
00:17:34
Yeah, I got it. You know, mid to senior software
00:17:36
engineer just just hammering away at code, right?
00:17:38
Yeah. Like how much code do you get
00:17:39
out between them getting stuck? Right.
00:17:42
Like, yeah. I would say yeah, maybe half an
00:17:43
hour of like human, human work, but.
00:17:45
You get it. You get it in 3 minutes or two
00:17:47
minutes you get. It in a minute, you know, or 30
00:17:49
seconds usually, which which is which is mind boggling.
00:17:51
But like the dream is you get, you know, 4 hours of human human
00:17:55
work or 8 hours of human work or eventually, you know, weeks or
00:17:58
just starts fixing itself and you never go in there, right?
00:18:01
I mean, you know, I spent a fair bit of time vibe coding a month
00:18:05
or so ago. I do think there are some self
00:18:09
driving car aspects in the sense that like the last 5% or 10% of
00:18:15
a problem is very important. And like it's like, oh, it looks
00:18:19
close. It looks close, but it's like,
00:18:21
sure, it's really good when I'm like copying the sub stack
00:18:25
design, dropping a lovable like, oh, rebuild that.
00:18:28
But like at some point I feel like you've just been working
00:18:31
long enough and you have some minor tweak you want to make and
00:18:35
it just starts getting stuck and has no idea.
00:18:37
I mean, partially that's why I'm trying to do no code and no code
00:18:41
knowledge. But I do think these programs
00:18:44
are really good at being enticing in the beginning, and
00:18:47
then they sort of get overwhelmed as the project
00:18:50
starts to expand. Maybe this project that you were
00:18:53
doing, you won't be able to ship because of that last five mile
00:18:56
problem or whatever. The fact that you even started
00:18:59
it is kind of amazing that you did that.
00:19:02
Like you wouldn't have been able to even start the project three
00:19:05
months ago or six months ago or something.
00:19:06
Sure. Yeah.
00:19:07
So I don't know, it's just kind of interesting, like where that
00:19:10
heads is. You know, I think eventually you
00:19:12
get to the point where it's a finished product and then
00:19:15
there's just like way more coding projects happening in the
00:19:17
world, right? And then, you know, I think
00:19:19
within companies that already have an engineering workforce
00:19:22
like it actually can get across that that last mile.
00:19:25
I think another important theme that always comes up is like if
00:19:30
we froze this moment in time, how much value is there that
00:19:34
people are still understanding versus depending on continued
00:19:40
progress in the models that's that's revolutionary.
00:19:42
Where are you guys on how much harvesting could be done with
00:19:46
what would already exist versus what we're waiting on?
00:19:50
I would say a lot like a lot of harvested progress.
00:19:54
Like I can't even think like trillions.
00:19:56
I don't know. I haven't decided of 10s of
00:19:58
trillions, but let's say trillions.
00:20:00
What's the measures? In value.
00:20:01
Oh, like GDP. GDP value of just harvesting the
00:20:05
stuff that we've done so far. But it's amazing you still see
00:20:09
people who are like, skeptic. I mean, I don't know.
00:20:12
I just don't want to be, you know, deluding myself here.
00:20:15
Yeah. I mean, I think like
00:20:16
fundamentally we've all had the experience, right?
00:20:19
Where something that used to take 8 hours now takes less than
00:20:25
a minute, right? I mean, like we've all had that
00:20:27
experience, right? And so if you just do the math
00:20:29
on that, you say, OK, I took something that used to be a 500
00:20:31
minute problem and now it's a one minute problem.
00:20:34
Unless you believe that thing had no value whatsoever, how
00:20:38
could you not see a 500X in knowledge work in front of you
00:20:41
and say, hey, maybe we haven't harvested all the value on this
00:20:45
500X improvement, though this thing I just did.
00:20:48
How could you not believe that that's going to have insane
00:20:50
ramifications throughout our world and and insane amounts of
00:20:53
value that eventually can be created.
00:20:55
You know, so I don't know. I just to me, I just look at
00:20:58
that, you know, 2 orders of magnitude differential we're
00:21:00
already seeing on labor and think it's got to be huge.
00:21:04
Here's a question for you guys. Like, I have a friend who's in
00:21:06
PE and he, you know, doesn't really, he's not really in the
00:21:09
tech world that much, but he, you know, tried chat GPD didn't
00:21:13
really, you know, find it super valuable.
00:21:15
He had never tried O3. And I just showed him what was
00:21:19
possible and he, like, you know, kind of changed his whole
00:21:21
opinion of what types of companies he should buy in the
00:21:24
future. Wow.
00:21:26
Well, I do think that's a real danger that there are, there are
00:21:29
things that I'm changing about my life and medical decisions
00:21:33
and lots of stuff off O3 BS. It's so persuasive that we'll
00:21:37
never really be able to back out the psychological.
00:21:39
And, you know, it's all right. It's like, is it having an
00:21:41
effect? It's like, yeah, it's deeply
00:21:43
affecting big decisions in my life.
00:21:46
You know, just because it's like the thought partner, it's there
00:21:49
just like, you know, if you were to have the friend that's there
00:21:51
while you're soundboarding an idea, like that's going to be
00:21:54
dramatic. And I think just the fact that
00:21:56
it's in the loop. Wouldn't you say overall that
00:21:58
it's in person giving good advice?
00:21:59
That's why I keep it in the loop.
00:22:01
Yeah, right. It's not like, but but it will
00:22:03
be hard to know if it if it hallucinated.
00:22:06
I don't know. It's like, is it gonna?
00:22:08
Yeah, hopefully in a couple of years I'll chase after me in
00:22:10
this shot. GPT will be.
00:22:13
I told you a couple of key things that I think you took to
00:22:15
heart and and now that I'm a little smarter, I'm gonna go
00:22:18
back on that. Yeah, we'll see.
00:22:20
Having lived through the iPhone experience of of having the
00:22:23
first iPhone in 2007 and having to spend the next four years
00:22:27
explaining to my friends why they had to get an iPhone and
00:22:30
why they. Should probably.
00:22:31
Consider getting an iPhone. I mean James, you had a
00:22:34
BlackBerry until what, 2011 or something like that?
00:22:36
Like 10/20/11 like. James, I'm sorry, I was the
00:22:42
lager to. I had a BlackBerry for a long
00:22:44
time. Yeah, and people would be like,
00:22:45
well, Max, what? Why?
00:22:47
Why do you, I need an iPhone. What do you do with an iPhone?
00:22:49
And I'm like, well, you can like browse the Internet.
00:22:51
And they're like, well, you know, I browse the Internet on
00:22:53
my phone. It's fine.
00:22:54
And I'd be like, well, these apps are pretty cool.
00:22:56
And they'd be like what? Like the the app that makes it
00:22:58
look like you're drinking a beer and I'm.
00:22:59
Like, well, I'm not. Using that every day, but just
00:23:04
the last point I have to make on this BlackBerry thing.
00:23:05
I'm sorry, BlackBerry, which obviously is a dead company
00:23:09
today. No one you know has a
00:23:10
BlackBerry, right? Their sales continue to grow for
00:23:14
four years after the iPhone was released right?
00:23:17
So O 708-0910 eleven BlackBerry sales continue to increase even
00:23:21
though we all today look back and even in the movie about
00:23:25
BlackBerry, they pretended that the day the iPhone came out it
00:23:28
was like P BlackBerry. Sorry guys, you missed the
00:23:31
future. And I'm just pointing out that
00:23:34
in the real world, sales continue to grow for the next 4
00:23:37
years. And so I'm just saying that when
00:23:38
you're in one of these sort of Roadrunner, Wiley Coyote chases
00:23:42
the Roadrunner over the ledge moments where you take a second
00:23:45
before you look down and realize that gravity is pulling you
00:23:47
there. I think similarly with AI, we're
00:23:50
all like, holy crap, we just went off the ledge.
00:23:52
Like shit is about to get really dramatically different here.
00:23:56
And you can still stand there in the air for a year, two years or
00:24:00
three years or four years before gravity really hits, right.
00:24:03
And and I think to your initial question, like we are in that
00:24:07
period where even if nothing changed, like we're already off
00:24:11
the ledge, you know, stuff, stuff is going to be
00:24:14
dramatically different. No matter what happens from
00:24:16
here, which I know we all believe, you know, the models
00:24:18
are going to continue to get better.
00:24:20
So the pace of acceleration is going to be even higher.
00:24:22
It's a perfect endnote. Let's move the conversation to
00:24:27
our predictions from six months ago.
00:24:30
Basically, we're not going to get 2 in the weeds.
00:24:33
It was a fun discussion. James, do you want to take us
00:24:37
through them, What the question was, where we each landed and
00:24:40
then we'll give a quick reaction and then go to the next one.
00:24:43
Sure. Sounds good.
00:24:45
Just to clarify also we back in November, we asked Claude and
00:24:49
Chachi BT to generate these predictions for us including
00:24:53
providing probability estimates of how likely they were.
00:24:57
And then so our job was to take the over or the under on each
00:24:59
prediction. The first one was open AI shifts
00:25:02
GPT 5 with a greater than 10 trillion parameter model Max was
00:25:07
the over, Eric under and I took the over.
00:25:10
Thoughts. Where is it?
00:25:12
GPT 5 here. Yeah, so far.
00:25:14
So far I'm correct. Six months.
00:25:16
I mean, I think we much hung up on they wouldn't use the name or
00:25:20
maybe there was a high chance that they would abandoned the
00:25:22
name. It seems like they're going to
00:25:24
do it honestly right now. What the vibes are that they're
00:25:27
going to kill the O series and just make 5 the overall?
00:25:32
I feel super good about the name and it's shipping this year.
00:25:35
At this point. I think the only thing that
00:25:37
could hit the under would be the 10 trillion parameters because
00:25:40
didn't Brad, like at the CEO literally say they were going to
00:25:42
ship GPT 5 this year and it was going to be called GPT 5?
00:25:45
Like I'm pretty sure he said that like last.
00:25:47
Seems like a. So I would still smash the over
00:25:50
the 10 trillion parameter. Thing is, I guess the one.
00:25:52
Certainly I would buy the over now.
00:25:55
I might take the under because of that 10 trillion like I oh.
00:25:59
You're I'm sticking to my bet. Whatever I we're the best locked
00:26:01
in. Yeah, I can't be wrong.
00:26:03
I was trying to figure out like so Claude 4 Opus just launched
00:26:06
and there's no comment on the parameters that I can find, but
00:26:11
the best estimate I could get from Claude was 2 to 3 trillion,
00:26:15
so I don't know if that's nice. Nice.
00:26:18
I mean a lot of the improvements seem to be post training and
00:26:22
chain. Of thought and if they're.
00:26:24
Going to merge, you know, the O series with GPD 5.
00:26:28
It's partially because they think they need part of the
00:26:30
improvement off these reasoning models, and so maybe it's a sign
00:26:33
that the parameters aren't getting as big as we thought.
00:26:36
Also also it seems like adding that many parameters you know
00:26:39
creates huge issues with inference cost and just serving
00:26:42
them it's. Expensive.
00:26:44
Yeah. Anyway, the next one was three
00:26:48
or more countries enact national rules regarding AI medical
00:26:52
diagnosis. Max, you took the under, Eric
00:26:56
took the over, and I took the over.
00:26:58
We had it at a 70% probability. You guys are crazy.
00:27:02
You guys are crazy. There's no way you think in the
00:27:04
next 6 months three countries are gonna enact regulations.
00:27:08
That happened so far, medical diagnosis.
00:27:10
Here's what I had. Oh, you.
00:27:14
Found something chachi BT is saying.
00:27:16
Yes, EU AI Act. Exactly, Yeah.
00:27:19
UKSMHRFUS to regs online chachi BT things we're like in good
00:27:24
shape all. Right.
00:27:25
Well you took 70% so never forget that I get 2 to one.
00:27:28
Yeah. I don't know how you're going to
00:27:29
do the overall math at the end, but I'm sure whatever.
00:27:31
Onward and upward. All right, Tesla full staff
00:27:34
driving approved for unsupervised driving in one or
00:27:37
more US state, 40% probability. Max had the over, Eric had the
00:27:43
over, and I had the over. I think we're in good shape
00:27:46
there, right? Isn't Texas gonna happen like
00:27:49
yeah in June, supposedly. Like, yeah, in two weeks,
00:27:52
correct? But it does seem like right?
00:27:54
Chachibi says not yet Pilot cars on private roads in Texas.
00:27:57
No public permit. It's coming to Austin in June.
00:28:00
Now. I think this gets into the
00:28:02
letter of the prediction, though.
00:28:04
Is it a state law or city law, whatever.
00:28:08
OK, we'll come back to that in a few months.
00:28:11
AI will write the copy for greater than 50% of a major news
00:28:16
outlets articles. 30% chance Max had the over, Eric had the over
00:28:22
and I took the under. I just feel like my, my under is
00:28:27
great here. I mean, nobody's.
00:28:28
Nobody's. Claiming this nobody's first
00:28:30
newsroom. Everybody is like, what's the?
00:28:32
Eric's gonna do this by the end of the year.
00:28:35
Just to go build. A new media company.
00:28:37
I won't be major though. I don't think I'll I'll qualify.
00:28:40
You could be major by the end of the year.
00:28:41
Why not? Apparently, the Chicago Sun
00:28:43
Times inadvertently ran an AI generated book list filled with
00:28:47
errors, sparking backlash. Clearly not paying for 03 there
00:28:50
I think. I just think we're gonna see so
00:28:52
many of these things, like hallucinations, like, aren't we
00:28:54
already seeing this from the Trump administration?
00:28:56
Like just random things that don't make any sense.
00:29:00
I certainly don't want, you know, AI to replace the
00:29:03
newsrooms. I'm just expecting a lot of like
00:29:05
stories over the next year about like academic papers and news
00:29:09
articles just having obvious hallucinations, right?
00:29:11
Because people are going to be using.
00:29:13
But shouldn't we be more disturbed that the Trump
00:29:15
administration is like, I feel like if Democrats were in charge
00:29:18
and they were releasing government reports that were
00:29:21
clearly written by AI, yeah, it would be like the biggest
00:29:23
cultural story of the moment. It's like, well, the governments
00:29:26
already phoning it in with Trump, it's like, oh, at least
00:29:29
they're using it is. Literally what I was going to
00:29:31
say. I was like, I would honestly
00:29:33
prefer the AI to be doing this than the miscellaneous Trump
00:29:37
administration employee. Like knows nothing about
00:29:40
anything. I just want Sam Altman to give
00:29:42
every administration staffer free O3 so that they have O3
00:29:46
right. Reasonably intelligent fake
00:29:47
reports instead of real ones. I do think this core take that
00:29:51
the Internet split everybody's view and this is part of why
00:29:55
originally Marc Andreessen was so pro crypto and anti AI.
00:29:59
I do think AI is going to bring us potentially closer together
00:30:02
where people are asking rock like is this bullshit true or
00:30:05
false? I mean, it's possible people
00:30:07
build crazier models, but for now, while the models are sort
00:30:11
of generally in agreement with each other about how the world
00:30:14
works, it could be a major force for cultural consensus over the
00:30:19
next couple decades. It's basically the network TV of
00:30:22
the Internet essentially, right? We're all watching the three
00:30:25
major channels and they all broadcast kind of like exactly,
00:30:29
but not family friendly content. And your take is that Marc
00:30:32
Andreessen? Marc Andreessen doesn't want to
00:30:35
bring us together. He was weird, crazy shit.
00:30:37
Yeah, that's what crypto is all about.
00:30:39
Like, do whatever you want, like, and AI is a source of
00:30:42
conformity. But then Andreessen Horowitz
00:30:44
basically saw what was winning, and it was like, we not need to
00:30:47
go where the money is. And, you know, that's how I see
00:30:49
the story playing out. I mean, they were resistant at
00:30:52
first. Khosla led the open AI venture
00:30:55
round. Anyway, keep going.
00:30:56
Didn't Peter Thiel say like AI was communist or something?
00:31:00
Yeah. Yeah, communist and crypto was
00:31:02
libertarian or whatever. Right.
00:31:04
OK, Fully AI scripted and AI rendered feature film gets a
00:31:08
theatrical release 25% chance. Max with the under, Eric with
00:31:13
the over and James with the under.
00:31:15
This was this was Eric using his insider knowledge of pay to play
00:31:19
tactics within the movie industry to to try to grab us on
00:31:23
this that someones going to pay a theater chain to take their
00:31:25
movie even if it sucks. Basically, make sense?
00:31:28
I mean, we still have time. I had to double check this that
00:31:30
it wasn't hallucination, but according to chat GVT there is a
00:31:35
fully AI generated movie releasing soon, Pirate Queen
00:31:40
Zheng Yi Sao, billed as the world's first fully AI generated
00:31:44
feature film. Exactly.
00:31:45
That's going to get a festival run.
00:31:46
I gotta watch it. All right, moving on, more than
00:31:49
three major smartphone OEMs chip phones with AI Co processors
00:31:55
running 7 billion parameter plus models on device. 7 billion is
00:32:01
pretty high. The rumor mill on Apple is
00:32:03
saying they're going to be able to run 3 to 4 billion.
00:32:06
So even if you believe in James's claim that the chip
00:32:09
they've had in there for 12 years is an AI chip, it still
00:32:12
might not be able to do a 7 billion parameter model.
00:32:14
So we were specifying an Apple, Samsung, Google, Chami,
00:32:18
dedicated AI Co processors that run 7 billion perimeter LLMS
00:32:23
fully locally. We had the probability at 75% to
00:32:29
unders Max and Eric and I took the over and I yeah, I was
00:32:33
counting the the current technology as capable of running
00:32:38
those types of models. But yeah, to your point, Max, I
00:32:40
think 7 billion might be a bit high, right?
00:32:44
3 to 4 is what the rumor mills saying for this year, but we'll
00:32:47
know in a few more months I guess.
00:32:48
Next. Anthropic releases a model
00:32:51
scoring over 90% on U Bar, the unified benchmark for AI
00:32:57
reasoning, which does not exist according to our own research.
00:33:02
During the podcast recording last year, Complete Hallucinated
00:33:07
prediction from Claude. How are you feeling about that
00:33:10
one guys? Great.
00:33:13
I should, yeah. Certainly raises red flags,
00:33:16
yeah. Moving on to #8 / 4 Fortune 500
00:33:22
firms 5 or more cut greater than 25% of their middle management
00:33:28
roles by the end of this year, crediting AI explicitly with a
00:33:34
25% probability. Actually interesting.
00:33:38
Here we have an over and over from Max and over from Eric and
00:33:44
on under from myself. I'm feeling pretty good about
00:33:47
the under. Yeah, yeah, look, I'm.
00:33:49
Gonna do the research on this. We're only asking for five firms
00:33:52
to cut 25% of only middle management.
00:33:54
So that's a pretty that's a low bar.
00:33:56
According to my research with ChatGPT, this has not
00:33:59
materialized. Many large companies are
00:34:01
experimenting with AI and none have reported cutting 1/4 of
00:34:05
their management. Yeah, maybe we overestimated,
00:34:08
first of all how much we thought it would give them air cover for
00:34:11
all sorts of things. But yes, exactly.
00:34:13
Maybe we didn't want to get right into the AI narrative.
00:34:17
If we got a tariff induced recession, this actually might
00:34:19
happen so. OK #9 Deepmine and Google
00:34:22
discover a new drug that clears phase one trials within 2025.
00:34:27
We gave that a 20% probability and Max took the under, Eric
00:34:33
took the under, and I took the under.
00:34:35
I think all looking pretty strong here.
00:34:38
And D mine spun out isomorphic, which would be the start up.
00:34:42
I think that would potentially do this.
00:34:43
So there's a there's a chance that even if it happens, we can
00:34:47
all claim technicality that it doesn't.
00:34:49
But I I think it's, it's not looking likely, right?
00:34:52
They have to be in phase one trials already.
00:34:55
Yeah, I mean I. Think we?
00:34:56
Yeah, I think that they are planning to enter trials by the
00:34:59
end of this year, so unlikely to have completed phase one trial.
00:35:04
And lastly, we have the international AI treaty with
00:35:09
greater than or equal to 15 signatories, including three of
00:35:13
the US, China, EU and UK. 50% probability we all took the
00:35:19
under. Seems like a good bet so far.
00:35:22
Yeah, the AI believes too much in human institutions.
00:35:25
Right, because US, China both seem unlikely, right?
00:35:29
Is Europe getting EUUK viable getting the US which under Trump
00:35:35
is now like no AI regulation and China which is we do what we
00:35:40
want. The fact that the EU and UK get
00:35:42
separate credit here doing a lot of work, but three of the four
00:35:46
seems high do. You know, do you know what the
00:35:48
score is? So just pulling together the
00:35:50
scores, I asked Chachi PT to create a scoring system.
00:35:54
It came out with a formula inspired by Breyer style scoring
00:35:59
which I had never heard of but. Yeah, that's how close you are
00:36:01
to the probability. Yeah exactly.
00:36:03
Yeah, seems like a good scoring system.
00:36:06
It gives 3rd place to Max Dinged for betting against AI film
00:36:13
better calibrated on his conservative bets like hardware
00:36:16
and policy. 2nd place Eric solid instinct, slight overconfidence
00:36:21
on a few misses and in first myself.
00:36:25
Great balance of bold but accurate calls.
00:36:28
Thank you, Chachi I. Love it, love it.
00:36:30
I think I think this AI film take is complete bullshit.
00:36:33
So I need to I need to score adjusted for that.
00:36:38
This is just a check in, no medals awarded yet, but I will
00:36:44
take the pole position and see you guys in a few months where
00:36:49
we can do the final tally. All right, let's do our fantasy
00:36:53
draft. Max, you want to talk us through
00:36:56
the game and then we'll get into our picks.
00:36:58
Yes, we invented an ingenious game based on fantasy football
00:37:04
that allowed us to draft teams of startups into imaginary
00:37:09
rosters. We've done two different drafts.
00:37:12
We did one about a year and a half ago and one about six
00:37:14
months ago. We restricted the draft list,
00:37:18
the draft board as it were, to companies that had raised over
00:37:21
$100 million at the time. So if there's obvious omissions
00:37:25
that come up in your mind, it's probably because they hadn't
00:37:27
raised 100 million at the. Cursor Cursor.
00:37:29
Being the most. I think like there was some like
00:37:32
where we've that we didn't include because.
00:37:34
Yeah, We also excluded specific like chip based companies,
00:37:39
Chinese companies, I don't know for robotics, I can't remember
00:37:42
healthcare, we anything we thought we were even Dumber than
00:37:46
normal about we left off the list.
00:37:47
So we all drafted teams. We did a snake draft.
00:37:51
Most notably, the first draft involved a discount that we had
00:37:57
one person had to take for getting the first pick because
00:38:00
the first pick was very obvious. It was Open AI.
00:38:02
Eric paid, I believe, $75 billion in handicap to draft.
00:38:07
Opening auction 1st and I and I really good about it.
00:38:10
Aged pretty well since they're valued at 300 million right now
00:38:12
3. 100 billion, would you say? 300 billion.
00:38:15
I'm sorry. Yeah, like coughed.
00:38:16
Yeah, it doesn't matter. And I will say before we say our
00:38:21
teams, I have not yet had the first pick in any draft.
00:38:25
So I just want everyone to remember that when making our
00:38:27
teams, OK, I will go through my team first, which I will admit
00:38:32
upfront is in last place right now.
00:38:34
All right, my team, Databricks, my star worth $62 billion,
00:38:38
Cohere AI, model company, modular AI, language company
00:38:42
Scale AI, Sierra AI, Sakana AI and Hebia.
00:38:47
And if you know all of those names, you are far too online.
00:38:51
I'm bullish on Sierra. I think that's, I think scale.
00:38:54
Has room to run as well. Oh yeah, yeah, and obviously
00:38:57
Databricks isn't going anywhere. But Anthropic surpassing data
00:39:01
bricks has been a bit of a sob story for my team because James
00:39:04
got Anthropic at Crazy crazy money if I recall.
00:39:07
The 4th pick that wasn't even your third pick because it's a
00:39:10
snake draft. So it was me then Max the data
00:39:13
bricks, then James with you'll say in a second, which doesn't
00:39:17
make any sense, and then fourth with Anthropic.
00:39:20
It's so embarrassing in retrospect, just.
00:39:23
To just to clarify, like we drafted these teams originally
00:39:26
in 2023 and then we did, I don't know, we drafted it.
00:39:30
We, we had an ad drop waiver period last November and this
00:39:35
again is a mid year check in. No, no ads, no drops, but
00:39:39
checking in on the teams. So Max, I have your score so far
00:39:43
right now at 93 billion, mostly because some of your teams have
00:39:48
not raised or exited since you drafted them.
00:39:51
I have no valuation on Sierra and scale is at 25.
00:39:55
I'm somewhat optimistic on both of those.
00:39:57
I managed to somehow pick the only foundation model company in
00:40:00
the world that isn't wildly overvalued.
00:40:02
Cohere SSI thinking machines, Anthropic.
00:40:08
Like just throw a dart boarded foundation models.
00:40:11
You've got a $40 billion company.
00:40:13
But I'm. So Silicon Valley, the show said
00:40:16
this from the beginning. No.
00:40:17
Revenue is so much better than revenue.
00:40:19
Cohere is a real business, so people can value.
00:40:22
And. SSI and thinking machines are a
00:40:25
dream. I have made a huge real.
00:40:29
Revenue I I think SSI has real revenue, but anyway.
00:40:32
Yeah, Max, what's your learning from this so far?
00:40:35
I would say we already knew this, but you know the winners
00:40:38
keep winning, right? Obviously Open AI swamps
00:40:41
everything else that's happened in the entire draft.
00:40:43
So we have a true power law which is nothing else matters
00:40:45
comparison to Open AI even with the handicap which.
00:40:49
Which we knew, which we knew was the risk when we came into it.
00:40:51
Was happening, but regardless it still happened.
00:40:55
Secondly, I would say Databricks is, you know, a merely a $60
00:41:00
billion company, but that looks lame compared to, you know, like
00:41:05
XAI being valued 80 like. You know, all right, all right,
00:41:08
let's not spoil. James, you want.
00:41:10
To go next. My team with the first pick on
00:41:13
my draft that you guys were making fun of just the moments
00:41:16
ago hugging face, No value because they have not raised
00:41:21
since 2023. I've drafted them because they
00:41:24
had one of the highest valuations at the time of the
00:41:27
draft. They were valued, I think, over
00:41:28
a billion dollars. I thought they were valued at 4
00:41:30
or 4 billion. Dollars at the time, yeah,
00:41:32
something like that, yeah. Yeah.
00:41:33
OK So Anthropic giving me 61 1/2 billion value replit hasn't
00:41:42
raised since the original draft. I exited Adept at 1 billion.
00:41:47
I snagged XAI with the first pick last November.
00:41:52
I'm. So jealous of that.
00:41:53
Locked in 80 billion of value right there because they raised
00:41:57
earlier this year and runway also raised $4 billion
00:42:02
valuation. 11 Labs also raised at a $3.3 billion valuation and
00:42:08
poolside no raise recently. Also one of those foundation
00:42:14
model companies that we have yet to really see appear on the
00:42:18
draft board, but I am happy with my overall team and my score of
00:42:23
close to $150 billion currently. You know, something I just
00:42:27
thought about, You're extremely lucky that XAI purchased X and
00:42:33
not the opposite way, because if it had been X, it purchased XAI.
00:42:37
We'd be able to like, force you to disown whatever growth, but
00:42:40
now you get to benefit from this combined monstrosity, which
00:42:45
kudos to you. I would just say had I been able
00:42:48
to draft first, I would be in James's position of being in
00:42:51
second place with XAI, so I don't personally think that a
00:42:55
coin flip should be dictating my performance right now.
00:42:59
What is you? I know.
00:43:01
What was me I I will say to give James credit here, I I truly
00:43:05
believe that anthropic is is the pick of the draft or, you know,
00:43:08
so far I think that just getting anthropic at the fourth position
00:43:12
in retrospect looks insane. And so I think that that is the
00:43:16
the greatest. I don't even know if I call it a
00:43:18
sleeper is sort of a semi sleeper pick, but that that
00:43:21
clearly to me has had the most appreciation.
00:43:23
Eric, why don't you all? Right.
00:43:25
So yeah, I picked Open AI with a $75 billion handicap.
00:43:29
Now it's worth 300 billion. So I'm getting basically 225
00:43:33
billion for that. Inflection sold for 1.43
00:43:37
billion. Character sold to Google for 2.5
00:43:41
billion. Glean we're scoring at 4.6
00:43:45
billion, but rumored to be raising at 7 billion.
00:43:49
Miss Straw AI worried about that one. 6 billion right now.
00:43:54
Perplexity. Oh man, I'm getting no credit
00:43:56
for that. That's going to be a good 10
00:43:58
right now, but it will be 14 is apparent according to the rumor,
00:44:02
so we'll see. Safe super intelligence.
00:44:05
I knew this was buzzy, but I don't even know if I could have
00:44:08
seen this one raised at $32 billion already.
00:44:13
It's more perplexity like That's insane.
00:44:18
Kodium sold for 3 billion do. You want to explain that it's
00:44:23
they renamed to Windsor for their name.
00:44:25
Kodium is windsurf. Yeah, they sold to Open AI and
00:44:28
then Harvey. No credit right now but rumored
00:44:31
to be raising at 5 billion. Total value $274.5 billion.
00:44:39
Yeah. I feel really good about this.
00:44:41
I mean, hysterically, as I think I mentioned on the last episode,
00:44:44
I I wrote a bear case about open AI after this at 157 billion, I
00:44:49
think, but whatever. So I'm getting it narratively
00:44:52
both ways. But yeah, I mean, I'm proud of
00:44:55
all my picks. I think even my sort of singles
00:44:58
are selling and I'm bullish on basically everything except I
00:45:04
would like to hear what's going on with Mistral.
00:45:07
But yeah, it's I mean, it's a power law business.
00:45:10
It's crazy that like I'm like, oh, glean that's that's a good
00:45:13
company. I was totally right that that
00:45:14
would be a good company, but it it doesn't really matter for my
00:45:17
my performance. Yeah, I mean I think you are
00:45:19
consistently hitting singles and double s, but you could have
00:45:22
nothing on your team except opening eye and be beating us by
00:45:25
100 plus billion dollars at this point.
00:45:26
So it. Doesn't, no.
00:45:28
Which is why I made a that we could not randomly assign the
00:45:31
first one and I I said we know. You were right.
00:45:34
So I'm complaining, I I'm not complaining.
00:45:36
You made the right decision 100%.
00:45:38
It's just it, it is remarkable. It's like the whole game is just
00:45:42
like, open. AI who drafted Open AI who
00:45:44
drafted one Open AI go back. To that episode.
00:45:47
Yeah, exactly. Yeah, No, we talked about it.
00:45:49
I mean, I think it was it would they were raising at 90 at the
00:45:52
time. And so you ended up with a $75
00:45:54
billion handicap on a $90 billion company, which seemed
00:45:57
like a reasonable deal to us. But you know, we were.
00:46:00
We were all wrong, obviously. Or at least we're wrong with us.
00:46:02
Check in. These do have to last, right?
00:46:04
It's like 5 years or something. We're we're like, yeah.
00:46:07
Five years, yeah. Yeah, we had a good shot with
00:46:10
Sam Hoffman getting fired of your team.
00:46:13
Going up sync. But but then he came back in
00:46:16
force. All right.
00:46:17
What's come on the market that you think we'll be looking at at
00:46:19
the end of this year? Thinking machines.
00:46:21
Cursor for sure. Manus.
00:46:24
Oh yeah, yeah, yeah. Thinking Machines, cursor and
00:46:27
manus are the ones that come to mind for me.
00:46:28
Yeah, I mean, it's, it's going to be a tight band because we're
00:46:30
we're picking them up and we're only interested in ones What
00:46:33
that. They have to have raised $100
00:46:34
million. All right, well, that's that's
00:46:36
basically our episode. We're gonna have two more before
00:46:40
the Cerebral Valley AI Summit in London on June 25th.
00:46:43
And then at some point, once we've gathered ourselves, gone
00:46:47
to that wedding in France we mentioned and relaxed a little,
00:46:50
we'll come back to you and give you our thoughts from the event.
00:46:53
I'm super excited about next week.
00:46:56
No pressure James, our our game master over here, but we're
00:47:01
trying to come up with some good concepts, but we'll be talking
00:47:05
about voice and video and certainly in light of what VO3
00:47:10
Googles new video creation model, it's an exciting time in
00:47:14
video. So see you next week.
00:47:16
Thanks guys. See ya.
00:47:18
Thank you.