Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
Our Official Pivot To AI
Recorded: Feb. 27, 2026 Duration: 1:08:12Space Recording
Full Transcription
. Thank you. what's up everybody how are we doing how are we doing how's it going just trying to figure out
how we get this stream on the timeline now that Daddy Elon is saying that you have to go native on Twitter and we're not using native on Twitter.
So bear with me a second.
Yeah, I'm completely throttled on Twitter as well.
The Nikita algorithm is not doing well for crypto folks.
But wow, we're here.
We finally did it.
We finally found the time
to do a show together.
Running back the Blockmates AI themed show.
I'm so excited that you guys
invited me to run this.
And I'm so excited to tap in
with the community here
that you guys have built.
We're officially announcing the BlockMates pivot to AI.
Crypto's done.
Sell everything.
We are officially pivoting to AI.
And this is our show to demonstrate that.
If you're in crypto, pivot to AI.
Hopefully this is a bottom signal no but we're going to be here to talk about everything involving the frontier specifically ai right now is very hot we've got
a bunch of different tools that we're playing with a bunch of different workflows there's all
kinds of news we are in the singular. There's things happening all the time.
I feel like I'm behind every single hour of the day. News that happened earlier in this week felt like it happened a month ago. There's just so much to cover. There's so much breadth. There's so much
depth in the breadth that we had to create an entirely new show about it. And I'm super stoked to come
on here and just yap away because there's so much to yap about. And there's so much cool
information bouncing around and cool tools and cool workflows that we wanted to give you all
an opportunity to use these things to better your lives and your job and your trading and your investing
and your research.
I mean, even in your cooking, right?
Like I think 563 is now becoming a Michelin star chef.
But no, we're not actually quitting crypto.
We're not.
We're always in.
Once you join crypto, you can never leave.
Like forever.
So we're just expanding we're
broadening our horizons we're presenting you guys with new information um that we are using every
day and consuming every day so we think it's valuable for for all of you to to to know it as
well that's really the crux of it i'd say, yeah. I think the more I've kind of came to think about it,
like originally when it was just me writing on a small Squarespace blog
that was like $20 a month and nobody was reading,
maybe like five people,
and then shit started getting real with blockmates and all of that.
Like if you boil it down to it,
it was just presenting what's happening at the forefront of forefront of tech really um and trying to make like really technical topics really accessible to
the vast majority of people on earth um so that was the plan and the way that everything's going
i feel like everything's just merging into one cosmic soup of really interesting on the fringe
technology so it's it's kind of like the areas that we want to
cover that's just interesting to us we've kind of put a few test pieces out over the past few months
we thought there might be a splitting of an audience if it turns out that everyone's just
interested in really cool shit so yeah we're gonna there's no way that we'll be cooling down
on any coverage of the crypto stuff. So that's who we are.
As Brody was saying, you're here forever in the industry now.
So just expect to see some weird and wonderful coverage of crypto AI, robotics, quantum, biotech,
like anything that is interesting.
Like there's some really interesting hardware stuff that we'll probably cover in the near future as well.
So and if you like that, we're going to consolidate it on Friday.
So there'll be a stream once a week and the blockmates newsletter will get taken over but everything that we just previously
mentioned on friday as well that should land in your inbox within the next hour and if you
do want to find it i'll nip onto twitter in the next two minutes and put a link under there as
well but uh yes we should we get started fellas so, it turns out there's a large overlap between people that like playing with DeFi ponzus
and who like playing with Cloud Code.
It's the same people.
The trenches are widening, though, so that's nice,
over in AI land.
Yeah, it's just not an immediate gratification trench anymore.
Like, you actually have to do something now.
I can't just, like, punt money at Bowdoin and get like a 30x.
You're telling me I have to go actually build an app now and then get customers
and then convert them to paying customers.
It's a lot more work, a lot more hoops than MemeCoin trenching.
Might be more sustainable though, just a little bit.
Yeah, and you might learn a thing or two.
This is the new PVE.
The new paradigm.
Yeah, the new PVE is just punching GitHub commits
every hour of the day and then waking up to your agent
punching more GitHub commits and still feeling left behind.
But no, let's get into it, man.
I love this Jason tweet. I just wanted
to just to talk shit about it real quick. Like, it is funny because he is victory lapping right now.
And then again, no, we're not we're not leaving crypto entirely to pivot completely to AI. We're
doing both. Hopefully this tweet is a bottom signal but yeah let's get
into like why we are actually doing this i mean this image is crazy if this is true of course
which not everything you read on the internet is true guys not everything you hear on the internet
is true either so take it with a grain of salt but if each dot is representing 3.2 million people and this is the entire population and you are using like cloud code, cursor, any of these coding models, if you're just building something, you're in what is that?
The top what percent?
0.04%.
So if you're listening to this, you're already by default probably using this stuff. So you're
already in the 0.04% of people who have adopted this technology and are using it to change your
life. That's crazy. So if you think you are behind, the chances are you're ahead of 99.6% of people,
of people, or sorry, 99.96% of people, if this is accurate. And I don't know about you guys,
but I feel like now when I'm constantly online, I'm chronically online, I have been for the past
few years, but more so these past few weeks, I just feel like there's a, there's the, the disconnect between the people who know what the hell's going on versus the
people who have not start,
tried to learn is, is widening very fast.
And even when I try to explain to people kind of what's going on,
like I'll send this a training piece and we'll get into this a training piece
because that's, that was a huge part of the news cycle this week,
but I'll send the Citrini piece to people. And like like they don't even understand what he's talking about you know like
none of those concepts have ever even occurred to them so it's it's really difficult for them
to even establish a baseline foundation of what the next 18 months looks like in that future that
citrini painted i I don't know.
I just feel like we're in a really weird spot right now.
What do you guys think?
You go.
Yeah, I mean, even the people that have used these tools, you can see there in the green,
almost all of them are just using the free versions too.
So they don't really understand what you can what you're able
to do so i think that's why the citrini piece hit so many people because at this point we're getting
to the the point where these tools can actually be useful writ large and it's not just by people
that know how to code or are already building it's it's people that have the agency to go and try new things yeah for me like i put it i
put a tweet out after way too many uh beers on the couch watching the football but i think it
there was definitely some truth to it even though i was kind of pissed at the time it's like this is
this is one of those situations you know when you think back and be like oh i wish i was around the
dot com bubble i wish i was around during like be like, oh, I wish I was around the dotcom bubble.
I wish I was around during like the birth of the internet.
I wish I was around blah, blah, blah, blah, blah, blah, blah, blah.
This is like this generation's opportunity like that.
And the only time that I've really felt like there was such a blue ocean of opportunity mixed with excitement and also a healthy dose of oh fuck this is absolutely insane was like very
very very early days of crypto and then very very very early days of defy right um 0.04 percent of
people who've used coding models that is an absolutely insane stat and all it is and if
you're really lazy like me and use whisper flow or a dictation app like whisper flow not a sponsor if they do want to
sponsor we we are open to bribes but literally holding fn on my keyboard and talking into
claude saying write me a great prompt and then taking it over to claude code and then dropping
it in like i know several idiots like i know i probably know more idiots than clever people
they would be able to do that, right?
Me included.
What more do you need?
What more do you need?
Sometimes when I'm using whisper, I'm like, damn, I actually need more time for my brain to process what I'm trying to say.
So I need to just type it as opposed to even speak it because sometimes I ramble.
Yeah. I'm getting whispers listening to me. But yeah, we're going to dive deeper into,
into our, our, our stacks as well. And whisper is on my list of things to talk about.
But yeah, I think you made a lot of great points. Um, the, the, the, the friction and
barrier of entry is decreasing so quickly that as long as you literally just do something,
just do something, you are in the top 1%. If you pay for a plan and you're getting,
if you just have access to the best models, it's like, you're still in the top,
whatever percent that was right. Um, so you are best models, it's like you're still in the top whatever percent that was.
Right. So you are still early, despite it feeling like everything's moving around you so fast all the time and waking up and being like, wow, my competition just dropped a completely degree where it makes sense to consume information that helps you.
But a lot of the doomer porn and a lot of the stuff that makes you feel like gives you anxiety, just try to block it out.
Stay focused. Build your thing. Stay in your lane. Don't get too distracted.
It's really easy to work on six different things right now.
Just pick one or two. Go all in. P out of it, execute, change stance, change course,
get feedback, pivot, you know, but don't get, don't get too distracted.
Yeah.
Why do you think that obviously the algorithm is obviously rewarding, like breaking, you
won't believe what just happened, like blah, blah, blah, blah, blah.
And it's just like, it adds to that sensationalism.
And I feel like a lot of the accounts that are talking about this shit don't have a fucking clue what they're actually talking about. just happened like blah blah blah blah blah and it's just like it adds to that sensationalism and
i feel like a lot of the accounts that are talking about this don't have a fucking clue what they're actually talking about and it's just they are just purely there for the
dopamine the engagement and this is like flavor of the week and now this is where you got all
the engagement so we we cover that got people sending out like mad like mds and mad skills and like mad GitHub repositories,
like could be just riddled with like vision links or malware or anything.
They don't even know what they're sending out half the time,
but they just want the engagement.
So there is that, but it can lead to a sense of, oh, fuck, I'm behind.
So it's really important to see infographics like that,
where it just puts everything back into context and I always use like Carpathia as like a bit of a that's the benchmark like whatever he's
talking about that's like the rational thing he knows more than everyone else in the room
so whatever he's talking about I'm just gonna like lean on him for that's like the benchmark
of where I'm putting everything and he's talking like mad about open closing it's a really really good step um so yeah it's like don't get lost in the sauce too much
about it like just try and keep up a little bit but at the same time this is what these weekly
streams will be for as well right right we're gonna try to filter the noise and only present the best stuff um every week but yeah let's move on there's there's um
the frontier is under attack just you know we're talking about we should stop being doomers well
it's under attack you need to be you need to be scared no no okay that's not what this means no
but basically over this past week,
it came out that a lot of these Chinese models were distilling
Anthropic, Claude, Opus 4.6. Basically what they're doing is compressing and maybe
563 could probably have a better technical explanation. I don't know if you've, if you
understand distilling and all those capabilities, but basically they're just ripping off the capabilities of Claude models and providing them open source for way cheaper.
They do this through a number of ways of attacking it to identify the weights and parameters of this model so that they can replicate it on their end.
parameters of this model so that they can replicate it on their end. But this was an
interesting point in time, or I think this was an interesting piece of information this week,
because it shows that these Chinese labs may not necessarily be on the same plane as the
American frontier labs. And we had this deep seek moment, which was kind of like a fake out as well.
we had this deep seek moment was which was kind of like a fake out as well and this was kind of
another moment where moonshot may not be like these chinese labs may not be innovating at a
faster pace than the frontier labs which has been a big concern um for western national security
purposes over the past like six months big big concern big topic. This news feels like it was a month old, but this probably happened on Monday. That's just goes to show how quick things are moving,
but this is a cool infographic. And I don't really want to go into the depths of it. And
maybe we can send this out and on a tweet later today. So you can take a deeper look at it if
you want. But overall, this was a big piece of information,
a little bit more of a technical piece,
but curious 563, if you've been playing around
with these Chinese AI models,
have you noticed similar patterns and behaviors
with those as you have with Opus?
I've used Kimi a little bit.
Yeah, the whole idea here is they're trying to understand
how Opus thinks. So they're trying to understand how opus
thinks so they they're going to prompt certain things to try to understand what's happening in
the background and then using that as data for how they're going to train their models later on
i mean this is always a risk when you put out a model because it has to respond and it's i don't
know i think this is just part of the game like obviously these models are trained
what like six months ago three months ago so by necessity these are going to be behind because
they're working off of old data um but yeah it's part of the game i i do find it interesting that
all these chinese models are open source or they're trying to go more towards the agentic side too
they're leaning into that more than it seems like the western ones are yeah i think i think there's a a lot of like really interesting
ways that you could actually take that i remember that deep seek moment where they released that i
can't remember which model it was but it was the mixture of a experts model that came out and there
was a huge sell-off and everyone's like oh well if you can get the same output uh get a higher output
by per gpu demand then maybe like i'm
pretty sure nvidia sold off quite hard off the back of that actual release and maybe that was
overblown and what you what you're saying there is is quite important but it's also like the same
history of everything that's ever gained before it like us and the west tend to actually lead and
then there's a easier way to it the whole like i think five six three
explained this to me before it's like mcdonald's will spend millions upon millions doing real
estate research trying to look for foot traffic um and you know have that extremely large capex
and then berger king just goes and opens across the street because all the hard work's done for
them effectively that's right but if that if that is the case what is like what is the second and third order effects of that do like we continually get this digital arms race
towards this new frontier of technology which has kind of been touted and i feel like there's a lot
made of this and i think it i think it does a good job for the frontier models that are in the u.s
because it's like oh well if we don't
continue to get this funding and like you're seeing sam talk about bit government bailouts
and stuff like that the lever that they're going to pull there is because we're in an arms race
with like china like that that is like the fallback like de facto fallback plan if they
don't get the funding and continuous like government subsidies and all that blah blah blah
um i don't know what that does for that um kimi
2.5 is fucking good um travis good who we had on the blockmates podcast for uh last week this week
it's all merging into one he talked about this before this news actually happened and he said
there's going to be a strange strange strange convergence because all the the data that's
getting fed into them is kind of finite at this point you're gonna have to start pushing towards
synthetic data the chinese models can distill distill distill distill and it's really
really interesting that the landscape's going to change i don't know how this plays out um but if
i can get i think there was a an open source model that you could basically run on a mac mini
um i think it might have been mini max i got released or a QN model I got released this week for free.
If you can run that on a Mac mini or a Mac pro for free
and you run an open call instance or whatever,
that's better than the 200, 400 API credits
a month that you're going to be spending using Clawed Opus,
which I have been absolutely rinsing.
It's replaced my Binance liquidation email.
I'm getting absolutely tortured on that front.
So I just think it's better for finance liquidation email i'm getting absolutely tortured on that front so
i just think it's better for the end consumer but um we'll just have to see how it plays out on a like a geopolitical scale i think i mean talking about geopolitical angle to all of this stuff
we had some big news yeah where dario basically says go yourself to the department of war um and no you cannot use
our models to uh to kill people and to take out military operations tulip king had a crazy crash
out on this already i don't want to i don't want to you, steal Tulip King's like limelight here. But yeah, I mean, I don't know.
What do you guys think that this means?
Like, how do you tell the Department of War that too bad?
You can't have it.
Someone's going to have to fall in line, right?
Do you think OpenAI falls in line and just gives them what they want?
Yeah.
And I think they've definitely done that already.
Yeah.
It's already happened. Yeah. just gives them what they want yeah and i think they've definitely done that already yeah this
already happened yeah and um i know that elon is trying to broker a deal with the pentagon as well
for x that was actually top story in the newsletter uh this week but yeah i mean someone will fill the
gap but what capabilities does claude actually have that we don't know? And that goes back to Alexander
Good's had a take on this as well, Ruth. There's going to be a white-labeled or white-listed
API access to anybody with an enterprise account, and they're not going to allow
regular consumer accounts to have this high-performant model. Maybe they already have that.
high-performing model maybe they already have that
um how does it look like training a model on on military data like yeah they probably have ones
without with fewer rails like uh fewer guardrails i should say that's what i would imagine like
yeah like ones that they could trust your gas yeah honestly mistakes pretty much yeah we we
joked i think it was on tuesday we we joked and said um extract maduro claude extract maduro make
no mistakes because apparently that's what they'd used it for in that mission for getting him out
of the country so i don't i don't know man like what What the fuck do they want to use it for? If those are the
two points of contention with regards to US civilian surveillance and total autonomous
weapons and drone usage, if those are the sticking points that's more concerning than like the whole
use of ai for department of war and of itself in my opinion but what the fuck do i know anyway
yeah somebody dropped this project they made in claude that was all of the data although all the
all the ways in which your data is leaked and all the ways that you can be patched together to create a profile on you, basically like a social credit score. And it was crazy with like your ISP,
all of your messages, your cell tower data, your location data, your DoorDash orders,
basically everything. And it can create like a profile around you that's basically exposing you
for all of your frauds, your internet search history.
And it can create this entire map of your psychological state, your physical location.
All of your activities are all mapped to you and it can create this profile around you and the type of person you are and all of your deepest, darkest secrets and use that as leverage over you. I mean, they've written complete sci-fi novels around this type of stuff. And it doesn't, guys,
it doesn't really surprise me, but it doesn't really seem too difficult to create and map out
a social credit score for every single person in the world based off of their internet and online activity.
I mean, now that we have the tooling to easily scrape together all this stuff,
you don't have to manually patch it together. It's kind of concerning. The surveillance state
is becoming more oppressive and easier to create. And I wonder how that kind of unfolds over the next 18 to 24 months.
If everyone, I've also had a thesis that like, if everyone has their private information doxxed,
then how is it actually that interesting to have your information doxxed? Like,
you know, here, here's a profile of all of the bad things that I've done in my life. But if
everyone has all of their secrets told,
then does it just mute the fact that we've all been doing bad shit behind the
behind the closed doors and that we're all kind of just humans at the end of
the day? I don't really know.
I guess it depends on how bad the shit you're doing actually is.
But I mean, I feel like I'm clean, I guess. I don't know.
I don't need one of these 483 people to try to create a scrape on me right now.
But I don't know.
I feel pretty good.
The extension of that is the news of OpenAI
about to drop a not suitable for work section.
So like naughty chats.
So that for me just feels like... It's a honeypot right there yeah that's like
epstein walked so you're gonna have like all of the world using chat gpt who have only just
convinced themselves that they might like the amount of tweets i've seen is like from
like instagram screenshots saying as if people pay for chat gpd is absolutely
hilarious but that's a tangent if they have a section like on the back end called naughty chats
where you can only tap into it when you're talking about like weird or getting your
kinks from chat gpd and all that weird that is the greatest like digital epstein honey
that is the greatest like digital epstein honeypot device of all time
so yeah just be fucking careful what you're doing just go to just go to thailand bro
get offline go feel the sensation of a of an english a 20 year old english man who's left
who's left europe for the first time
and going to thailand holy shit talk about the amount of english people in thailand man
oh yeah yeah you guys love that place yeah yeah but moving on kings over there
it's awesome um but moving on we've got some NVIDIA news. NVIDIA reported earnings this week and
they absolutely crushed, but the stock's down. Asset allocators, in my opinion, are just in a
constant flux of confusion. They don't actually know how the hell to allocate money right now.
I mean, data center revenue is up 75% year over year, right?
I mean, they're just completely crushing.
They've got so much demand.
They've got all these partnerships.
They've got all these buyers of these GPUs.
They're innovating on the Rubin platform now, which is the next generation of the Blackwell,
which the Blackwell just came out.
And they've got so much demand.
They've got so many buyers.
And yet they're
crushing earnings, the stock's down. And it's a weird place right now. The NASDAQ is kind of
plateauing, as you can see here. I think this is from the past three months. The stock has really
plateaued. It doesn't really know where it wants to go. What more good news could you ask for out
of this stock and out of this company?
And yet it's still finding difficulty to attract new capital into it. So is the party still
continuing? I think. I don't know. Seems like inference demand is only climbing.
Inference demand is only climbing.
People need more inference.
AI is going into everything.
People need more data centers.
People need to serve that inference.
Feels like it's a trade that keeps going up,
but market doesn't really quite know yet.
Yeah, people still think we're in a bubble.
So everyone's looking for a reason to want to sell.
But I just think we're in a total new paradigm.
So the old laws don't apply here.
This time is different.
I think it really is.
That's the problem.
And if you look at all the Mag 7,
I think NVID is like uniquely positioned right
because the demand is obviously there it's just like you said people are just super defensive
right now because there's so much uncertainty with everything else yeah chumbawamba basically
made the entire point of his thesis around uncertainty climbing and uncertainty is
inherently bad for equity in risk markets. But where do you put your capital? I guess metals,
that's where it's kind of going right now. I mean, silver's up, gold's up, copper's up.
I haven't checked it this morning since market opened because we had this stream,
but gold was trading like 96 dollars now it's
trading at 93 um yeah yeah halo thesis around like really high defensible companies energy
like exxon coca-cola's so that you're ripping you know like these really really um hard to disrupt businesses that
you're just not going to be able to replace overnight with ai so like there's been a really
good pair trade with regards to that versus either sass or anything that could just get wiped out by
someone one shot in opus 4.6 you know yeah there's an alt super cycle just not in the alts that you hold, just the stock market alts.
They are trading like crazy alts.
But I guess the party is still going on.
I mean, OpenAI just raised $110 billion this morning from NVIDIA SoftBank.
Maybe Oracle snuck in there.
I don't know.
Amazon snuck in there this time.
Oracle said, I'm done spending
money on this thing. But Amazon said, let me get a piece of that. So they're using OpenAI. I will
start using the new Terranium chips from Amazon as well. Amazon, a lot of people are calling out
Amazon as a sleeper contender in this AI race because of the teranium chips.
It's kind of similar to in cloud. It's kind of similar to Google's TPUs, if I'm correct,
just an alternative ASIC for inference. A couple of people, I think Gavin Baker had an interesting take on how the general purpose compute platform is actually more advantageous as an asset to acquire on a balance sheet than as an application specific inference chip.
And it was something to do with like, they hold their value better.
That's an interesting thing that we could probably tag here.
If you're curious to learn more about the chip space, Gavin Baker, he posted it this week.
You should go check that out.
He's been covering NVIDIA for like, I think 25 years now.
So he's a legend.
He's also wrote a piece about space data centers, if you're interested.
But yeah, I think the party's still going on, I guess.
110 billion out of $730 billion valuation. I think they're the highest valued Frontier Lab
right now, unless Anthropic is closing the gap.
I was at 380, I think.
380 the last round. I mean, I'd expect them to probably double. I had a tweet.
I was like, the best pair trade this year,
if you're trading private markets
and you can create a pair trade,
which you probably can't,
and you probably can't trade private markets like this.
Well, I know you can't actually,
but the best pair trade would be Anthropic OpenAI.
Anthropic Flipping OpenAI.
You think they'll flip?
I think so, man.
They're on track too. If you look at how fast they're growing compared to open AI. They're shipping way faster and people actually like,
like Anthropix product focused vision is working really well. I mean, I'm using cowork like crazy.
Granted, I'm also using codecs like crazy within cursor, I haven't been with I haven't I haven't used the open AI chat GPT subscription in like three months
yeah obviously the value accruals obviously
concentrating around the coding models so like that decision to primarily go after that.
I still think like GPT 5.2 for just everyday,
really lightweight use sometimes is preferential
over using either Sonnet or particularly Opus
if you just ask them for like random shit.
I feel like I flip between all of them
for different use cases and kind of got like
workflows of why i use one for one thing one for another but um yeah i don't know it's just like
their expected revenue that they were churning out like some charts that were getting thrown
around from amthropic is absolutely incredible like and the thing is where those 200 million
dollar uh grants from the government and investments from the government,
they don't need that money.
They do not need that money the way that they're going.
It's absolutely insane the revenues
that they're going to be pushing out.
That catch-up trade with regards to Anthropic and OpenAI
is probably the most easy telegraph trade in existence.
And OpenAI pushing out towards ads for the free models
and the adverts that they pushed out
which are nailing the coffin for me.
Just have to see how it plays out
because if Anthropik and Dario don't bend to what the US do,
we'll see the full wrath of what happens when that doesn't
go according to plan for the government.
They can make your life absolutely
fucking hell if they want to.
See how that plays out in this whole thing as well
because if they take that stance, I guarantee
Sam's going to take the other stance and just
cause you up to them.
Yeah.
Yeah. I don't know.
I guess we'll find out um we'll see i think i do think like the whole chat gpt thing is so stuck in the um casual i don't want to say normie i feel like normie
is so derogatory nowadays but i just it's just so it's so captured within the um
the casual user's brain that they refer to all ai as chat gpt like they really captured yeah
um well just yeah did you see the uh there was a few there was a few tweets where they were like oh instagram has just discovered claude
and it was like a load of these like female instagram female instagram accounts that were
like tweeting basically saying oh yeah i just downloaded a new one called claude but it's like
it's given me like really autistic vibes that's why we get along with it man yeah that's why we like it but yeah it wasn't it wasn't quite uh sycophantic enough for like the instagram female uh
influencer crowd and they were calling it like a little bit weird i feel like the i feel like
the aesthetic though is is a little bit more instagram model chic you know like beautiful brandon like open ai just feels like silicon valley
corporate slop just like middle of the line well it's like white black i don't know it's the only
one everyone who knows about that's what mike was saying his missus uses chat GPTs like mine does too. They use the free plan, dude.
My dad, he's been fixing his motor on his boat recently.
He's been using the free version.
He's like, this is fucking incredible.
I just saved $600 fixing the water pump on this motor.
I'm like, what are you using?
He's like, chat GDP or something like that.
Whatever.
Chat GDP, that is a reason.
I was like, do you pay for it?
And he was like, no, I just got the free version.
So I guess like people still get a ton of,
I mean, that's value right there.
I mean, he saved like 600 bucks
by not taking it to the shop
on the free version of chat gpt so
yeah but he's also gonna get served like the the nastiest trim and show style fucking ads
for the next 25 years around fucking boat motors yeah and the boat will just like stop working
one day because like it told him the completely wrong thing to replace you know what changes the game with that sort of stuff
though when like i'm starting to see like gemini float these um like the the snapping the snap in
a picture or something and send that into chat gpt or claude or gemini or whatever you're going
to use and like asking it for advice on what the fuck i should do here like we moved we've lived
in asia lived in europe blah blah blah if i'm looking at something like direct translate what does this actually mean instructions just basic
shit you know i mean i use that i use like the image like take a take a picture what the fuck
should i do here basically um the next evolution of that which em and i have been floating around
i guarantee the the best at it is like let's say let's say your old man's example there
where he's fixing up the boat.
Just direct video and inline chat.
That's the next evolution.
The next evolution after that is,
well, not just recording a new camera.
It's probably a wearable that you're recording
and it can talk you through the shit live in real time.
So I think that'd be great.
That's cool.
Yeah.
That's cool.
I never even thought about that.
It's getting to the point of invading.
So it's moving on from texting.
The dictation's great.
And sending the images is great.
And then obviously the next step from that
is just recording the issue that you've got
and helping you along the way live
with fixing response and shit like that.
So I'm all for it. I'm lazy as fuck kind of pen right
the video thing yeah with johnny hive yeah yeah that kind of gives me like
rumba vibes where all of that data was streaming to china and some guy hacked a Roomba and then it got access to all of the data,
like the Roomba house layout data. And in like, I don't know, man, I'm not saying that like these
platforms and these hardware devices would necessarily be easy to hack, but the fact that
the data exists somewhere, someone has it and it can be bought back to my previous rant about that security
um doxing surveillance tool that that person built um yeah it just seems you're just creating
more digital footprint for yourself and i don't really know if like i've i'm off instagram i just
like i kind of had like a weird moment the other day where I was like, I just want to stop contributing to my data footprint in places where it doesn't make sense.
Like here, I'm like, you know, we're talking about a conversation.
I'm using Claude.
I'm building a footprint in that regard, but I'm using it for productivity purposes.
But I don't want this thing to have a data footprint of like my psychological pathways.
Right. purposes but i don't want this thing to to have a data footprint of like my psychological pathways right so i'm like off these scroll these endless scroll apps because i just feel like that's like a data data footprint i don't want to continue to build yeah we we were talking about this
on tuesday on round trip as i said like that's going to be like a conscious movement i feel like the no wearable no ai safe spaces around cities and towns and bars and
restaurants is probably going to be a thing because as soon as people start wearing wearing
always on wearables it's game over like there's going to be have to be certain places like in
bars restaurant pubs like wherever like no wearables right that'll be a thing and we were saying i've got two fucking
rumbas my fridge talks to me my fucking dishwasher and washing machine like ai powered somehow i
don't know it just just provides me spending an extra fucking 500 on it i'm running open
on my toaster i was like this so i'm always, I don't know why I do this,
but just to be a dick probably,
like looking for like the equal and opposite reaction of that.
Analog devices are coming back, man.
I'm telling you now, there's a fucking billion dollar opportunity
for anyone who wants to create just pure analog devices
that have no IoT install on them.
But we're just moving into a really weird world.
Like always on wearables.
It's like, I don't want to have to like have a weird interaction
with someone like you meet.
Imagine going on like five years time,
you're going on a first date,
you met on someone with Tinder
where your open clause probably selected them
based on like fucking when you're off work
and it's like lining up with their fucking estrogen cycle
or whatever the fuck.
And then you're both wearing like these always on wearables
and you're not actually saying what you want to say. You're suggesting like you're both wearing like these always on wearables and you're
not actually saying what you want to say you're suggesting like you're taking the most efficient
route so hopefully that you're going to convert it's fucking it's just weird dystopian world that
we're entering into they made a whole black mirror episode about that there's a couple yeah there's a
couple like that no yeah i mean i i like in hindsight i remember watching blackmail and i'm like that'll never
happen that's crazy and we're like pretty much damn near in it so should re-watch it
like oh yeah like that that was that was like we'll be watching in 2020 2030 like oh that was
so 2026 but yeah moving on we still got a lot of cover guys we only got 17 minutes we're gonna
have to start going rapid fire.
I wanted to touch on a Cetrini's post here because it was a pretty big,
big piece of the news cycle.
I don't know if you guys have talked about it on other streams this week yet,
but basically Cetrini laid out his,
his 18 to 24 month doom or I guess thesis that was pretty doomer.
Basically fictionalized sci-fi fanfic piece that basically explains step-by-step how AI replaces the workforce and the downstream effect
of all of that. You basically are laying off all these white collar workers that knocks out the consumption,
the primary consumption base of the U S economy. Um,
those people reshuffle into lower paying jobs or no jobs at all. It is highly deflationary because we're able to produce all of these goods for,
for basically nothing now. Um,
and that has an impact on the mortgage market, price of goods, like I mentioned,
and all these different bonds and debt and all these different markets and pension plans and
all this type of stuff. And it goes up in 2026 is how he describes it. We have a great year of
productivity games. People are getting laid off.
The margins expand.
Stocks go up.
Everyone's getting super rich.
And then once the consequences of all those layoffs actually materialize, we realize that
there's unemployment is skyrocketing.
No one can afford anything because the price of goods has yet to match the deceleration of deflation with the job market and the economy is kind of in flux.
So far, a lot of people like over the past three days were trying to like discredit this.
We're saying, well, you know, technically, you know, we'll actually, you know, all these like the Redditor
neckbeard type, well, actually that's not going to happen because whatever. He was basically
validated yesterday with the block laying off 40% of its staff. Granted, there's some nuance with
overhiring, post-COVID era overhiring, people probably weren't doing a lot of their work,
whatever. But the
first point of his thesis was validated four days after he posted the thesis. That counts for
something. And I mean, this guy's been right for the past two years on basically everything that
he's thrown out there. He's predicted every market trend over the past two and a half,
He's predicted every market trend over the past two and a half, three years to a T.
I mean, he's crushing.
And that's why when he released this thesis, literally a lot of these asset managers took
it verbatim as a guidebook and started selling off all of the exact tickers that he mentioned
in the article.
And I mean, if you guys have been paying attention to Citrini, like he'll post a piece and it'll be like some obscure company and it'll just re-rate 25% after hours or pre-market.
It'll just rally like crazy because people are copy trading like him like crazy right now.
So yeah, so American Express sold off.
A couple of these other companies like had huge sell-offs.
But overall, Grant, what do you think?
But overall, Grant, like what do you think?
Yeah, I mean, how much of it is, and it's probably a mixture of all of this,
but a lot of people already thinking that and then just got the confirmation bias
that they needed to pull the trigger.
And then how much of it is not so much that everything he's doing is true, but it turns into this Kenzian market situation
where if enough people,
like not the fact that what he's saying is to be true,
but enough people are going to believe it
that it's going to cause like a huge fold momentum
with regards to buying or a huge sell-off.
So then people are like,
oh, well, it happened last time you did this, so it's going to happen this time you did it and then it just
becomes like well if you say it it just becomes a self-fulfilling prophecy we see all the series
in microcosm across like a lot of the crypto markets and stuff but um i think there's a lot
of truth in it i think a lot of people are burying their head in the sand and don't want to believe
it but like my other stance on it is it's just going to create infinite new jobs in
different verticals that we can't even comprehend at this point in time it's like the way i keep
explaining that it's like the new uncharted areas of the map on gta that you haven't walked into
just yet so we'll have to see no one has the answers that's why it's that's why it's the beauty
of like predicting the markets on these things yeah i think he's directionally correct. It's just the danger is that these things
happen so fast. Like if these things happen so fast, the market doesn't have time to react and
people don't have time to get new jobs and new industries don't have time to spin up to replace
the workers. But I think that's the thing where maybe it's a little bit off base just because
not everything is software, like 100% software based. There are way more hurdles in place in the real world to this technology.
And not everybody's in the echo chamber of using the latest and greatest.
And the adoption will take time.
Like even if it is the best tool out there, it still takes a lot of these legacy businesses
a long time to adopt this tech.
So I do think it's not going to be as bad as what he's saying.
But yeah,
I mean, that is the general direction. But on your point of taking a long time to adopt the tech,
if people see, or I mean, not if, tech CEOs woke up this morning and saw the block
up 30%, I don't know what it's trading at now.
After they reduced their headcount,
like if they're being rewarded that significantly
for accepting these changes,
I think it could happen a lot quicker than we think.
In tech, yeah.
In tech.
I think it's different when you have a board,
when you have like things that aren't just software, making all the decisions.
Right. Because with software, yeah, one software engineer can now do the work of five.
Right. So maybe you can chop the heads off of.
Yeah. Like what he did here. But I don't think that's necessarily applicable to the entire workforce just yet.
Not yet. Maybe co-work has plug for every role in the next year or so.
Yeah, fair enough. Well, moving on, we've got 10 minutes left.
Let's use this last 10 minutes to not talk about the news because we've been inundated by news
all week. I'm sure you have been as well. We could probably talk about the news because we've been inundated by news all week. I'm sure you have been as well.
We could probably talk about the news forever. So let's talk about actual signal that you can
take away from this show. How are we actually using this stuff? We've talked a lot about
doom and gloom. Let's talk more about the optimistic side of all of this, is that we're actually using this stuff to better our work and to make things cooler, do 10x the amount of work that we were historically doing.
And I mean, this whole slideshow presentation, if you like it, you might not like it, because I don't know how I feel about it yet.
But this was basically one prompt. I put all the show notes into this and created this in one prompt, which is pretty cool.
You can do cool stuff like this now.
And that was with Claude, by the way.
But if you're in go-to-market, there is a huge universe of different unexplored tools
and workflows that you can begin using.
And Cody Schneider is a great
place to start. His entire page is dedicated to go-to-market workflows being composable systems,
not one-off prompts, integrating them to Facebook ads, Klaviyo, Beehive, whatever your email
management system is, whatever your marketing dashboard is,
whatever your CRM is, creating all this custom stuff that monitors and AB tests all these
different like campaigns, I guess, and creating easy, integratable systems within your Slack,
right within your team Slack so you can visualize all this stuff. I guess we got a new word for it. And to your point, Grant, we're going to create
completely new jobs out of this. And this is the go-to-market engineer. Nowadays, everyone's an
engineer. Everyone's an analyst, I guess. I mean, I saw someone who was like a content analyst. I'm
like, brother, you make TikToks. Like content analyst is crazy, but we got go-to-market engineers now basically some
barking marketers with with some with some vibe coding skills but if you're a marketer
shit go check out cody schneider he's got a lot of cool stuff i've been spamming the block made
slack with some of this stuff because i think it could be very valuable yes it's great isn't it
because this is like the the easy response to everything that's happening now and the easy
kind of like you know they're like negative people who just don't create anything and just like
poopoo every single idea and just a should be banished to the permanent underclass in my opinion
but these people just effectively show what is possible in multiple
different verticals so it's not just directly for engineers it's not just directly for people who
are like looking after admin multiple different areas that are still ripe for the right for the
taking in my opinion like marketing front end i know we've got the Figma, MCP integration, the clock code. All of that aspect,
I still think is pretty blue ocean.
I'm being completely honest.
I don't think that's entirely solved.
And there's a lot of like solutions
hacked together
where you're probably going to see
a little bit more streamlining
over the coming weeks and months.
But these people are just like
trying shit out.
They've seen like firsthand
how difficult it is to automate
all these kind of things.
That's where you steal like 10, 20, 30 hours a week back.
You really do.
If you can get this shit nailed down.
Yes, it's hard yards to get it set up
and get it out the gates,
but it's always going to be.
It is always going to be.
It's only going to get easier,
but at the same time,
it's worth putting the time and effort in now.
Even if it doesn't amount to anything
at the first instance,
you're going to have so many lessons learned that you carry on
maybe to the second and third and fourth and fifth iteration
of what you actually try.
And you're going to be so much further ahead than everyone else
because you've got that prior learning.
You've been through the shit.
You've been through hacking shit together with Zapier and Ithead
and stuff like that.
And then when it becomes seamless across all workflows,
you're so much further ahead because
you just know how this works so it's worth it's worth tinkering i think in spare time even work
time to be honest yeah start bumbling through it now right it's how you're going to learn it's not
going to work the first time but you're going to be way farther ahead of everybody else that's
hasn't tried this stuff out and they're looking for like a guide on how to do it just go do it
just ask claude ask chat gpt how to do it that's a good point that's a great point really good
point don't there's only so many twitter articles and youtube videos you can watch just go give
yourself a very very basic task and just do it like that's the best way to learn yeah and stop stop pitching people on ideas just do the idea like ask for forgiveness not
permission just do the thing it's so easy to do the thing now you don't need to create a whole
document that's like hey i want to do this for for my role i think i think this would be valid
just go do it just plug that document into cloud yeah just go freaking rip, man. Go rip. It'll be a bad version of it, but yeah.
I wish I had the graph where it was like AI's application to specific sectors.
And it was like 80% software engineering, 70%.
And it was like going down the list.
And it was like this whole area is blue ocean.
That whole area that was like less than 15% of, of, of people that are applying AI to their jobs. It was like legal healthcare,
transportation, logistics, all that stuff. Like that's crazy opportunity for you guys.
And if you need an idea, again, just ask Claude, try to figure out where, where AI has not been
applied yet and just start, start, start building stuff and start posting it.
Post it.
Send it to people.
Get feedback.
Show people you're doing this stuff because then they're more likely to, A, send you relevant stuff.
I've told five friends I'm building this thing, and they sent me a bunch of different resources that could help me.
Like, oh, you might find this interesting.
It increases your surface area for luck if you if
you just post about it and you should you demonstrate it and get feedback um and just
keep going that's at least that's that's that's my advice yeah i think like there's also people
are a little bit stuck but i'm starting to slowly get their head into these automations even if you
want to stay local with like small medium enterprises like that hyper local theory there's so many of those businesses
that i might have heard about it that might have drastic need or you know like reduced bandwidth
and could take additional leads but they don't know how to sort all that sort of shit out
vast majority the best the big chance is like if you listen into this and you can kind of
slowly start to piece these pieces together,
I feel like there's ample opportunity even just locally,
again, small and medium-minute prices in your areas and stuff like that.
It's all ripe for the token, in my opinion.
100%.
Now we're going to go through our stack real quick,
and then we're going to go through underrated, overrated.
I have a lot of opinions on this, but I'm going to try to keep it level-headed because i've had a lot of
opinions this whole entire time but my stack is right now i've got opus 4.6 i'm accessing that
through co-work i love the co-work interface it's amazing it's so easy to integrate i have got i've
got all my projects i'm working on on the. I've got my entire knowledge base. They just dropped like a context window and memory update as well this
week, yesterday, that like expanded the memory by a large margin. So I've noticed that my
conversation history isn't compressing as quickly. So I've got much longer conversation history.
history isn't compressing as quickly. So I've got much longer conversation history. I'm using
Codex 5.3 as my staff engineer. So I'll actually write my PRDs with Claude via Cowork, write the
PRD to the PRD folder, have Codex review the PRD, send the feedback back to Cowork. Cowork reviews
the feedback and it says, yep, good to go. And then I'll have Codex actually implement the code. And I like structure my PRDs based off of the kind of
vertical that I'm working within. And then I create a multi-tree work. This is going to sound,
I'm sounding probably schizo and ADD as fuck, but I'll create a multi-tree workflow where then
Codex actually spins up multiple agents to complete and parallelize
that development process. But again, the only reason I know how to do this is because
I've got like, I've been playing with this shit, you know, for the past two weeks.
Then I got, yeah, I said a cursor and coworker on there. Um, Gemini for like, kind of like
supplemental Google search. Um, and then a whisper for dictation.
Mike has a tip right here.
Yeah,
continue skills for context
window. Okay. Nice.
Yeah, I mean, there's all kinds of different
skills and I even skipped over
the the part right here where i mean people were talking about you know building all kinds of new
skills um there's no shortage of skills out there and different workflows to try and test
just kind of figure out what fits for you and i found that this workflow fits for me because it
gets very granular and helps with my education on what i'm actually building um but i'm curious to hear how are you guys doing it yeah i mean like for people first
like figuring this out as well like do the opposite of what you're currently doing so
instead of you constant consistently asking like these models what you should be doing like
get the model to ask you questionnaire style it's going to set out 10 15 questions of what you should be doing. Like get the model to ask you questionnaire style. It's going to set out 10, 15 questions of what you,
it can then kind of extract the best amount of information
directly from you with the most strategically
like structured questions.
I think that works really, really well.
If you're kind of like stuck at the loose end,
which way should I go with this?
I've currently like the main thing I'm trying to figure out now
is, like, my full open claw setup in Telegram.
So the current setup is using Sonnet 4.6
because that's drastically reduced the cost of what I was actually doing.
I was running it on Kimmy for a while.
It seems to get quite buggy when I'm running Kimmy,
for whatever reason that is.
I really can't figure that out.
I've got it locally plugged in to claude code in case anything goes wrong so just kind of like debug
on the back end there if anything just crashes because it can be a little bit buggy and
still i still think they're working through that quite a lot um i'm just trying to like
through the average day i'll be jumping on podcasts i'll be jumping on like client calls
i'll be jumping on trying to like do strategy we've got a lot of events coming up like
wearing a million different hats um can i just reduce bandwidth across all of those by delegating
out some of like the repetitive boring tasks that i know i put off all the time but then they end up
backlogging and i just end up like having to like actually strategically work through them and it
just takes way too much time so like anything that can help automate away.
I'm putting in the hard yards now.
So it's just like repetitive.
Don't want to do it.
Is it like digital?
And can I actually just physically do it by getting an agent to do it and
then separating out in telegram,
in a telegram group,
just one agent,
but basically giving it different skills across different,
I'm calling them departments.
So whether it's like podcast prep notes whether it's kind of uh content strategy ideas whether it's social media repurposing whether it's like potential bd leads blah blah blah and then
separating it out that way so that's what i'm working through currently um it's going good
i just feel like there's a constant update with
OpenClaw and there's this constant battle of
Claude
looks like it's getting closer and closer
to being able to do everything that I needed to do
like the scheduled tasks have just dropped
like much greater memory
as you just kind of pointed out
I feel like they're slowly just drip feeding that out
like on this
I guarantee they've got every single feature that they need and they're slowly just drip feeding that out like on this perpetuity.
I guarantee they've got every single feature that they need and they're just going to keep that gravy train rolling.
So it just feels like they're constantly shipping stuff.
And then the end product just looks like an open core anyway.
So I'll be planning on it.
And I mean, they've got data on how we're using this.
So it's like if we're all creating our own interfaces to manage our agents we'll just drop cloud co-work
right like yeah and they dropped remote access this week as well so if you've got right got
to run on the background on your on your local machine you want to tag into it but via your
mobile then there you go yep coming for you coming for you open claw if the people what they want
is is what they're doing.
Yeah, Mike's asking why Codex over Claude code for building.
I mean, for me, I usually start off in Claude code
just because it's been easier as part of my workflow,
and then it'll just hand off to Codex,
and it'll fix everything that Claude code got wrong.
And generally, it'll just keep going for like an hour or two,
and I don't have to interact with it at all, which is pretty nice.
I don't know about you, Brody.
I don't know.
I feel like every time I've had Codex review a PRD, there's been technical gaps in what Opus has written and considerations and tradeoffs that were not considered.
and considerations and trade-offs that were not considered. So I kind of like to
have a different opinionated model for different tasks, especially if the tasks intersect.
I mean, it's like if I duplicated myself, right, and I'm working and I tell myself to do something,
myself is just going to think, oh yeah, I should just go do that because that's what i think right so i again yes to mike's point he just made another comment
so i so i pit them against each other and i and i steal man what the other person what the other
person oh my gosh i can't believe i just did that i literally can't believe i just did that
i steal man what the other agent says not person cheese um you're getting one shot you're gonna be
in those naughty naughty folders on on chat to me too no country for old men i saw your header
i looked at your profile gran i was like the no country for old man header one of the greatest
movies all time um but yeah moving on over versus underrated. And I think this could be a whole show as well.
At some point it will be maybe, but, um, for me, my overrated is these video models. I've already
crashed out to Josh about the video models. Like I already went on my, my, my rant on that. Um,
open claw, I think is overrated during the time that I was actually creating these slides.
I think it might be underrated because I need was actually creating these slides. I think it might
be underrated because I need to get back into it. I just was having difficulty setting it up
in the beginning. Overrated is doomerism. And that's more of like just putting it there to
remind myself it's overrated as opposed to actually thinking it's overrated.
as opposed to actually being this overrated.
And then Gemini plugins,
like those things are trash.
Underrated perplexity investment research tool.
Pretty cool off the rib from first testing.
Optimism as a competitive advantage.
LTR is big on this.
And then belief in yourself.
Underrated.
Dude, you can do anything you want in this world now. You need to believe in yourself andrated do you can do anything you want in this world now
you need to believe in yourself and your ability to go do it that's underrated
it is underrated it couldn't be easier now if you can speak or type you can do it
if you have a if you have a brain you can do it. 563, what's your overrated under?
These are some good ones, man. I like the call out for Doomerism.
I think Doomers deserve what they have coming to them.
Just believe in yourself, like you said here.
Yeah, all it takes right now is an idea.
The execution itself, you don't have to be super technical.
You just have to try hard enough at this point.
You don't need to outsource anything
yeah i'm with you on the video gem models i think they're a bit overrated currently like i still
haven't tried to see dance i'm seeing like everyone's turned into announcement video
creator which is cool um it's still not there yet for me. Like when you actually try them and use them.
Underrated is like voice dictation.
Like it's just the best.
And ChatGBT's model dictation.
So if you're just speaking into not the voice model,
which is shit,
like the just like the voice to text dictation model,
ChatGBT's is way better than Anthropix at present.
Interesting.
I'm too lazy to type now, so I just do that.
My thoughts don't move as fast as the agent building does,
so I'm still pretty good with typing and collecting my thoughts.
Otherwise, the prompt just gets so messy.
But I guess I could prompt it and be like,
I guess fix this prompt before you send it?
Yeah.
Optimize for being lazy as fuck.
Yeah, we'll know nothing
and we'll just have infinite intelligence on our hands.
I guess that's the meta.
But that's it, guys.
Thank you so much for tapping in with us.
This has been really fun.
Really great kickoff show.
Thanks to the market and the Frontier Labs for providing us with an abundance of material to have a conversation about.
I think this will be a cool series.
I had a lot of fun.
I'm so excited to have a bunch of cool guests on.
I think there'll be a lot of people interested in coming to hang out and having a chat. A lot of people that we haven't really heard from either
who have been building their own unique and cool things. So yeah, you can find us every week
at 10 a.m. Eastern on Fridays and find us at Blockmates. Make sure that you subscribe to the Blockmates newsletter, which
the AI one comes out now every Friday around this time as well. A lot of the information that we
covered is in that newsletter and it's in written format so you can take some time to digest it.
Hell, hook up your Claude, hook up your coworker to it, start ingesting those Gmails,
hook up your co-work to it start ingesting those gmails start start giving up your privacy for
for uh start giving up your privacy for efficiency you know there you go love it yeah
great the uh i did below this if you're watching this live on twitter you will find the newsletter
link where you can just click it and join it's for for free. There's like 17,000 other people in there. So if you want to join those guys
or be permanently in the...
I'll demote it to the permanent underclass
if you don't subscribe as well.
So yeah, just be wary of that one.
No, we're being optimistic now.
No more doomerism.
That's overrated.
Chris, thanks fellas.
Sweet, so fun.
