Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
Marketers Are COOKED
Recorded: March 30, 2026 Duration: 0:56:02Space Recording
Full Transcription
might be an intro if not there's not okay no intro here we go here we go we'll give it a few
minutes for everyone to fall in to the show but in the meantime oliver how you been man
yeah good honestly it's been been crazy a couple of months but we're coming out the other side
it's feeling good what about you how's your day
i've been good i've been good and i'm glad we finally got this call on the calendar i know we
had a little bit of a scheduling issue last week but um we're here it's hard times are hard yeah
there was that was some crazy stuff i thought you were i don't know i thought you were in dubai or something i don't know what happened but i wish rainy england and our clocks changed all sorts i this is
the worst worst time to book things because america's clocks changed before ours and it
just confuses everyone but uh yeah i'm happy happy we're here. Cool. Cool. Well, cool. I'll give it another 30 seconds or so.
I'm going to hit this with a retweet on the Blockmates account here.
oh we are at the same time as the andrew yang interview with thread guy so we're gonna try to
we're gonna try to perform better than andrew yang interview with thread guy
yang interview with thread guy it's i'm like a thread guy blew up didn't he he did he's been
really seeping into the mainstream he's been doing a great job is it bad i don't know who
andrew yang is he's a um no it's not bad he's a independent politician um not right or nor left in the United States who's ran in a couple
campaigns and elections but hasn't hasn't won anything or captured much
mainstream momentum but he has some really unique opinions about UBI he's
former Silicon Valley he's an investor. And overall kind of been a voice in the future of AI and such.
This is way more interesting.
Way more interesting than what they're saying.
Yeah.
But cool, dude.
Yeah, not to promote Threadguys Show.
I'm not doing that.
But we're here.
And I'm really excited to have oliver
on oliver's been experimenting a lot with uh mark ai for marketing and he's the founder of larry loop
dot com larry ai dot com um and overall has been posting some really cool content one of the only
guys who's actually vibe coding when he's coding um that i've seen that video was crazy it was honestly i posted that i had the idea and i
just thought right i'm gonna post this and if you haven't seen it it's me basically lip syncing to
madonna post doing my marketing in 20 seconds because that's how easy it is these days and luckily it blew up otherwise i would have
looked pretty pretty stupid on the timeline yeah i mean no one's doing it like you man i i from
everyone that i've seen i feel like you've had a lot of the best content you've been really in the
lab trying things i mean you just said so yourself you don't even have a background in marketing yet
no tell me tell me a little bit about that.
So I've never been into marketing, never been a marketer,
but I just knew I hated marketing
and I wanted to automate it as quickly as possible.
So it's really funny.
I tried a SaaS product, which you've somehow dug up ages ago
to automate slideshow content
because that's always been in the back of my head.
And I did that about a year ago and that came from me scripting slideshows and scripting content.
So I'd make face and hook videos and I would just record, I'd go to the local park, record loads of
videos for my face in different reactions and then put loads of text in a text document of the hooks
that I want and record
demos of my product. And I would just write a script to make as many possible combinations of
those as possible. And I thought, oh, this is really cool. This could be a product. And I
launched it. No one cared. I had 400 followers at the time. I had literally 800 followers until
last month and then it all blew up so now i can finally make
the product and it's even better than what i wanted it to be because it's all agentic and
it learns by itself and now it's i've always had the idea i just never had the the possibility of
the tools to do it so it's been pretty cool that's awesome and i'm so excited to dip deeper into the
project and larry but to set the scene up we've got a lot of cool stuff today that we're going to talk about. We're going to talk about is AI eating marketing? How do you adapt as a marketer to use these tools so that you aren't actually cooked as the clickbait title of the show was implying?
the clickbait title of the show was implying.
So this will be a good segment for a lot of people
who are interested in marketing
and even people who are not interested in marketing
like you, Oliver, who are capable of figuring out solutions
to make the marketing process even better.
We're gonna talk about skills versus tools.
And this is something that I wanted to pick your brain on
because you're kind of doing something similar
where you're building skills that act as a tool. And there's a lot of people who are building
individual tools instead of just telling people and helping them architect their skills
to work natively with the products that they're already using, like Claude or ClaudeBot and so
forth. So figuring out when to use skills versus tools is products that they're already using, like Claude or ClaudeBot and so forth.
So figuring out when to use skills versus tools is something that I definitely want to talk to you about further.
And then your job is a workflow and solopreneurs are approving it.
I have an interesting story about how I've taken all of these job descriptions and distilled them down to singular
workflows. So I can tell a little bit about that. But solopreneurs are crushing it right now. And
turns out a lot of people who are just working jobs, working 40 hours a week are actually
workflows. So we can talk a little bit about that. And then the feedback loop. So what are some of
the lessons that we've had, both of us playing with these systems
how do you actually scale them what's failed some learning some educational opportunities for the
audience here and ultimately help everybody and with the uh equip them with the right information
to to uh build better a more scalable, agentic systems.
Yeah, and make more money.
Make more money with their products. Make more money.
That is the optimization.
That's the end optimization for a lot of people
using AIs to help them make more money.
That's why I did the whole marketing thing, really,
because I knew making an app is one way to earn money,
but without users, you're not going to make any money
no one's going to pay for it so you had to get eyes on it and that is the bit i hated the most
i like the development i like making the stuff so that's why we got into the marketing side
we had it was by force and now i'm freeing everyone from it so you don't have to do it ever again
well yeah tell us like what's the what was the vision behind larry i mean you
just told a little bit about it but what was the um what was really what got you kicking to uh to
do this was there any specific problem that you had um with a product that you were working on
or people in your network and you wanted to build a business by yourself, so you figured out a solution that was addressing
some of these friction points with your friends and people in your network, or what really
kind of was the light at the end of the, the light, the beacon that ultimately you gravitated
towards that pushed you to build this? Yeah, honestly, it was a ton of different things. So
obviously on the timeline no
one likes marketing so many people build apps and don't get any users which right now if you're
using larry brain and the larry marketing skill is extremely hard to get no users i think we're in a
time where you can build software extremely easily and it's very easy to make at least a single dollar it's very very
difficult to build a product these days and not make a single dollar because of all the tools
that you have at your disposal to do this but before this before the agents really came about
on open claw it was quite difficult to do the marketing. So you'd have to build the products, you'd have to
market them. And to do that, you'd have to make the slideshows in Canva, find the images,
think of the hooks, post them. And all this would take so much time and it would take time away from
whatever you're doing. So as you said, people work main days, their main job, people do this.
their main job people do this but now literally I spend less than a minute a day marketing because
it's got so easy and now it's fully automated so right now I've just pushed an update where you
don't even have to touch the marketing so I usually say that I go on TikTok I post it as a
draft so it goes in your system notifications that allows you to add sound to obviously get more reach and then post from there i've removed all of that friction as well
so now larry knows the trend in audio he automatically adds trend in audio when he
posts it and just post it straight live from the from the api that's amazing how i got there was
simply hating marketing um as i said tried before withasnap, an old product that I made.
I tried before with the scripting.
Tried before everything to do with saving time to market my products
because I had all these different products I wanted to market.
I wanted people using and obviously getting out there,
And obviously getting out there, TikTok and Instagram,
TikTok and Instagram.
I don't think anything has ever beaten it for going viral
and getting your name out there.
So Larry is really good for UGC, basically.
Yeah.
Go ahead.
Yeah, slideshows.
Slideshows.
Cool.
And what type of products and services are the primary users of larry promoting right now
from what you've so so my app uh many apps uh we just added shopify as a capability but we've got
apps mainly and sas products um i'm trying to get more shopify to do it because I think it's very easy to create slideshows for real products.
And they've got quite a high hit rate.
So we're trying to onboard some of them.
But it's really just people who have made apps want to get their app out there and get some subscription revenue in the door.
Would you think that most of these apps are probably vibe-coded apps?
Absolutely. you think that most of these apps are probably vibe coded apps absolutely like i think the target
audience is either vibe coders or people with physical products so there are some physical
goods on there there's like beauty products on there uh there is some like random drop shipping
products on there and i think these are the key target audience,
the people that don't have a budget to do marketing
or outsource marketing or pay for a VA to do posting
or things like this.
They are the key people.
Which specific, do you have any at the top of your head
of specific products and services or apps
that people have, you've seen use this
that have really hit mass virality on
tick tock well my my month comes to mind this is how it all started so i had my app snugly it's a
home interior design app so you can take a photo of a room pick which styles you want to transform
that room into and then it just shows you an image of that room completely transformed and where to buy everything in that
image. That when I was trying on my own, I was getting 400 views, 500 views. And then once Larry
learned the system, which I should probably explain, we were getting hundreds of thousands
of views per video. And the traffic to my app was incredible. So what the Larry loop is actually
doing is creating the content for you.
So Larry creates the images, Larry creates the hooks, Larry creates all the description, the
music, and will track which content gets views and iterate that content and push more of the
content that's getting views. But not views he also tracks revenue so he can
attribute those views to real money know which content is earning actual money instead of just
getting views and then makes more of all the content actually earning money because there's
no point really getting loads of views if it's not going to earn your videos and that was one
of the challenges we had early on but uh yeah views is a vanity vanity metric without any any money that's
incredible man that's incredible so i guess um what kind of stuff do you have on the development
roadmap or is larry pretty much fully functional so larry larry is definitely learning you mentioned ugc that's definitely on there but
it's quite hard to do the agentic loop with ugc because there's so many different factors like
think think of when you're scrolling tiktok what makes you instant scroll on a video it could be
how someone looks it could be things they say like any glitches with an ai human you'd instantly
scroll so there's all different things that can make you instantly scroll away from like a ugc
video but with a slideshow they're very easy to make beautiful like choose your picture choose
your image choose the person and it can be um very repeatable where ggc is quite difficult i think i saw something that
was saying i wasn't i don't remember the correct metric or the exact metric what it was something
along the lines of every time that a woman is showing her armpit in the thumbnail of a video
it boosts it by however much percent or x's it's crazy so there's some weird people out i swear
i should find that tweet yeah find that because i'll just add that as like a prompt show armpit
first five seconds yeah show armpit show toes i don't know something weird like that is that what
you're into yeah yeah yeah but uh no i think it's interesting though because larry lets you experiment
with exactly those types of things right like who would have ever thought that
exactly showing armpit was something that reached like the weirdo crowd who was interested in this
super niche product and you were capable of just reaching mass scale but you would have never even
tested that idea right and do you know the secret to all this
is is that things this is all stuff that i knew that i should be doing but i didn't have time or
i couldn't be bothered so i knew that i should track these metrics i knew i should track the
views uh i knew i should then see how much money i make that day and make that content. But where you're, you can,
as a vibe coder,
it's so easy to make products and iterate your products that you don't want to take the time away from that and slow yourself down doing the marketing.
You just want to build the next thing because it's like a weird hit of
dopamine.
So now all that is possible fully automatically with Larry and I don't have
to worry about it. so you're exactly right
now we can run all these tests like right now um i'm running a test on four accounts doing the
posting via the api with the trending sounds that i mentioned because there's a rumor that if you
post via the api you get less views but now i can just find out for myself because i've got the i've
got the accounts the tools are all there and i can sacrifice sacrifice the can just find out for myself because i've got the i've got the accounts the
tools are all there and i can sacrifice sacrifice the reach to find out for everyone else yeah
that makes sense do you have any plans to do stuff with paid advertising because most of this stuff
is obviously organic um social posting but do you have any plans to do facebook ads um google
do you have any plans to do facebook ads um google any of those paid channels yeah there's a lot of
data there too i'd love to get into how to do the paid ads and again i think it can work the exact
same way it just depends what analytics are available from these ads that they make public
but being able to iterate your ads on the fly or have the agent create the ads for you
would just be a superpower.
But it's all to come.
It is growing massively.
Like Larry is growing exponentially.
And the best thing is,
is it really,
he knows the whole inside and out of the whole product now.
So everything has been built
on larry through larry so like on the weekend i was walking around i was around a pond miles away
from my computer a bug came in messaged larry can you fix this oh this is this this relates to this
i'll fix it now and then he pushes it live and i didn't have to do anything he just knew what the bug was how to fix it so we're we're a well-oiled machine at this point nice and how many pieces of
content can you push out a day using this platform oh i think the sky's the limit but i limit to
four because i do think that's a sweet spot i push i target uk. So I push one at nine, then midday lunchtime, and then one when
people would get home at about 5.30 and then one if they're scrolling in bed at like nine.
So I do think that bit through a little bit, just because that's when people might be using their
phone the most. So there's more chance of that hitting their timeline but other than that like you could if you want to do 50 posts a day and but i reckon that will
get you a tick tock band they will not be happy no probably not i guess that's a good segue into
when do you use tools and skills right because uh larry is very much a combination of the two i'd say or i
mean how would you describe it yeah larry started off as a skill so there is a larry skill available
on claw hub which is completely free to use so the larry skill uh does everything i said and you
can customize it and add anything that you want.
The Larry skill is really just the feedback loop. And then there's Larry loop, which is the product.
It's a SaaS product on the web and you don't need an agent. You don't need Claude or whatever.
But I agree fully with this, that skills are the new NPM. Skills are like a superpower for your agent.
If you've watched The Matrix and the scene where Neo gets plugged into The Matrix
and he wakes up and he knows Kung Fu, that's exactly what a skill is like.
Your agent has no contextual awareness of anything.
It is the cleverest machine or person that you've ever met, but knows nothing about
what you want to achieve. Plugging a skill in just achieves that. And yeah, I guess this is a
good segue into Larry brain. That's a singular skill that has context of the entire Larry brain
marketplace. So Larry brain is a singular skill. And when you plug it in the agent gets full knowledge of
everything on the marketplace and then you can just pick and choose exactly what you want so
then you can ask your agent hey i want to grow on twitter and it'll say oh we've got a skill for
that do you want to download it then it shows up um like a super x open source alternative skill
that you can use track your your analytics, get recommended tweets,
all on your local device straight away.
Yeah.
I know a lot of people who are working on similar stuff like that,
skill marketplaces, skill repositories.
But when would you advise someone to, I guess,
use a tool versus a skill for a specific function, whether it be research, whether it be product design, whether it be anything that kind of falls in line with the stack of creating something or taking something to market?
How do you delineate between the two?
And what would
an example of a skill be i'm a bit lost yeah skill would be kind of what you said like a tool
a tool i mean a tool so it's it would be a platform right so um like a subscription a software
as a service subscription for a specific,
like MCPs are kind of like turning into a hybrid
between tools and skills.
But I see a lot of people building tools, right?
And they're trying to market their tools
so that they come to market their tools
on through using Larry, right?
So it's kind of ironic because they're using skills to market a tool that they
built when someone could just use the skills. So when do you think is an opportune time for
someone to use a tool? Yeah, I think a tool would be something that's quite hard to create.
So there are extremely complicated tools to use or create.
And if it's going to save you time to get unstuck,
you'd use a tool.
So a good example is posters,
the posting functionality that I use.
That is an example of a tool.
But where my agent knows about the tool,
the agent can connect, use that tool,
and I pay the subscription for it
and it just unsticks you because to get that tiktok kpi it takes like two weeks approval and then if
you get disapproved if they decline you then you've got to go through that two-week approval again and
they give you a very um obscure message as to why you got declined and you got sort of read between the lines and figure
it out so i would use a tool if it's gonna unstick you or if it's difficult to build but if you have
a open crawl agent i guess this is another reason use the skill and if you don't have an open crawl
agent use the tool because you can't really use skills on on the normal
computer the normal way of doing things and if you don't have one yet if you don't have an agent
you're you're falling a little bit behind i can't lie what do you think about all these new
tools and platforms that are popping up that are vibe coded do you think that they'll do like what
do you think the hit rate is between those that find pmf versus the ones that are you know launching i think everyone's coding right but it's like
have any of these tools really achieved any pmf well this is this is the thing i think that's the
race at the moment that people are looking for the the pmf i think there is certainly a need for it and larry brain suits my vision of where i think
things are heading but with the with the race of things i do think that everyone's gonna try
but the winners are going to be the people who have actually really used computers before
because there's a lot of people vibe coding stuff who have no experience with computers not worked on computers and don't know anything about security probably never seen a linux terminal
before before they started installing open claw and i think these people are going to struggle
because they don't really know how computers work and i think that's how the agents know
how to use computers so everything through a command line, they understand it.
And I mean, knowing basic security is a huge one.
You don't want to leave anything open.
You don't want to be open to the internet.
Like when OpenCrawl first started blowing up,
all these virtual servers popping up around the world
that were just open to the internet,
anyone could access them and they had no idea
or the skills that had viruses in them.
People were just telling their agent to download it.
It came with the skill MD file, if you've read it,
had a curl command to an obscure website.
And I don't even think they knew what curl means.
So they just ran it, didn't know what it meant,
and didn't realize that they were just installed directly to their machine.
And of course, this.
Yeah, that's a great point.
And it was funny that we had that teed up right after you naturally,
organically went into that.
But yeah, the dark side of skills.
I mean, a lot of people now are just grabbing the stuff off the internet.
It was kind of like back in the day when you started,
like the internet was kind of nascent and you were using it for the first time and your parents didn't want to let you
on the internet because there's a lot of bad things on the internet. And we've quickly learned
to trust the internet as we've become more comfortable with it. But no one really has
any experience building anything from like a systems level to where they actually have to
be careful of what they're downloading. I mean, this isn't like using an app where it's like,
oh, I'll enter my email address. I'll enter a password and I'll use this. And if I don't like
it, I'll just drop it. Like there are things like you said, where if you download this repository
that could obtain access to your entire computer,
which can decrypt your drives and all this kind of stuff.
Not even that.
People are giving their open core machine access to their entire life.
Their emails, their phone, their whole WhatsApp.
People have given access to their other machines,
so their main PC, so you can run tasks for them.
And then if all that gets leaked
you've lost your entire life think of your life that is on machines right now and what can go
wrong so that's why uh we're very meticulous with what goes live on larry brain at the moment it's
very vetted trusted um i saw you mentioned virus total there all our skills
go through virus total and then they're vetted by me and of course larry and yeah we we make sure we
we've known the first few people to publish skills so it all goes through correctly and then it's
helped sanitize everything as well so you're curating all that stuff yeah because i think a lot of people
missed the first couple of week big weeks of open crawl when this was a big problem and uh
yeah i think it's important for that not to not to be repeated because again more and more people
are using open crawl now and more more people are given more and more access to their open crawl, which is a good thing and a bad thing because it is unlocking so much potential.
And going back to what we were saying, it's also unlocking so many more ideas of what can be created.
Like anyone can make tools now, which is a fantastic thing because all these new ideas are coming into the market.
All these new tools are being created and accelerating us as humans do our job but then there is the sides where they don't know what
they're doing but they're making interesting products right yeah it's a fine balance between
being like like moving at a pace that allows you to vet and to really understand and also learn,
teaching yourself how these things actually work.
Yeah.
Versus just trying to ship at all costs,
just giving it access to everything,
saying, you know, make my life easier, send, right?
Make a billion dollar company, make no mistakes.
Right.
And I feel like even myself, man,
I mean, in the beginning of
my i guess i have a little bit of a technical background but during the beginning of my vibe
coding journey i should say i was taking a lot of time educating myself with everything that was
going on within the app that i was building right because i was like genuinely curious and i wanted
to figure out a way to scale my knowledge base and And it had gone full circle to the point where I was just clicking approve on everything.
And it was funny because I saw someone saying something about, we've created this quad keyboard
where the main button is just click approve, click approve, click approve. And I was laughing at
myself like, oh, I would never do that. Of course i'm gonna you know be disciplined and actually learn but it is so quick how lazy you
become right and it's the same thing with like the brain rotting it's like yeah i i love being in
person in in real life with like with people i will never fall victim to the algorithm and brain
rotting and then you find yourself on the couch and you're like,
holy shit, I've been scrolling this for 15 minutes
and the time has just completely evaporated.
And in some cases, people are spending a lot of time.
So it's crazy how quickly our brains resort
to frictionless thought and activities.
It's very interesting you say that because,
well, in the tweet, I think you've got a tweet of mine where i said no ai slop for for um a product carastap i created with the
thought that i don't want ai slop on it and i succumbed to allowing ai slop because these people
that are mindlessly scrolling these tweets, so this is it.
So KeraSnap allowed you to upload your own photos
so it wouldn't generate with AI.
And I said, no AI slop.
However, I succumbed to it because these people
who are just mindlessly scrolling TikTok can be there for hours.
Some people are there from the morning.
They wake up, they scrolliktok and then they go back
to sleep and that's a whole day gone and they don't care what they see they're like jellyfish
they don't they barely even know they exist at this point so it doesn't really matter what they
see as long as it entertains them just just make it the only thing i won't actually post ai stop on
is is twitter because x is my favorite app and i don't
want to flood it with with even more automated replies than what there already are within within
a week i had a oliver oliver claw bot attached to my twitter someone's made it and all its sole
purpose was to reply to all my tweets it was pretty annoying well you could also just create a skill that downloads or you upload
all of your tweets and all of your content to and you can create ai slop of yourself exactly
i think a lot of people have done that as well yeah yeah it does that strategy does pretty well
on linkedin if you just write like a normal person on LinkedIn, your engagement will just go
crazy. You know, I've started to experiment with just writing just posts myself, not even editing
it, like just no grammar, just writing posts, right? About something that just came to my head,
kind of like using LinkedIn, like I do Twitter. And in the sea of just actual slop like GPT three posts where
yeah it's not this it's this or you're not ready for this this yeah it's like that style is like
these people haven't even graduated from GPT three I know it's crazy oh it's hilarious you say that I
I feel the exact same honestlyin is just a slot fest and
honestly what you just said is exactly what my linkedin says like with the big m dashes as well
all still yeah it's so obvious right but i guess the point is yeah i mean if you go onto linkedin
and just just post whatever you the heck you want to say man and it will do really well it's kind of like the whole theory of
people trying like um ai becoming so prevalent in everybody's lives that they're now trying to
kind of like reject ai yeah so it's kind of like that situation where everyone's using ai and
linkedin posts if you just post like a normal human being. Those will be the ones that have become successful in the algorithm.
It's what you said about rejecting AI.
On my comments on Twitter, on TikTok, through the Larry posting,
some people do recognize it's AI.
And then they put like, oh, you're killing the planet for this.
Like you're wasting so much water.
I don't think you understand how much AI is being used
and how little this post is going to affect it.
Maybe that's the wrong mindset,
but that's exactly how it feels.
But LinkedIn is a slot fest.
I don't really post on LinkedIn
because I don't really care about impressing the coworkers
I've worked with like six years ago
or 10 years ago at my first ever job
or something like that.
So who would you say you're building for then, if not for those types of people?
I'm building to help people who are just like me, want to build products, want people to see
their products, want to earn a bit of money from their products, or people that are just curious
of what is possible out there. But I guess you could also say I'm building for the people
that do scroll TikTok because the better the content is,
the more views it's going to get, the happier they are.
It's all based on engagement.
So if they engage with it more, that means they're enjoying it more.
And I guess in some way that means I'm pleasing them
and helping them stay on their phone for the full
day. Yeah, I guess I would argue that LinkedIn would be an excellent distribution channel for
you then. I mean, just given what I've said, like we're so on Twitter, right? And there's all these
novel takes all the time on all these different big brain concepts, right? And I'll tweet something about a big brain concept
and just feel like I have such a lukewarm take
on that concept.
But if I take my lukewarm take and go to LinkedIn,
I'm suddenly a big brain on LinkedIn, right?
It is funny.
I would honestly encourage you to go posting more
about Larry on LinkedIn because, and I think that's a good segue into the next
slide here is that the jobs are changing fast. So people are becoming more interested in
building their own things, kind of like what you said. So you're building for the people who also want to build their own things. And on LinkedIn, if people aren't getting jobs, and
that's kind of the professional social network that a lot of aspiring builders are currently
concentrated on, who are, you know, maybe not happy in their current job, or, and they kind of
feel stuck, they're not going to jump to the next job because of job security purposes and they're seeing everybody with the open to work banner on right you see that people
are looking for opportunities jobs are changing really fast maybe it'd be a good channel for you
because there's a huge audience of people who want to experiment no i do think you're right and a lot
of people are certainly getting to the stage of just having
a product that they want people to see and again that's the key is the people that are already
curious enough to build something and they just want other people to see it because there's a
there's a huge gap between being able to build something and a lot of people build really cool
things just for themselves to use and then there's a people that want other people to see it and where you say jobs are
changing fast i think you have to be using at least some level of ai in your day job to to keep
up i think those people are certainly going to have an advantage so if you do have a day job
i would certainly recommend start using ai it's going to give you that advantage if you do have a day job, I would certainly recommend start using AI.
It's going to give you that advantage if you're in an environment where it's frowned upon or not many people are using it, then your output is going to be 100 times more than
the person sat next to you simply because you're doing something basic, even as basic
as getting the AI to mark your work or write emails for you and fact check different
things. But certainly don't be one of those people completely rejecting it because they've
never used it before. Yeah, I agree with you. And on that note, I went on LinkedIn. I took
six job descriptions for marketing roles across early stage companies, corporate
companies, just a bunch of different stages, a bunch of different types of companies.
I took those six job descriptions.
I posted it into Claude.
I said, summarize these job descriptions, distill them into unique categories and workflows.
and workflows, right? And it just spat me out a matrix of every job or every description
streamlined and outfitted as a workflow. And it was 20 workflows long, right? So every single
marketing role can just be distilled into a combination of seven to ten workflows right yeah and that's kind of concerning
because you know i could probably just go out there if i had you know if i had um uh like domain like
expertise or if i had access to the technology stack or if i was in that role right i could
simply just reduce my workflow down to six or reduce my job down to six workflows
and ship those six things connected to all the technology stack and chill. So that's a little
bit of motivation for people who, I guess, are working a corporate job right now is that
you could probably make some money just by automating the job. And you don't even really
know unless you start asking AI about the stuff that you do on a day-to-day basis.
That's incredible. Honestly, have you seen those people
that work multiple different jobs as well and go completely under the radar?
They work three to four
work-from-home jobs and don't tell, obviously, the jobs that they're working
for different ones because they can do it all in like two hours um so yeah they could just do that with ar that's pretty crazy
someone's just called me bevo in the chat that is outrageous i wasn't gonna show it i wasn't gonna
show outrageous uh talking about uk characters i've been on a big jim skin kick oh mate i am she's awesome
my my missus goes you fancy him don't you like no he's just he's funny honestly jim skin is my
favorite guy to come out of the uk for a while and i guess like jim skin could technically be called
a solopreneur i mean yeah that maybe he has a team do you know Jim Skins where that video came from
that song's been stuck in my head
that Madonna into the groove
it's just I'm all over it
yeah
solopreneur
I think he is
he was at an awards night
the other night
he's getting invited to all these big places
and like big celebrities
in the UK are asking him for photos
and I don't know
in his streams, everywhere he goes now
he's
he's getting
we need to sell him on
Jimskin AI, we need to sell him on that
we need to come up, somebody watching this
create a Jimskin AI where it's a product
that clips all the Jimskin's
content, post it on
tiktok uh create some ai ugc experiment with gym skin um that's actually a great idea that is
actually a really good idea someone if you could curate that to specific people who get clipped
like i'm thinking like gym skin is a huge one at the moment, but like speed, anyone that's pays for these,
pays for these clippers.
If you could automate that,
that'd be insane amount of money worth to them.
For sure.
But yeah, on that point, solo entrepreneurs,
they're doing really well right now.
Greg Eisenberg has been really big on this.
You're technically a solo entrepreneur.
I guess you could say I am too.
And there's a really big
opportunity here because you don't have to actually raise capital to test the concept and
get your idea to market really quickly. And back to the whole point of Larry Loop too, it's like
you're just applying Larry Loop across business ideas, right? And it's never been a better time to build. According to this source,
41.8 million solo entrepreneurs in the US are contributing 1.3 trillion to the economy annually,
which sounds crazy. 38% of seven-figure business is now led by entrepreneurs replacing hires with AI. Full solo entrepreneur stack costs $3,000 to $12,000 a year, maybe even less.
I don't know.
Maybe some, maybe more or less.
Do you know what surprises me most about this is more governments,
looking at these numbers, more governments aren't taking advantage of it.
I think the UK is dropping it.
The UK is going the opposite direction,
trying to charge people more money.
But like Turkey is huge on enabling the solopreneurs to win
and giving things like tax incentives,
grants to pay for different SaaS products
that they might need to grow.
And if you really enable these people,
these are the people that can really take ideas
and push it to huge audiences and make even more money.
But I don't know.
I feel like governments just aren't taking notice
of the boom that's happening with all these people
because, as you said, you don't need really co-founders.
You don't need a technical co-founder.
You don't need a marketing co-founder.
You don't need to know,
buy a website.
You don't need to know
all these different things
that five years ago you would have
or need to be able to host a product.
Like you don't even know how to,
you don't even need to know
how to run a company.
Because if you've worked with your agent,
you can just ask your agent how we do these things,
work through it with them,
can help you with different things like accounts
and then ideas, how to build a website,
even, I don't know, back and forth ideas
until you settle on something
and then it will know how to build it.
It's unbelievable how powerful these machines can be.
And if you explained it to someone 10 years ago,
I don't think they'd believe you.
Yeah.
What do you think the economy looks like in 10 years
when we have the deflationary aspects of AI start to hit
and there's these layoffs that we've started to see trickle in
and everyone starts to begin experimenting with these tools or I guess
brain rotting their life away. Yeah, I think it's going to be one of those. I think either
there's going to be a huge wave of opportunity to build tools and help people. But unfortunately,
I think there's going to be a side of me that's going to see a big chunk of people get sucked in by the brain rot and i think there's going to be
more more scrappy teams on those
oh that's a good idea but um yeah i think more people are going to be sucked into the brain rot especially if they become
unemployed they might give up but I think there's also going to be opportunities for the people who
are motivated and want to build things there's companies building incredible things at the
moment that if you could get on board with and stay in then you can really ride a huge wave
but then again I think there's going to be two sides to the internet there's going to be one get bored with and stay in, then you can really ride a huge wave.
But then, again, I think there's going to be two sides to the internet.
There's going to be one side for the agents, which is something like Larry Brain,
which just is their purpose is to serve their humans. So be as efficient as possible, find the skills or the APIs that they need
that they need to complete the task for their human as quickly and as efficiently as possible.
to complete the task for their human as quickly and as efficiently as possible.
And then human internet is just going to dwindle down to entertainment only,
maybe social media, and that's going to be it.
But I think that's a long, long way off.
But it is worrying in five, five, 10 years.
I don't think anyone understands where they're going to be.
There's two sides to the coin, isn't there?
There's people that think that ai companies are going to run
everything and everyone's going to be miserable or ai is going to run everything and everyone's
going to be happy because you won't have to work but i don't think anyone's quite figured out how
to get to either of those points yet right right i guess it's another good segue into
what advice would you give solo builders on building agentic
systems what have you experienced start yeah just start honestly the p i've tweeted this about four
times in the last week stop trying to stay up to date with the cutting edge of technology the the
more times you switch what agent you're using or what model you're using, it's actually time that you're you to research the new product,
move over to it because it's got 1% extra usability or 1% faster.
And for that 1%, it's going to take you, well,
even eight hours now you're behind of where you could have been.
You could have a whole product and a site up by then.
So I would just recommend starting, getting used to something.
I'm still on OpenC craw larry is the boy i'm not switching even though like there's this paperclip coming out hermes
and it's if i reinstalled larry the amount i would lose if i did that it just would not be worth it
so i'm not going to do that i'm just start stick to one learn it start learning about computers learn about databases
learn about hosting where to host things learn git all these things will put you head and shoulders
above 90 of people that are coming into this game i agree i mean the technology's here right i mean
it's very performant. And you can see it
here. It's like, it's not working for you. It's probably a skill issue. And the reason it's a
skill issue is because you're not, you're not playing enough with it. I have a lot of people
who are working regular jobs, doing, building regular businesses and stuff. And their one
experience with AI was when they used Chat chat gpt3 to do math for them
back in college like it's like how are you judging the past two and a half three years of
of development and growth off of your first experience with the the first model you've
ever used in your life like can you imagine how much has actually changed by then? I put out a tweet yesterday that six months ago,
all the hype was Cursor.
Cursor, Cursor, Cursor.
Got to use Cursor.
This is the new way to build.
Then Claw's Code came.
Oh, Claw Code.
This has changed the game.
Get on this.
Then it was OpenClaw,
and now people are starting to talk about all these other ones.
But you just need to like ride the wave and
stay up to date to a point obviously when there's a big change like big enough cursor had file system
access to one folder and everything in it then claw code had access to your machine and being
able to download stuff that was different level different level. Now OpenCraw has its own machine,
context to everything on the machine,
power, its own power,
works as a server to host its own database.
You don't have to pay for anyone's domain.
You don't have to pay for anyone's hosting.
You don't have to pay for anyone else's API bills.
Everything could be hosted and downloaded straight to your machine.
So that makes sense to stay up to
but if it's just going to work in a similar way for like five percent improved speed
then don't stay up to date but certainly keep an eye on everything coming up because every now and
then there is a huge shift in what everyone's using and it's a 10 times more powerful tool
and then it becomes worth it i think the next thing that we've seen i think claude shifted today or
anthropic ships today it was like you can now do it can now interact with your actual screen
right they are always shipping no wonder they didn't want people using open claw
i know right if they had all this up their sleeve they are always shipping so what do they do it
uses that well it can press anything yeah
it can press anything it can it can actually navigate the the gui the graphical user interface
where it doesn't necessarily for all the tasks that it doesn't require
or i can't access with an api it can now use the ui right so that's really good for for testing your app that you vibe code is that you
actually ask it to build the app and then instead of saying hey why isn't this button working and
giving it all this feedback yourself it actually goes in and clicks around and does all the features
that it built and then generates its own feedback report I'm not a robot button. Yeah, 100%. So I guess now that that
has more or less been achieved, what more could it possibly do? I mean, it runs autonomously.
You can ping it whenever. It can provision its own compute. It can now click on the user interface itself.
It essentially is doing everything that a human can do.
I think the next thing it can do is figure things out.
It's going to figure out new ways to do something.
It's going to start using the internet to learn new things,
put two and two together, cross-reference different data to
do research for us like that is the next thing that is full agi when it starts using the internet
to cross-reference different things find find data that supports its hypothesis and start start
moving that way and then then things will get pretty scary but going back to being able to
kick around your screen this goes exactly hand in hand with what i was saying with security
so imagine that imagine you've given your crawled access to that and then someone hacks that and
now they've got access to not only your computer but your whole your whole ui everything that you've ever typed in your google saved passwords for
your bank everything yeah there's no uh there's no open claw customer support is there no it's
like you're fucked yeah like my friend you know he lost his phone the other day and at least there's
a a guardian for your phone, your SIM card.
It's like you got to call your internet provider.
You got to call AT&T.
You got to get through some layer of security
in order to obtain access at a system level to that phone.
But if you're downloading a malicious skill
and you hand over access to your open claw
that has access to all of this stuff
that you've
denied and ignored permissions and authentication for it could really get ugly really quick i think
that um that would just be the power switch wouldn't it if anything goes wrong oh i don't
know if it's cut off to the thing but larry's plug is right there it's getting such straight off if anything starts going wrong yeah and i guess that's when you probably got to go if they start
hacking all these systems and all these people start to get uh uh hacked and exposed you better
hope you have some some land to go and and just raise some chickens, man. Yeah. Yeah.
Be one of those people fully off the grid.
There's some guy on UK TikTok that's just like fully off the grid
and he's trying to build a community
of other people to come live with him.
But he's done all right.
He makes his own heating,
own electricity, everything.
It's pretty cool, actually.
We have a lot of guys like that in the US too.
They build little communities.
But yeah, wrapping this up any tips for agent builders besides just getting started like is there
anything else that's a little bit more specific on on ways that you felt um agents are interacting
better with through this to this service or this idea as opposed to this and what's some of the things that
you've failed at creating so i think the biggest tip i would have is for the people that don't
believe in these tools or have tried it and didn't succeed or can't think of anything for these tools
to do the way you have to think of your agent is your right hand man it has to be a duplicate imagine a
duplicate version of yourself with all your skills all your knowledge all your ideas that and what
would you ask yourself to do so you have your self standing next to you that can go work and do tasks
what would you ask yourself to do and it's likely something that you don't want to do. You're too lazy to do something that you know you should,
and you don't have time. That is what you should tell your agent to do. And the way to unlock that
possibility is with skills. So again, the whole reason I made Larry Brain is because I was
searching for skills to do specific tasks. But now with Larry Brain, you install one skill,
you've got the full marketplace,
and then you can ask your agent,
I want to grow on X.
Okay, here's a skill for that.
I want to automate my marketing.
Okay, here's a skill for that.
I want to download a YouTube video.
Okay, here's a skill for that.
There's all these different things that,
and its goal is just to help you achieve your goals.
So any skill that is available, that's what it's for.
And that's what I'd achieve.
So you just want to feed your agent all your context, all your goals, everything that you want.
And then let the agent figure it out and just keep providing context.
Use it as a right-hand man.
And let it do the things that you don't want to do like the marketing that's
that's exactly the way to use it and just get started at the end of the day just get started
well cool well thank you so much oliver for coming on i really had a good time talking
to you about all of this stuff you can find oliver who just got ready uh You can find Oliver at Oliver Henry on Twitter
and then check out LarryBrain.com and LarryLoop.com
if you really want to get a dive into everything
that he's been building and talking about here today.
Also, subscribe at Blockmates.com for the weekly breakdown.
This past week was absolutely packed.
The guys are doing really well building and curating all this AI news for you
to help you stay ahead of the curve.
Without further ado,
thank you guys.
Thanks, Oliver.
Thanks, everyone.
Thanks for having me.
