39:43Greg Isenberg · Interview
Making $$$ with Loop Engineering
Elie Steinbock walks Greg Isenberg through the build-verify-learn loop he's using to run SEO, ads, and product feedback on autopilot.
July 13thThe argument in one line.
AI has not shrunk headcount or the need for human judgment so far — it has shifted where value sits, rewarding people who build an audience and take responsibility for outcomes over people who only execute, while opening a short window for whoever figures out what AI agents will pay to access.
Read if. Skip if.
READ IF YOU ARE…
- You're building a product, agency, or media business and want a concrete sequence for getting your first paying customers instead of building in a vacuum.
- You sell a service (design, dev, consulting) and want to know whether AI tooling actually threatens that business or just changes what clients pay for.
- You're trying to decide how much goal-setting and strategic planning a small, pre-$20M business actually needs.
- You're curious about the emerging economics of AI agents paying for web access and want a plain-English walkthrough of why it might matter.
SKIP IF…
- You're looking for a technical tutorial on building or deploying AI agents — this is a conversational reaction show, not a how-to.
- You already run a venture-scale, funded company where formal goal-setting and OKRs are non-negotiable — the advice here is aimed at smaller, bootstrapped businesses.
The full version, fast.
Two bootstrapped founders argue that building an audience beats building software as the starting move for any new business, using a small-podcast-funded event and a design agency's pivot from execution to strategy as proof. They cover practical advice — a framework for sequencing audience, community, then product, and why service providers lose to competitors who simply follow up with something concrete instead of a generic pitch. The conversation turns to Cloudflare's new gateway for charging AI agents micropayments to access web content, framed as an early, high-upside opportunity similar to being first in social or mobile. It closes on a viral chart showing AI spend per employee climbing into six figures while headcount at AI-adopting firms keeps growing rather than shrinking, undercutting the narrative that AI simply replaces workers, alongside a case for local, open-weight models as a way to keep proprietary business data out of frontier labs' training pipelines.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Who's talking.
09:33cohostJonathan Courtney
Where the time goes.
00:00 – 09:33
01 · Countdown
"Starting soon" countdown clock before the livestream begins.
09:33 – 16:12
02 · Cold open: the Shrimp Millionaire origin story
Greg introduces cohost Jonathan Courtney and the two explain how Jonathan became 'Shrimp Millionaire' after catching shrimp in his mouth at Benihana and getting mislabeled 'millionaire entrepreneur' on a prior show.
16:12 – 31:45
03 · AMA: first 100 customers and the age of the generalist
Audience questions on getting first customers (the ACP funnel: audience, community, product) and how a graphic designer with no sales background should learn to sell; a full-time-job side hustle question closes the segment.
31:45 – 39:31
04 · Worst things to do in 2026, and the lazy-outreach rant
A reader asks what would completely sabotage someone in 2026; the answer is building software and never building an audience. Jonathan riffs into a rant about how badly people pitch him services, and floats a free-SEO-article outreach tactic using Claude.
39:31 – 47:18
05 · Are design agencies screwed by AI? (plus SEGA's surprising history)
A viewer asks whether design agencies still have a moat when AI tools can generate brand books and landing pages; both hosts argue agencies that already moved from execution to strategy are fine. A tangent covers SEGA's origin as a slot-machine supplier to US military bases.
47:18 – 1:02:51
06 · Why goals and vision statements are overrated
Discussion of a tweet arguing goals/targets aren't necessary below a certain scale, using Nintendo's directionless early history and Shopify's 'arm the rebels' mission as examples; lands on a 'one number, one sentence' minimalist goal-setting rule.
1:02:51 – 1:10:38
07 · Cloudflare's agent economy: charging AI agents to browse the web
Breakdown of Cloudflare's Monetization Gateway waitlist announcement, which lets site owners charge AI agents stablecoin micropayments over the x402 protocol to access pages, datasets, APIs, or MCP tools; framed as an early-mover opportunity as agents outnumber human web visitors.
1:10:38 – 1:18:24
08 · Paul Graham: people are going to stop reading books
Reaction to Paul Graham's tweet that reading will become a rare advantage, a rebuttal calling books an inefficient format, and a personal story about getting a daughter to read by removing her bedtime; closes on phones and brain rot before bed.
1:18:24 – 1:26:11
09 · iMessage agents and the rise of vertical AI apps
Discussion of an iMessage 'app store' directory for AI agents (Caddy, Poke, Tomo) and why narrow, polished assistants like Poke and Lindy can feel more usable to non-technical people than general tools like Claude.
1:26:11 – 1:33:57
10 · You just hired a 1M bad employees: the AI headcount paradox
Reaction to the viral article/chart (sourced from over 21,000 US firms) showing AI spend per employee climbing into six figures while headcount at AI-adopting companies keeps growing, contradicting the pure job-replacement narrative.
1:33:57 – 1:43:41
11 · Local LLMs, data privacy, and the sign-off
A viewer question on local LLM use cases and compute cost leads into a discussion of keeping proprietary business data out of frontier-model training, an Apple-style 'privacy bubble' UX concept for local-first AI, and closing remarks.
Lines worth screenshotting.
- Building the product first and finding customers second is backwards — one host funded a $130,000 event from a niche podcast that averaged around 300 listeners per episode after 40 episodes.
- The single biggest way to sabotage yourself in the AI era isn't ignoring AI, it's spending all your time building software and none of it building an audience or getting anyone to use it.
- Most cold outreach for services could be replaced by an 8-second AI-generated demo clip or a free small mockup, yet the overwhelming majority of pitches skip that step entirely.
- Service businesses that never moved past pure execution work are the ones most exposed to AI tools; the ones that pivoted to strategy and took on outcome responsibility are not.
- Clients keep paying humans for work they could technically do themselves for the same reason people hire cleaners for jobs they could do alone — knowing how and wanting to do it are different things.
- Below roughly $20 million in revenue, most formal goal-setting and mission statements are theater; a single tracked number and a one-sentence direction is enough to run the business.
- A new Cloudflare protocol assumes AI agents will soon carry their own wallets, email addresses, and spending limits, turning ordinary web pages into toll booths that charge machines instead of humans.
- The next wave of wealth is expected to go to whoever figures out what data or access AI agents will pay for before everyone else does — the same arbitrage that rewarded early movers in social, mobile, and cloud.
- You don't need to be exceptionally smart to win — a bias for action beats a great idea that never gets executed, and plenty of highly successful people are proof of it.
- Reading is framed as a compounding advantage specifically because it has become rare — people who still read books are expected to be, with few exceptions, the only ones who think and write well.
- A narrow, single-purpose AI assistant can feel more useful to a non-technical person than a general-purpose tool, because it hides the complexity instead of asking the user to master it.
- Data pulled from over 21,000 US firms shows AI spend per employee climbing into the hundreds of thousands of dollars while headcount at those firms keeps growing, the opposite of the pure-replacement narrative.
- The privacy argument for running AI models locally is really a data-leverage argument — every prompt sent to a frontier model is also a live feed of a business's pricing, workflows, and competitive position.
Build the audience before the product, and keep goal-setting brutally simple.
WHAT TO LEARN
The clearest advantage in an AI-saturated market isn't a better tool, it's a small, specific audience built before the product exists and a willingness to actually act on ideas instead of endlessly refining them.
03AMA: first 100 customers and the age of the generalist
- Start any new business with an audience (a social account), then a community around it, then build the product last — not the other way around.
- A tiny, low-listener podcast can still fund a five- or six-figure launch if the audience is specific enough and the pitch happens directly inside the content.
- Learning to sell is not optional for a business owner: a large share of the owner's real job is promotion and marketing, even if delivery is what they trained to do.
04Worst things to do in 2026, and the lazy-outreach rant
- If you want to sabotage your own progress in this AI window, the fastest way is to spend all your time building software and none of it getting people to actually use it.
- Most outreach fails because it's generic; a small, concrete example of the work (an 8-second AI-made demo clip, a rough mockup) beats a templated pitch almost every time.
05Are design agencies screwed by AI? (plus SEGA's surprising history)
- Service businesses that already shifted from pure execution to strategy and outcome-responsibility are not threatened by AI tools; the ones still selling only execution are.
- Clients keep paying for work they could technically do themselves because they don't want the job, not because they can't do it — sell against that, not against capability.
06Why goals and vision statements are overrated
- Below roughly $20 million in revenue, elaborate goal-setting and mission statements are mostly theater; one tracked number and one directional sentence is enough to run the business.
07Cloudflare's agent economy: charging AI agents to browse the web
- Treat AI-agent-facing monetization (per-access micropayments for pages, datasets, APIs) as a genuine early-mover opportunity, not hype — being early to a new access layer has paid off in every past platform shift.
08Paul Graham: people are going to stop reading books
- A bias for action outperforms raw intelligence in practice; most people default to collecting more information instead of doing the thing.
- Reading is becoming a competitive advantage specifically because fewer people do it regularly — protect the habit deliberately (e.g., removing phones from the bedroom) rather than assuming it will happen on its own.
09iMessage agents and the rise of vertical AI apps
- A narrow, single-purpose AI assistant can out-convert a powerful general-purpose tool for non-technical users, because it removes the burden of figuring out what to ask for.
10You just hired a 1M bad employees: the AI headcount paradox
- AI adoption has not reduced headcount at adopting firms so far — spend per employee is rising alongside headcount, so plan around AI raising the cost of a seat rather than eliminating seats, at least in the near term.
11Local LLMs, data privacy, and the sign-off
- If your work involves proprietary pricing, workflows, or strategy, treat every prompt sent to a hosted AI model as a data-leverage decision, not just a usage decision — that's the real argument for running models locally.
Terms worth knowing.
- ACP funnel
- A three-step sequence for launching almost anything: build an Audience first (a social account), fold that audience into a Community (paid or free), then build a Product or service for that community — audience before product, not after.
- x402 protocol
- An open protocol referenced by Cloudflare's monetization plan that lets automated web requests settle small payments, in this case over stablecoins, so a webpage, dataset, or API can charge per access.
- Monetization Gateway
- A Cloudflare product opening a waitlist to let site owners charge any web page, dataset, API, or MCP tool for access, aimed specifically at AI agents rather than human visitors.
- Open-weight model
- An AI model whose trained parameters are published publicly, letting anyone download and run it on their own hardware instead of calling a hosted API — the basis for local, private AI setups.
- Vertical AI agent
- An AI assistant built to do a small number of tasks very well (e.g., a personal-assistant app) rather than being a general-purpose tool, trading breadth for a simpler, more polished user experience.
Things they pointed at.
35:30toolOpus Clip
35:00toolClaude
1:21:00toolOpenClaw
1:35:20channelPewDiePie's self-hosted local LLM rig
1:03:20productCloudflare Monetization Gateway
1:15:20productEight Sleep
29:40productEpidemic Sound
Lines you could clip.
34:10
“If you wanna really mess yourself up, just spend the next couple of years building pointless stuff and not having anyone ever use it.”
1:09:25
“It's not it's not advice, but you do not need to be a genius to win in life. You just need to be someone who has a bias for action.”
1:11:30
“If you're one of the few people who still read, you'll have a huge advantage over everyone else.”
1:26:20
“AI was supposed to replace human labor. It did the opposite. For the first time in history, humans are cheaper than software, and AI is creating more jobs than it eliminates.”
32:35
“I still think the core of running a business is building an audience and getting people interested in what you actually are creating.”
41:10
“Your job is actually to promote, grow, be coming up with a strategy for your business, so that you just don't get trapped in it forever.”
Where the conversation goes.
16:12 – 31:45denseAMA: first customers, sales, side hustles
31:45 – 47:18denseWorst mistakes of 2026 / outreach quality / design agencies
47:18 – 1:02:51steadyGoal-setting and mission statements
1:02:51 – 1:10:38denseCloudflare monetization gateway / AI agent economy
1:10:38 – 1:18:24steadyReading, attention, and brain rot
1:18:24 – 1:26:11steadyVertical AI apps (Poke, Lindy, iMessage agents)
1:26:11 – 1:33:57denseAI spend vs. headcount ('1M bad employees')
1:33:57 – 1:43:41steadyLocal LLMs and data privacy
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
metaphoranalogystory
09:39Oh, hello, sir. Alright. You guys are live.
09:41And we are live. Welcome to the second edition of SIP live where we livestream, ask answer your questions, look at the timeline, give a give you our takes.
09:56I'm your host, Greg Eisenberg, and I'm here with my cohost, Jonathan Courtney.
10:02He's a shrimp millionaire from Berlin, originally from Ireland. Welcome to the show, sir.
10:08Hey. How's it going? Should we or shouldn't we explain what Shrimp Millionaire is?
10:15I feel like we had it last time and no one really noticed it. So I feel like people just immediately were okay with it.
10:22That was the problem. Which was somewhat concerning.
10:27So, yes. So, yeah, with this AI show and one of the guys on there is successful because of shrimp somehow.
10:35Exactly. So how did you get the the nickname shrimp millionaire? Before we get into the AMA, before we get into the timeline takes, the people need to know.
10:45Look. Two things happened. Number one, I was in San Francisco with you.
10:50We were at some AI event. You there was, like, lots of stuff going on. After the event, you were like, let's go to Benihana.
10:57If any of you guys know about Benihana, if you're in The US, you know about this thing. If you're not in The US, you need to learn about Benihana.
11:04I'm living in Germany. I didn't know what that thing was.
11:08Greg and his team took me there. It's a hard to describe experience. It's kinda like a mixture of a show and food, and they're throwing shrimp at you, and that's what happened.
11:18I got like I think at some point, the guy threw shrimp, and I caught it in my mouth, which is a normal thing to do at this thing, and then I kind of I I really like shrimp. So that's where the shrimp thing came from.
11:31Right? That that was it just locked on. And then I was on Greg's show, and myself and my team are watching it back just to see how it was, and it was already on.
11:41It was already live, and he put underneath my name millionaire entrepreneur or whoever is editing his videos put millionaire entrepreneur as my title, and I just found that so funny just as like, it's so not on brand with me personally, but I did find it very funny.
11:59And so from now from that moment on, which was about two weeks two months ago, my team also referred to me as millionaire entrepreneur. And so now we brought them together for this show, shrimp millionaire. There we go.
12:13I'm the shrimp. I mean,
12:15I I think it's by the way, the comment section on YouTube was going wild about you being a millionaire entrepreneur because they're like, what does that even mean?
12:25What I mean, I guess the the do you wanna know what millionaire entrepreneur means? I can explain it if you want. My company make the companies that I run make more than $1,000,000, and the combined value of the yearly revenue is more than multiple millions, and I'm the owner of all of them.
12:50So I guess the I guess what it generally means is the net worth of the person.
12:57But regardless, I don't want to be called a millionaire, as in that don't start you guys in the audience, don't start calling me a millionaire entrepreneur. That's that's just like a a fact.
13:09That's not my title.
13:11Like, that's not my job title. Yeah. I mean, I think what's interesting about you, especially, like, in the context of, like, we were in Silicon Valley as and we were just around people who'd raised, like, so much venture capital.
13:24And what's cool about you, what I've always liked about you is you've done the whole bootstrap thing. You actually make millions of dollars. Yes.
13:30You don't make it from shrimp per se. You know? I wish.
13:34You've got your design agency. You've got facilitator.com. But, anyways, I'm happy you're here.
13:43We're gonna get right into it. Yesterday, I posted, what's the one thing you'd ask me and my fellow shrimp millionaire over here if you had me on the phone.
13:54And we got a 139 responses. I'm gonna go through, you know, three or four of them with you, Jonathan, right now.
14:04And then at the end of the stream, we're gonna go through a few more. And in the middle, we're just gonna go and talk timeline takes. So how about I'm gonna I'm gonna scroll down.
14:15Which one I want you to pick one of the replies, and we can go into it.
14:22I can't actually see that. Okay. So I'm You gonna know what?
14:26I can go to the actual livestream and see it but it's slightly out of, uh, sync. Let me see. Okay.
14:32So I'll just go through and
14:36and sort of see what what's interesting me.
14:39So I'm I'm I'm in the dark here. You know? I just get I just get told, hey.
14:44Turn up, and we're gonna shout shrimp at you.
14:47Sounds good. So at at the jesh warreb says, how do you get your first 100 customers? And I think we talked about this on the last livestream where everyone is building like crazy right now, but no one's really or most people are not focusing on actually how to get customers.
15:05Now to answer this question, I have to know what you're building realistically. But Yeah. I have a framework for for building anything, which is the ACP funnel.
15:15So if I'm building anything, I generally start with a, which is an audience. I'll start with an Instagram account, an x account, a TikTok account, c, which stands for community.
15:25So I funnel them into some sort of paid community, a free community, IRL events, and then I create a product or service to help those people.
15:34I think that how you can get your first 100 customers is you probably shouldn't be building, vibe coding, whatever, your software, your product, your AI agent. It doesn't make a difference to begin with.
15:45You should be thinking about how can I create a media company first, and then from there, go and build the product?
15:53Yeah. I mean, Greg, the the the cool thing is I'm literally doing this right now from scratch, and I started a new business last year, so I wanted to sell creative retreats, which is just not within my it's not anything to do with what my business does, and so it's a new business.
16:11And I was thinking about, like, how am I going to find customers for creative retreats? I want to do creative retreats. I wanna do art retreats, physical in person art retreats, large scale, like, events, essentially like an art festival.
16:27But I didn't really have customers to sell these things to, and so what I did is I started a media company, a very small niche podcast about being an entrepreneur who also likes creative stuff, which essentially nobody listens to.
16:46This is the important part. This time last year, every episode was getting even after, like, 40 episodes, maybe 300 people were listening to this thing.
16:56And at the peak of where I first sold out my first creative retreat, and we're talking about an event that, like, made, you know, a 100, 120,000 okay. Dollars.
17:07A $130,000. That was from a very small list of people, a very small audience, but very niche from doing a weekly YouTube video audio podcast that's long form, unedited, and then I just did the pitch right on that show and sold it right from that show.
17:29And so that was a perfect example of building, like, a very small media company. It's still tiny. I mean, I get Greg gets more views on one of his videos in the first second than I get in thirty days.
17:44It's that niche and that small. We're talking about my podcast, by the way, guys. And, yeah, that has allowed me, building that has allowed me to create a lot of mini businesses out of it.
17:55And so, yeah, starting with content is a great way to sort of find your people. I guess there the other way is like, actually, depending on what you're building, is I always look at it like there's the easy but expensive way.
18:12Yeah. And then there's kind of the hard and organic way. The hard and organic way, which I think a lot of people should do, is you just sort of build your audience.
18:20Right? And you build the audience slowly.
18:22The easier but expensive way is you you do ads. You do ads. Facebook ads.
18:26Mean, it's either you're either renting it or you're building it yourself. So I think, like, have to make that decision yourself and what works for you and how much capital you have.
18:35But I do think, like, that's the way to think about it. Thanks for the question. I'm gonna go through two more really two or three more really quickly, and then we'll get to the timeline takes.
18:45By the way, chat, I see you. I got you on my iPad right here. I got my x chat over here.
18:50I got my YouTube chat on over here. So I need I need that chat going. I need that chat blazing because otherwise, I don't know if I'm talking in me and my shrimp millionaire friend are just talking into the void.
19:04We wanna add as much value as possible, and if you, you know, throw questions in there because other people will answer them, and also I'm looking at them. So I see you I see you. I see you people.
19:15Alright. I also see you guys. Yeah.
19:16And and you I'm in.
19:19I'm typing in here. There it is.
19:23Okay. So how to build you you know, how to get your first 100 customers.
19:28I at the r h n or r h neutrino says, I've been a graphic designer, a web developer, but never a sales guy. Any advice?
19:38I mean, you have to like, this is the age of the generalist, period.
19:43So if you're a it's amazing that you're, like, thinking, okay. I was a graphic designer.
19:49Now I have to be a web developer. Now I have to be a salesperson. And I think salespeople, on the other hand, have to be like, okay.
19:57Now I need to be a web developer. Now I need to be a designer. Right?
19:59So I think what's happening in general is everything is converging. People re and the people who are going to do really well over the next five or ten years and by the way, just as how there was an insane amount of opportunity between, you know, 2020 and 2022 where people made insane amounts of money, huge opportunities, I believe that 2026 to 2029, there's gonna be that same wave in the sense of if you know how to use a lot of these tools, if you know how to build right now, it's an incredible time.
20:29So good job on you, step one, that you're actually doing this. How do you like, I would be like, how do I if I would want to be a salesperson, I would go and just, like, read the the biggest sales books, honestly, to get, like, frameworks around sales, and then I would actually just start making cold calls and start doing the thing.
20:50The only way to learn anything is by doing the thing. Right? So you can, you know, ask, you know, Google, Chat, GPT, whatever, what are the best sales books that I need to read?
21:00Get the frameworks, and just start making cold calls, start getting reject rejected, start meeting people, and over time, you put in your ten thousand hours, you're gonna learn.
21:12What do you think? Yeah. I think I think it's a, like, very important question.
21:16You know, how, you know, I am the I am the person who can do the delivery, but what do I do about the sales part? It's one of the most common questions entrepreneurs ask me as well.
21:26So to the person who asked, I'm sorry. I I didn't catch your name. I've been running a design agency for sixteen years now, and when I started out, I was just a UX designer.
21:38I didn't really have many sales skills, but because my company start start started to grow, I kind of just got put in the position of, oh, okay. I am that.
21:48I I that's actually my job. What I always say to to entrepreneurs is if you're the owner, if you're the CEO, and even if that's a one person company, then you kind of are automatically the CEO. A large proportion of your job, if not 50%, is to be the promoter of your business, to be both the marketer and salesperson for your business.
22:09And just before this call, someone on the call in the in the background was saying, I wish I had more time to do my YouTube videos, but I'm going to be stuck doing delivery and calls, etcetera, and actually, what I wanted to say is, actually, your job is to be the promoter, and you want to find people to deliver the work, not to, uh, not you permanently delivering forever.
22:32Your job is actually to promote, grow, be coming up with a strategy for your business, so that you just don't get trapped in it forever, if that's your goal. If your goal is to be a consultant who does the same thing over time, that's also fine, but learning to sell, learning to market is not really an option.
22:51You do have to do it, or, and I Greg, I don't know what you think about this, what I say to entrepreneurs is, if you truly can't stick to that, you actually have to find a co founder, and they have to own 50% of your business, because if you find someone who can find you clients, they will either take your clients or own your business very soon.
23:12Totally. Justin
23:14actually has a really good Justin Perea from I think it's on YouTube chat.
23:20He says, now how do I actually do that while having a full time job? So, like, we're basically like, yeah, you gotta go learn this stuff, but how you know, if someone's working forty hours a week, maybe they have a child, maybe they have a family.
23:32Right? Like, how how can they actually do this?
23:37I mean, well, to so I think that's obviously the the first thing you just have to admit to yourself or admit is that that will be more difficult.
23:46You know, if you have a full time job, especially if you have kids, it is gonna be a little bit more difficult. What I did is I had a full time job. Now, I didn't have kids, just to be clear.
23:55I had a full time job. I asked them, can I take Fridays off because I wanted to start building my own thing? I took Fridays off for, like, five months.
24:03They allowed me to do this. I didn't do anything on those Fridays because I had the kind of . I did nothing.
24:11I actually played the video game Skyrim and finished it, like, and did every every little thing in Skyrim. Amazing game.
24:18Um, so I did nothing on those Fridays, and eventually I realized if I don't fully quit and have to make money by doing by, like, being forced to start my business, then I'm gonna do nothing, and so I actually had to quit. So for me, my personal anecdote is I was unable to do the side hustle while working a full time job because it always The job tired me out, and I always, like, just felt like, oh, I'll get around to starting my own business later, but because there was no real deep pressure, I didn't have to.
24:50And so for me,
24:52that's just my I'm not it's not advice. It's just what happened with me. So Justin responds.
24:57He says he actually is working forty hours a week Monday through Thursday, and he can build Friday through Sunday. So if you could build Friday to Sunday, that is enough time. And that's enough.
25:07Right? That's like it's almost too much time. You know?
25:11It's amazing what you can do in the weekend. I actually find, personally, I'm way more productive on a Sunday than I am, like, the whole week.
25:21Like, I get more done on, like, a Sunday, like a like a locked in Sunday than I do in a whole week.
25:27I'm extremely lazy, so I I'm not I'm I'm I'm the lazy one of the two of us. I mean, you can just tell by looking at us.
25:35Okay. Canon Idea says, the worst things to do in 2026
25:39if I wanted to completely screw myself over going forward, if I wanted to let this golden age of opportunity elude me.
25:47That's a really fun question.
25:49What do you think? Who who asked that? Sorry.
25:51Who asked that quote? At canon ideas, who you would think would have a canon load of ideas.
25:57Canon. Yeah. I mean, okay.
25:59I'll just say, I'll if we're just talking about from a pure financial perspective, um, I will tell you that the things that I'm seeing people doing, it it's a combination of two things. One, I would assume that if you ignore all the completely ignore all the AI stuff, it's probably not a good idea.
26:15Like, if you actually decide, hey, you know what? This is all hype, whatever, I think it's probably not a great way to approach things just because what I'm gonna say next might make it sound like I think you should do that.
26:27I think the biggest mistake people are gonna make, and if you would want to absolutely mess yourself up, what I would do is spend all of your time building apps, and all of your time building software, and all of your time building loads of complicated systems, and none of your time actually selling things, getting users, and building an audience.
26:49I still think the core of running a business is building an audience and getting people interested in what you actually are creating. And then so I think if you wanna really mess yourself up, just spend the next couple of years building pointless stuff and not having anyone ever use it.
27:06I would say that that will also make you feel very demotivated.
27:10Yeah. I agree. I also think it's like a spectrum.
27:15There's like a spectrum of people. So there's some people who are like, okay. I know AI is a thing, but I'm just barely gonna use it.
27:23I think those people are kinda screwed on this at this point. Then I think it's like people on the other end of the spectrum, you have people who use every single tool like crazy.
27:35Yeah. And and every new tool comes out, they have to use it. They have to understand every single piece about it.
27:40I actually think those people are get are are kinda screwing themselves over a little bit.
27:46Addicted. They're overstimulated.
27:48Overstimulated, and they get you know, and I under I get this too because I I love a new tool. I love a new thing, a new shiny object.
27:56So whenever I hear something, I know, like, gotta try it. I actually think in the middle is the best place to be. You spend time building and marketing with, you know, Google, Anthropic, OpenAI, Grok, like, the main the main players.
28:16And then once a month or twice a month, you see something new that you think has the opportunity to really that just interests you. So if, like, for me right now, it's, like, AI, what they call, like, vibe filmmaking.
28:30So I feel like there's a few tools that are now getting really good that you could, like, actually direct commercials that are really, really good.
28:39And you can't do that today with, for example, you know, OpenAI.
28:44So you do need to, like, try new things. So I'm not saying don't try new things, but I think, like, the worst thing you can do is basically try too much or try too little. Just try the right amount.
28:55And I I think this was this was a really good question.
28:58Dude, can I may I just butt in for one second? Yes. Here so I run a couple of businesses.
29:03I get a lot of people contacting me to work with me and sell me stuff and, because I have a YouTube channel, and I'm sure you get this, Greg, as well. People contact you all the time, and they're like, I'm gonna make clips for you. I'm gonna do this.
29:15I'm gonna do that. And I'm always like, I wish I could just tell them how to actually sell to me.
29:22Like, why are they doing this so badly? There's so many different ways. Like, if you see someone who has a YouTube channel and you want to, if if you're like, I know how to use Opus Clip.
29:34Opus Clip is an AI tool for clipping up YouTube videos. Probably a lot of you guys know about it. It can clip YouTube videos, you can kinda edit how the clips work, you can kinda perfect it, and it actually does a very good job, like a fairly good job for a tool that's relatively automated.
29:51And I'm thinking, it's crazy to me that I run a business, that I'm trying to grow my my show, and 99% of the people who reach out to me won't even send me an example that could have taken them eight seconds to to make in something like OpusClip.
30:06An example of, oh, here's how one of your podcasts could look clipped up. Or, for example, like, even something as simple as people contacting me and saying, hey, by the way, I made this really simple, you know, lead magnet for your website, which probably would not take you more than four or five hours to create a really good one for a small business, and then reach out and say, hey, I made this for you.
30:29This is the kind of thing I do. I think this could work really well for you. I'm even with all of these tools, I still have people reaching out to me with just the laziest bottom of the barrel, hey, let's get in touch.
30:40Hey, let's have a call, and I think what these tools should be doing, if you're thinking about, like, not about building a massive business, but you could have a extra 2 k per month by selling some very simple things to small it doesn't always have to be to huge corporates or startups, to small businesses who have no like me, who have no idea how to build all this stuff, and just finding little niche ideas of what a lot of small businesses need.
31:07Here's a free one. The get in touch button on most small business That flow, how do you triage the different clients on a small business website?
31:19If you go to myfacilitator.com, click get in touch, see the way that triage works. I had to build that myself.
31:26I couldn't find anyone to do it. That kind of thing, you could charge you could have charged me 1,200 Euro to put that together, and you could do that three or four times for a small business.
31:38It doesn't have to be crazy complicated. It doesn't have to change the world. There's so many businesses out there that just are looking for small little efficiencies, and and you think they're being contacted every day a thousand times with really good stuff, but believe me, we're not getting contacted every day with anything other than people saying, hey, Grab a coffee?
31:59Question mark? The the it's it's really not a lot of effort is going into to any sort of outbound for us anyway. I mean, so what you're saying is
32:08is basically people are doing the bare minimum, especially when they're doing reach out. And especially in in the age of AI when there's all these tools, someone mentioned in the chat hyperframes, which which also, you know, I I haven't played with it.
32:20I think it's really interesting. Like, tool these tools are coming out that you can actually, you know, vibe or or build marketing assets or little, you know, versions of what they would get.
32:33You send it to them, and you're like, hey. I built this thing for you.
32:36It might not be perfect, but I think that you'd get some value out of it.
32:41It opens up the conversation, and then it it's sort of like the, you know, the same reason why if you go into a luxury you go into a car chase store, for example.
32:52Go into a car chase store. One of the first things they're gonna do is they're gonna give you a bottle of Perrier or a bottle of Evian, and they're gonna give you some French chocolate.
33:02So, actually, what they do is they say, do you want Evian or Perrier? Okay. I want Perrier.
33:07Then they give it they they give you, like, a like, on dinnerware, like, full on dinnerware, like a saucer, and and they give you, like, two pieces of French chocolate and a Perrier.
33:20And it's like, now all of a sudden, I wanna buy a watch. You know?
33:25I was just here for the air conditioning. You know what I mean? And now all of a sudden, I, you know, I'm I I feel good about myself.
33:32I got a little sugar high from from from the from the chocolate. I'm I'm I'm hydrated. So you can do that now in the Age AI.
33:40So I love this question.
33:44Greg, I'm so sorry. But is there any chance we could stay on this slight, like, for another minute? Yes.
33:49Because I think that is gonna be huge for your audience. Because they're not here they're not experiencing this from the side that we're experiencing it on. We're the business owners, and we're receiving this stuff.
34:00Right. And it was really the low it's really the Okay.
34:04I'll just give you one more idea and I'm gonna leave it. SEO is a huge, huge problem for small businesses. Like mine, don't contact me.
34:13I already solved it. Like mine, it is All you would need to do, open up Claude Fable, give it the task of figuring out a good strategy for a specific target small business that you're looking at, have it write one to two to three articles that would be part of that strategy, Contact that business.
34:32Say, I've already written three of them. Now, the annoying part for a small business like mine isn't writing the articles. It's the execution of all the SEO work, getting the h one tags right, getting all the the proper, like, linking and architecture between the articles, etcetera, etcetera.
34:49Write all the stuff for them for free. Do a little bit of Make sure that it's actually relatively good. Build their boy style guide brand stuff.
34:57Do it once, and you'll be able to do it for multiple businesses without having to tweak it. Then contact them and say, $3,000, I'll implement this for you, and then your ranking on AI tools is going to go up.
35:09It's it's such a simple thing, and it's something we all need as small businesses. It's so annoying to do, and none of us even know we need to do it.
35:18I only figured out I needed to do it in, like, February this year.
35:21Yeah. And I also think that there's this, like, arbitrage moment, and that's sort of what I was getting at earlier, which is, like, what has changed?
35:31There's all these tools now that allow you to do these things that a one person business can actually go and build these things. Also, someone who's nontechnical can now go and do these things. That has changed.
35:44Now in a few years, more and more people are gonna understand that. So there's gonna be there's gonna be more competition around some of these things.
35:53I'm not saying that it's you know, we talked about this last time.
35:58Like, I don't buy the permanent underclass narrative. I don't I don't believe in that.
36:04I think, you know, things will I I I think there is always gonna be opportunity for people to, like, build, do things, new pain points, and stuff like that. But my point is now is such a fertile time to actually do the things that you're saying.
36:20Selling is easier now because you can do those things. Yes.
36:25In the past, if you wanted to do what you were talking about, you had to have a team of people pay them, oftentimes, millions of dollars a year to execute on these things.
36:36Now you with the right workflows and the right agents and the right sub agents and the right skills, granted, you need to have the right distribution. You need to and, you know, you need to understand how to sell.
36:48But with the right recipe of all that, it's an incredible opportunity.
36:52Yes.
36:54Alright. That was a good one. Someone else in the chat said Kariat said in the x chat, at Kariat.
37:05And this is a question for for me, but it it it's relevant for you too. Do you think your design agency still has the same potential in the AI age where anyone with a brand book with Flora Fauna, I've never even heard of that, Tone of voice with Claude, landing page with Claude Code Codex, and build you a funnel with Cowork.
37:30So the question really basically is, you own a design agency, AJ and Smart. I own a design agency called LCA.
37:38Are we screwed?
37:42What do you think?
37:47I can speak from the LCA perspective. We are lucky in the sense that we started the business six years ago, and some of our early clients were helping companies build suites of AI native products.
38:02So we've worked with the biggest companies on the planet like Dropbox and Salesforce, and they've come to us and they they've been like, hey, we're we're stuck in the SaaS era. Help us build and move into the AI era.
38:13So our niche is you know, we've had the best year that we've ever had, 2026. So, you know, I think that from a design agency perspective, if your niche is a good niche and the type of client that you have is a good client, and you can figure out ways to increase margin, increase just like the way you you execute through systems with AI and stuff, it's it's good.
38:49Now, I think that 90% of design agencies don't have a good niche, don't have a good ICP, haven't figured this out.
38:56So I think that, unfortunately, a lot of them will suffer.
39:01Yeah. I I think being able to do this stuff was never the hard part of running a design agency.
39:08Like, the the I was running a UX agency that I'm still running AJ and Smart today, and that's mostly now product strategy, we realized relatively early on that the actual execution work was what the client was paying the least for, so as soon as possible, we moved from execution to strategy, and we try to move as far away from execution as possible, just because that was the thing that the clients kind of always intrinsically understood, well, you just throw people at it.
39:39You throw people at it and that will be solved. At least that was the client's perspective. And so the AI tools haven't really changed much for us as a design agency because of the fact that we already moved away from the pure execution niche.
39:57What I will say is that I think a lot of you guys who are watching this right now, you're in a bubble. You think your clients are just opening up these AI tools and building stuff with these AI tools and shipping them.
40:11Again, coming from the small business perspective, and I really want to speak for the small business owners here because, you know, we we also have a lot of money to spend on services like design agencies.
40:24I always have trouble when I when we for example, our website is currently built on Framer.
40:30One of our, uh, our facilitator.com website is built on Framer. I still pay an external agency when I wanna do something there, even though I can build it in Claude myself and I know how to use Claude code.
40:42Why? Well, because I don't want to do it myself, and that's still basically how it works when you're working with business owners. They don't want to do it even if they now know it's easier, And they shouldn't be doing it because that's not their job.
40:55They're, you know, they're doing their own thing. So You could make the argument, though, that in a year from now,
41:01like, these Claude will be so good that it will be as if it's, like you you basically could tell it, hey.
41:11I've got this website. It's managed on Framer. You do it now.
41:16I'm moving out, you know, and and it'll do it.
41:18I just don't see that for the business owners that I know. I think that I would always rather pay someone who says the the the reason you're also paying someone is that they take full responsibility of it, that they deploy it, that they take care of it, that they interview you and get all the ideas for what's gonna happen, that they check it on every device.
41:42I don't wanna I still think that external service agencies will need to exist because, you know, the the company that's selling custom couches in your city doesn't wanna have someone on staff who's a vibe coder.
42:00You know what I mean? And so they will need external people. I I I just I just don't see that going away anytime soon just because the AI tools get so good.
42:09You said two important things.
42:11One is you said the strategy component isn't going away.
42:16So in terms of finding incredible strategy, there'll be agencies that exist that help people find incredible strategy. Yes.
42:25That might mean the companies that focus on 90% execution today have to go and turn that into 10% execution or zero percent. Yeah. The second thing you said is responsibility.
42:40You'd be surprised how much people would be willing to spend to take their responsibility and just give it to someone else.
42:50Yeah, dude. That's that's the thing. That's the thing.
42:52Right? People because if something goes wrong, they have someone that they can Agency's fault.
43:00Complain and or a fall it's like a fall man a fall a fall man. Right?
43:04So to them, that's worth oftentimes a lot of money. And the sleepless nights that the agency owner has or the agency teams have, you know, that they're they're up at night because they're thinking, oh my god.
43:20What if this website goes down? What if this framework website goes down? Right?
43:23You hope a good agency, right, would do that. Yeah. So I think that's you I think that's where agencies are going.
43:31So where's the opportunity? Well, the opportunity is in more strategic, more responsibly bearing responsibility bearing agencies.
43:41And like I was saying, also, finding niches and ICPs that, like, you know, they're they're just it's they need your your services, which sounds obvious, but you know?
43:55It's it's I think it's one of those things where it's like, well, you so in in the chat, Justin wrote Yeah.
44:04You know, why should I reach out for them when I know they can do it themselves? Yeah. Why do I have a cleaner?
44:12Why shouldn't I just clean my office? Why do I, as a business owner, need cleaners to come to this office? Why don't I just do it myself?
44:21Think about it from that perspective, because that's the way business owners are thinking about it when you reach out to them, because they don't they're smart enough to know that they don't want to do this themselves because that's not their job. Yeah.
44:32Don't reach out to a vibe coding agency and try to sell them vibe coding, but if you're reaching out to, you know, my company facilitator, we work with corporates.
44:40We work with large corporates. They absolutely can do the thing we sell to them.
44:45They know how to they can easily train one of their employees up and then train hundreds of their staff. They would rather have us do it because they're a bank. You know what I mean?
44:56They're not trying to build, like, a internal facilitation training machine inside their company.
45:03That's not in their DNA. But they could, but it wouldn't make sense. So I think a lot of it is just, yeah, sure, people can do things by themselves, but it often just doesn't really make any sense to do it.
45:16Totally. Alright. I love it.
45:18Alright. Those are some of the questions. If we have time, we'll do a couple a few more at the end, so let's keep that chat going.
45:25I wanna move into the section where we break down some of the most interesting ideas from the timeline. You haven't seen any of these, and we're just gonna go and and go through it.
45:41Um, okay. This I just put in because I know you're a gaming nerd. So Oh, yeah.
45:46You're a gaming one. Let's go. So Shield Monote says, today I learned SEGA stands for service games and was originally a a Hawaii based supplier of slot and amusement machines to US military bases.
46:03I did not know this. Today I learned.
46:06I didn't know it either.
46:08What the hell? Sega thought Sega was a Japanese company.
46:13Who am I? I don't know. I don't even know who I am anymore.
46:15Right?
46:16What this is insane. Actually, anyone in the chat buy a Dreamcast? I was a Dreamcast boy.
46:22Really? Dreamcast, Jet Set Radio, Crazy Taxi. Let's go.
46:26If anyone is in the chat who's a Dreamcast boy, we're boys for life.
46:32Yeah. Jet Set Jet Set Radio was, like, iconic. Crazy Taxi, obviously, so good.
46:39Um, it it was like who? Crazy Taxi was kinda like TikTok in the sense that it it's just like a dopamine machine.
46:47Oh my god. I was so addicted to Crazy Taxi. Yeah.
46:50Yeah. Yeah. Yeah.
46:50Yeah. Sorry.
46:54Offspring.
46:54Offspring locked in. From a business perspective, I also think it's interesting that they their niche was US military bases.
47:03So they basically started a company, and they're like, hey. There's all these US military bases, and these people are bored.
47:09What can we sell them? Games. We'll be the game supplier for the service people.
47:18Like, serve you know, the people who are going to war. That was the business, and then they expanded beyond there. Very interesting.
47:23Weird.
47:24So weird. I did not know that at all. Yeah.
47:28Well, what you did know, here's another interesting tweet from the timeline. I've never had real goals or targets when building my businesses. I just started working on stuff and seeing if people wanted those things and went from there.
47:42I thought this was really interesting. You know, I think the when you're building a business, there's all all the literature says you should have goals.
47:52You should have targets. Why do you think you don't need to have them?
47:57Well, I think a lot of literature around building businesses is for large scale funded startups or public, um, like, growing public companies.
48:10And, I think, you know, anything pre 20,000,000 revenue, you don't really need that much goal setting or vision or mission or any of this stuff, And a lot of a lot of other people running I mean, I think Jason Fried also commented on this and and has a similar opinion for building what they build.
48:33I think, like, a lot of people get lost in the vision setting and goal setting and all of that around running a business, when it doesn't need to be that complicated when you're not trying to build a billion dollar machine, and most CEOs and most business owners, even of the very big companies that you talk to, they're sort of building a reverse narrative.
48:57They're like telling you that they had these goals after it all worked out, but if you actually saw the journey step by step, it's usually just and chaotic, and, you know, Nintendo is one of my favorite examples for for gaming. They started as, like, a trading card company that was, like, gam they were like a gambling company in Japan for a long time, and then they just kept trying different things and different things and eventually got into video games.
49:24There was no vision. There was no, like, vision statement around we're going to help the world play or we're going to get to a billion players.
49:33No. It was just we're making stuff, we wanna make money and we're gonna keep doing that and eventually kind of find your thing.
49:41And so, I've just never had a I've never had any sort of, hey, AJ and Smart is gonna be the, you know, we've hired consultants who've helped us come up with, like, these kind of statements, but I've never paid any attention to them. I've never stuck to them. I kind of did it for the employees because I thought you had to do that.
49:58But I don't know, man, I don't I don't do that. Usually, it's usually very basic. With with my current with my current con company unscheduled, my goal is, like, I wanna have 10,000 listeners per week.
50:12That's as deep as it is getting that's it. I mean, that is a goal though, isn't it?
50:18Yeah. It's a goal, but I also don't care if I hit it and I'm not tracking it and it just kind of is random. Mhmm.
50:24Yeah. I mean, the literature says the quote is, what isn't measure what isn't manage what is it what isn't measured is manage or the other way around?
50:37I don't even know.
50:38What what isn't manage is measured. No.
50:41I don't know. Someone in the chat will tell us. But, basically, if you don't, you know, measure stuff, it's just not it's
50:49you're not gonna get it. You're not gonna get it. You know, the p and l every month.
50:54We look at the profit and loss sheet. We're trying to see if we're making more money than before. Like, what we do is we try to look at how much money did we make June last year versus this year?
51:03How much money did we make q one q we we measure it in money, like, also it can be just measured in money. It doesn't have to be so complicated. So, yeah, we measure but, like, in reverse, like based on what happened and we're not very we just don't really do that, to be honest.
51:21Doesn't mean we're right. It's just I don't really do that. How do you do it?
51:25Do you have like a big goal for startup ideas pod like, when you started it, were you like, I'm gonna change the world using AI, blah blah blah, or is it just, I'm gonna do this and see if I can build a big audience, and then I'll figure that out later?
51:40I always start by just following my curiosity and experiment.
51:46Like, even with this livestream, like, I don't know how many of these livestreams we're gonna do. You know? If people like it, like, let me know.
51:52Let us know. Like it on YouTube. You know?
51:56Say things in the chat. But, like, as of now, if you were to ask, like, what do you what's your vision with SIP Live? I don't have, like, a huge vision.
52:06I just know that there's a lot of shows that I don't think are super inclusive and that don't actually give a lot of sauce.
52:15And so I was just like, how can I create a live show or livestream that is something that I would wanna watch?
52:24That's basically it.
52:30You know, if it it might turn into something down the line, or it might not. If it turns into something, at some point, I say to my, I I think it's important to have some sort of north star.
52:41So I really like the Toby from Shopify, like, mission where he he says Shopify's mission is to .
52:52And I just think that that's just a simple What does that mean? It means, like, he's got supposed to put on this? The rebels are the merchants, the people who wanna sell things, and his job is to arm them.
53:03So give them tools to, you know, fight in the war. And I think that's really interesting.
53:11And I like how, like, simple it is and stuff like that. I think I'm I'm not really a big goal or target person, but I am a big, like, what is the vision person.
53:21Mhmm. And then, like, every three months, I need to have some number that's Yeah.
53:30That is something that is basically like, I'm going in the right direction to that thing. That's all that I care about. And that's some it doesn't have to be money.
53:38Like, you know, money is one example of it. You know, users, retention, you know, word-of-mouth.
53:45You know, it depends what business it is, but one number, one sentence, and that's all you need when it comes to building a business, in my opinion.
53:56Abs I mean, that's that is absolutely all you need. I mean, I think on that front, you know, what I do every day for first, the unscheduled business is always the easiest one to talk about is I open the YouTube dashboard and I look at, like, how is how is the current performance?
54:13How is the current subscriber rate? And I kind of mentally think about does that because for that, I don't have a really big target, it's just does it feel like it's going in a positive direction?
54:25Am I seeing more green up arrows than I am red down arrows? And somehow that just already feels like the right that for that business, that that's all I'm looking for.
54:36Mhmm. Like that. Something.
54:39I like that. I've never to be honest, I haven't distilled it to, like, one sentence, one metric,
54:45but, no, I really like that. I think I'm just gonna do, um, I think for for AJ and Smart from now on, it's gonna be arm the rebels. I think I'm just gonna take Toby's one, arm the rebels.
54:57Uh, that's the new AJ and Smart slogan. Let's go. You're the new Toby.
55:02I'm the new Tobs.
55:04You're Toby, actually, now. You're Toby Courtney. What
55:07is it? Toby Courtney? What is it?
55:09Spot is he Spotify or Shopify, did you say? Shopify. Shopify.
55:13Okay. Iron the rebels. Toby, I'm taking your your slogan.
55:18Where is he from? Toby Lutke?
55:21He's from Germany. Is your
55:23From Germany. Yeah. Toby.
55:25Toby. Alter.
55:28He's watching. Right? He's watching.
55:30Toby, if you're watching, please meet Toby Courtney. He's a shrimp millionaire from Berlin.
55:36I'm your son. I'm your son, Toby.
55:39We're gonna move on to the next tweet from the timeline.
55:43I'll do.
55:44This one is really interesting.
55:48By the way, no. Not sponsored by Cloudflare. Not sponsored by anyone.
55:52I just thought it was an interesting announcement, and I was thinking of maybe doing an episode of the podcast on it.
55:59So they said, we're opening the waitlist for our monetization gateway, which will allow you to charge for any web page, dataset, API, or MCP tool behind Cloudflare. These the charges will settle on stablecoins over the four zero two open protocol.
56:14So, Jonathan, let me explain you what this is, why I think it's interesting, and why I think that there's a lot of opportunity behind it.
56:22I'm curious your take. So, basically, if you think about the web, the web was designed for human beings.
56:32Human beings access a piece of content, for example. And in order to monetize that, human beings put ads.
56:42So, you know, basically, the beauty about the Internet is it's free for the most part. Very small amount of the Internet is actually gated for that you actually have to pay to access.
56:54And because of that, we've actually had a lot of really interesting websites, you know, and apps come to be over the last twenty years.
57:02Now it's pretty clear that agents, AI agents, are going to be greater in number than human beings Mhmm.
57:14Relatively soon. I'm not sure exactly when. I don't know if it's gonna be a year or three years or two years.
57:19But, you know, in one to seven years, I think it's pretty clear that agents will overtake human beings. So that is gonna be interesting.
57:28In a world where the user of the Internet is most likely going to be an agent, they're going to be pinging a bunch of different websites to do things.
57:40Now, the problem with that is the monetization model of those websites assumes that it's a human being that is visiting the website.
57:50Right? It's assuming that this human being is going to look at an ad and maybe buy something.
57:59It's assuming that, you know, that human being is gonna fork over some data and then that the company can use that data or sell that data or something like that. What Cloudflare is saying is, no.
58:12No. No. No.
58:12No. We need to have an open protocol that basically allows website owners, publishers to add something to their websites.
58:24And basically say, if you're an agent, I'm gonna charge you some fraction of a penny. In this case, they're using stable coins, but it doesn't it doesn't matter how they're settling it in the back end.
58:35No one cares about that. All that matters is a a fraction of a penny to use to access a website.
58:44And I believe that this is what's going to happen. So I think that there's gonna be a lot of opportunity to actually, you know, as Gary Tan says, make things agents want.
58:59Basically, there's gonna be a lot of startup ideas around creating websites with the agent in mind and then adding the Cloudflare monetization gateway to charge for it and then figure out ways to get traffic not from human beings, but from agents.
59:21So you have to reverse engine you know, we talked on this on this pod on this livestream about, like, promotion, but more from a human being perspective.
59:31What does it look like when you're trying to actually get agents onto your website? And I think there's gonna be many, many, many, many wealthy people being created who figure this out over the next twelve, eighteen months, they're gonna have a head start.
59:47The way to get access to the monetization gateway is you do need to be a CloudFare customer.
59:56I'm you know? That's kinda easy, though. Right?
59:58You just You just, I'm pretty sure you can Sign up for, a domain or something. But really interesting.
1:00:03What are your what's your, you know, first reaction to this? Does it sound too sci fi to you? Does it sound like, you know, this is no.
1:00:12This will never happen. It says, you know, open protocol, stablecoins. Are you just like, this is hype?
1:00:18What are your thoughts on Cloudflare? Yeah.
1:00:21This is so far outside of my wheelhouse that I first need to ask you if I understand. So is the idea here that, you know, someone could build like a someone builds like a website, and for some reason, this website is attractive to agents in some way.
1:00:37Maybe there's data on that site that, you know, LLMs wanna keep pulling from because it's like the only place that has the live data for certain things that people might wanna search for. And when the agents are going to pull that particular data, they pass through a gateway where they actually get charged. Who's paying the bill?
1:00:56Is it the is it, like, Anthropic? Is it the person who's doing the search? How's that happening?
1:01:03So the assumption
1:01:05is that we're going to be living in a world where AI agents are going to have wallets. In fact, they're not just gonna have wallets. They're gonna have email addresses.
1:01:13They're gonna have phone numbers. They're gonna have What about shoe? What about shoes?
1:01:17They're gonna have some form of shoes. They're you know? They're gonna have shrimp of some sort.
1:01:24Know? Sun lotion, all that stuff. Right.
1:01:26And you saw this with you saw this with, uh, Mold Book, the acquisition, I think, that Meta made. Uh, $100,000,000, I think it was, where they bought a social network for AI agents.
1:01:40Um, it was really early on when they bought it, but I think it was it was interesting in the sense that it sort of, uh, foreshadowed a world where, yeah, you can have these AI agents, and it makes sense that they're gonna have wallets to do certain things.
1:01:51And it makes sense that they're gonna have email addresses and phone numbers and stuff like that. So, you know, behind every agent is going to be a human being ultimately.
1:02:01So Yeah. Think of it as, like, a company card. Right?
1:02:05Like, I use I use Brex bank accounts, and we have multiple companies and multiple credit cards that we set up on Brex.
1:02:17And it's really it's really cool. Like, I I'm able to see basically, like, who's spending what, how they're spending it, limits, stuff like that.
1:02:26And the same way that you can, you know, in an easy way, flip on a Brex, you know, company card or whatever, you're gonna be able to create company cards or cards for agents.
1:02:41Now I don't know if Brex is going there or who's doing it, but this idea of creating wallets for agents, and you'll have, like, limits.
1:02:53So you'll be able to see, like, okay. I'm I'll give my agent $20 a month, and Yeah.
1:02:59Let's see how they used it. Right? What was the value?
1:03:03I somehow find this really interesting. Like, first of all, I find it kinda cute, the idea that your agent is going around with a cute little wallet and everything. But the other thing is I'm like, the first thing that comes to my mind is I'm like, okay.
1:03:15Even if Cloudflare doesn't do this, what is it that I can build that l l people using LLMs are just gonna need, like, they'll just want that data, and their agent is gonna come back and say, hey, there's actually a paywall here.
1:03:29Jonathan's website, I can't get that data unless you pay $20 a month. Do you want me to pay it? And on the other side of that is what they need to continue doing their research, and I guess the easiest way to imagine this is, let's say, you owned Reddit, and the only way for your agent to go and and, like, you know, crawl through a Reddit thread is if it's if it pays for it, and I know these companies kind of do deals between themselves, but this is probably at a smaller, more fractional level, like, that actually is that so so is that basically one of the thing one of the theories that there's going to be businesses, there's going to be people who kind of get wealthy from owning a certain thing, data set, whatever it is, that has to be paid for to get access to, and it's a thing that generally a lot of agents will need access to, and so thinking about those things before everyone else does is kind of a weird arbitrage.
1:04:26I literally never even heard of this in my entire life. This is the first time I'm hearing about it, but somehow I find that weird and exciting. Is that it?
1:04:32Yes. That's it. And that's, you know, that's that's sort of like,
1:04:37of course, there's a world where this doesn't play out. Right? Of course, there's a world where you like, there's there's certain hypotheses that need to be true.
1:04:54Right? Maybe it's agents don't have don't don't have wallets or they don't have email addresses or maybe agents don't wanna pay for it or you know? So a lot of things need to be true.
1:05:03But if someone like Cloudflare, who's, like, a huge company, I think they're, like, an $80,000,000,000 plus company, they've been right so many times.
1:05:11If someone like Cloudflare is basically saying this is the future, you gotta, like, just listen.
1:05:18You know? It it there's because in the sense I I'm not saying it's a 100% certain that this is what's going to happen, but I think with a high degree of certainty, you're going to have some monetization gateway for AI agents at some point.
1:05:33So being early is helpful because, you know, you just have to look at any wave, social, mobile, cloud, whatever.
1:05:42The people who were there early had an unfair advantage. First mover advantage is is a thing for a reason.
1:05:51I wanna know I wanna know what the first one of these things is because it's actually like, you know when music went digital and then there was loads of piracy, and then there was almost like this assumption that no one's ever gonna pay for music again, but then something like Spotify came along, all of the new digital rights for music came along, and now, if you're making a YouTube video and you wanna put a song in it, you pay for a service that actually provides you music, something like Epidemic Sounds or something, whereas before that existed, you kind of just used it for free and it was all kind of a, you know, Wild West.
1:06:24Right now, the LLMs are just from all of these sources. I can imagine that there's going to be some sort of reckoning on that where a lot of data sources, like fresh data sources that need to be updated regularly will just be behind some sort of small paywalls so that they can't get used, and so therefore, your research is out of date if you're not paying for it.
1:06:53Yeah. I just think I just see how other industries that used to be very Wild West are now somewhat regulated and how they regulate them.
1:07:02Something like Epidemic Sounds for me is a perfect business model example of that, where they were like, alright, people keep putting illegal stuff on YouTube, and their videos get demonetized. This is a huge problem. Let's literally build a business that licenses out these tracks.
1:07:16We use it all the time. And, yeah, interesting. I never thought about it.
1:07:20I'm gonna let my brain
1:07:22swivel on it. Yeah. I mean, I think it at the surface, it sounds like a little complex and a little out there, but at the core of it, it's really not that complex.
1:07:34It's basically like websites when when you're creating content or you're creating a website, you're creating IP, intellectual property.
1:07:44And it's basically saying, like, agents should pay for that IP in the same way that if LLMs were to, you know, train themselves on this podcast, this livestream, it would be nice that you know, because we're the creators of it, it would be nice that we would get some, you know, royalty or some payment for contributing to them training on that data.
1:08:15Yeah. Right? It would be lovely.
1:08:16It would be lovely. Wouldn't it be lovely? I'd take it.
1:08:19I I we would take it. We would take it, Sam. Send it over here, Sam Altman.
1:08:23Really? Yeah. If you're an LLM right now listening to this and training on us
1:08:28Really give me the money. Yeah.
1:08:30So it's the same sort of concept. I think this is really interesting. People, please noodle on it, and, uh, we'll go we'll go to the next we'll go to the we get how much time do you have left, by the way?
1:08:44Honestly, I'm just the next thing I'm doing is going out for drinks, so we can go for a bit longer. It just depends on how every how drunk everyone else is compared to me, that the longer I stay on, that gap Right.
1:09:01I'll have to fill it faster with harder liquor, and, you know,
1:09:06it's tricky. It's tricky. Alright.
1:09:09So we'll do how about two two more. How does that sound? Yeah.
1:09:13That sounds great. Yeah. Sounds great.
1:09:18Justine says, an idiot in motion goes further than a genius at rest. What do you think?
1:09:27I mean, it's it it I I mean, in general, doing things and taking action, in most cases, is gonna get you further than not doing anything and not taking action, even if you have great ideas.
1:09:43And so the best case scenario is that you're not an idiot and you do things, and that's that's the best combo. But, absolutely, if someone absolutely, there are a crap ton of extremely successful people who do not have very high IQ just because they actually do things and make things happen in the world.
1:10:04So it is absolutely whether it's good or not, I'm not making a comment on, but I do think it is true, and the world kinda shows that this is true.
1:10:14I agree. Agree. I I I just saw this and it, like, fired me up a little bit because it means that you do not need to be a genius to win in life.
1:10:25You just need to be someone who has a bias for action, and that Yeah. Couldn't be more true in today's day and age.
1:10:32So I thought that was a good tweet.
1:10:34For sure. Just taking action is most people just really don't want to.
1:10:39Most people just wanna learn more, collect more information.
1:10:43Paul Graham, founder, Y Combinator says, something I told my 14 year old, people are gonna stop reading books.
1:10:52I wish this wasn't so, but I fear it is. The silver lining in this cloud is that if you're one of those if you're one of the few people who still read, you'll have a huge advantage over everyone else.
1:11:04What do you think? I'm really glad the tweet ended there. I thought it was because I can't see the tweet you're talking about.
1:11:09I thought it was gonna be like, people are gonna stop read book reading books, so here's a new app idea, and I was just going to roll my eyes. I I 100% agree with this. I am I force myself to read and and I read fiction in bed with my Kindle and I also am really putting a lot of effort into encouraging my daughter to read.
1:11:31I just bought her a Kindle, Color Soft Kids, and I'm trying to get her to read it as much as possible. Guys, if you have kids, I saw the best tweet ever and it changed the game for me.
1:11:40It said, if you want your kid if your kids read, it's just amazing for their brains. Everyone kinda knows that. But if you wanna get them to read a lot, what I've told my daughter, who's seven, is, hey, when you're in bed, as long as you're reading, there is no you have to go to sleep time.
1:11:57You can read as long as you want. You can read for five hours if you want. Don't don't bother me.
1:12:03Don't get up and bother me and tell me I have to do stuff, but if you want to read in bed, you can read as long as you want, and she sees this as, like, so I don't have I can stay up all night, basically. And so she she reads for like two hours straight in bed, whereas before I gave her that rule, she was too lazy and wanted me to always read everything to her because she just wanted to lay there, like, while I read.
1:12:25And so reading, I think, is a massive competitive advantage. I think there's a huge difference between consuming the written word and watching or listening to something.
1:12:36For me, I get massively different outcomes when I read a physical book or read on the Kindle reading words. I don't know.
1:12:44It goes indifferent, it it it's somehow something different happens. And so, yeah, and I do think it's also an attention span thing where if you're able to actually read a book, it says something about your ability to live within the modern chaotic mess of of of and not being so reactionary.
1:13:02The only thing I would add to it is don't just read non fiction. Like, fiction is a big part of developing your brain and also getting really interesting weird ideas.
1:13:14So I do think adding into that, make sure fiction is a big part of this and not just you're reading the stoic books on repeat or something.
1:13:23I had a visual of your daughter just, like, being in bed at, like, four in the morning. Like, her eyes are, like, bleeding because of, like, how tired she is. Yeah.
1:13:34She actually she is reading She she has taken advantage of it in a way that I didn't expect. I thought she was gonna read for half an hour and fall asleep, but she's, like, locked in, and it's eleven.
1:13:44And I'm like, oh, man. Okay. I need to probably need to change the the situation here, but right now, it's summer holidays.
1:13:50So Paul Graham
1:13:51Paul Graham says, the people who still read won't just be better informed. They'll be, with a couple of exceptions, the only ones who can think well. You can't think well without writing well, and you can't write well without reading well.
1:14:04And I thought this was the it was this was interesting. So this is the the rebuttal to Paul, and he responds.
1:14:12Uh-oh. Yeah. Who who rebutted?
1:14:14Who rebutted? So Kieran Ritchie says, books are a monumental waste of time. 600 pages for every one page of actually valuable content.
1:14:24I should know I've written three. You want information density? Blogs, papers, articles, and a well curated x feed.
1:14:32And then Paul responds and he says, most books are bad. Part of reading well is finding the good ones. And Kieran says, fair.
1:14:41I stick to textbooks these days for the higher signal to noise ratio. What do you think about this? I fucking
1:14:47know. I just don't like that that way of thinking about the world. I think that, um, you you sometimes I'll read an amazing book and I'll go on Amazon, like, I wanna, like, leave a review.
1:14:58I wanna, like, say that I like the book, and then I'll always see someone who says, this could have been an email. This could have been a blog article. They don't get the point that a book, what it helps you do is to, it repeats case studies, it repeats things, and it helps it get into your head in a way that's more lasting and more robust.
1:15:17And so, a non fiction perspective, I think that almost every non fiction book that I've ever read, when I look at the Goodreads or when I look at the Amazon reviews or when I tell a friend I've read a book that I really enjoyed, they're like, oh, you could have just listened to the podcast episode where he was on Tim Ferriss, and I'm like, you just don't you don't get it.
1:15:36Like, you just you're not this is not it's not about efficiency. Not everything has to be about efficiency and density, and I don't know who this Kieran person is.
1:15:45I'm sure he's a lovely person, but I 100% disagree. I think the same thing people say about fiction novels, they're like, you know, it was about 200 pages too long. Yeah.
1:15:55If you have if if if you're trying to, like, just absolutely get the most efficiency out of life, I don't think life is all about efficiency. I don't think knowledge gathering is all about efficiency either, um, and I really hate that sentiment personally.
1:16:11I'm not a fan of it.
1:16:12Yeah. I think, um, it's a barbell approach.
1:16:16I feel like you can like, both could be true. Like, you can read good books, and that's an advantage.
1:16:24And you could also have, like, a really curated information diet of, like, blogs and articles and x and stuff like that.
1:16:33You know what I mean? So I think, like Yeah. The people that
1:16:35Yeah. Just do both. Just do both.
1:16:37You know? I am I do both. I look at x and I also read both.
1:16:42Like, I just don't get the I I've always really been irritated by that sentiment, to be honest. Yeah.
1:16:48I think
1:16:51for people who and, you know, I'm I'm guilty for this because sometimes I I do this. Like, for people who spend a lot of time brain rotting, like, at the end of the night at the end of like, before you go to bed, you're kinda like you got one eye open, and you're just, like, scrolling through Instagram reels.
1:17:07And rot. Yeah. And then rot.
1:17:09Then it kinda, like, morphs your brain in a way that, like, when you go into a book and it's so slow, it just Yeah. It it it feels weird.
1:17:19So Yes. You gotta if you haven't read a book in a long time, like, and you you consume Instagram reels and stuff like that, do yourself a favor and read some books.
1:17:32It's it's it's gonna be helpful, and and I agree with Paul Graham. It's gonna be an unfair advantage.
1:17:40I also think the actually, I mean, look, it actually happens quicker than you'd think.
1:17:47So I I had a brain rot period over the last six months, just part of business building, part of just sometimes I go into a phase where I'm much more reactive and not very reflective and a bit more chaotic.
1:18:01And then I decided what always happens is I realize I'm using my phone too much and I realize my phone is in my bed too much in the morning when I wake up and this and so I said to my fiance, hey, we need to get the phones out of the room. Let's get our Kindles back, you know, like, to go, and so we're both in Kindle world at the moment, and it it just makes life feel slower and less reactive when you wake up and you don't have a fucking phone in your bed.
1:18:31The other thing I will say is that that I got the new Eight Sleep, which has all the controls on the actual bed thing itself, which has been a game changer. So that's another reason now I have no excuse to have my phone in my room. I think phones look.
1:18:44I, on one hand, love that everyone is not reading because it gives me an advantage in life over other people in a mean I'm just saying that in a mean way. I understand that that's not a nice thing to say. But, yeah, for my own friends and family, I would I wouldn't I would wish upon them that they have the opportunity to not rot their brains and be able to re it's like it's like a cure for your brain, honestly.
1:19:08It feels like it's doing undoing some of the damage of all the weird fight videos I get shown on Instagram.
1:19:16So a couple of things. One is I just this is unrelated, but I just I find when you say the word fiance, it's just funny to me.
1:19:25How do you guys how do you say it? Just because, like, you you're like, it's so fancy. It's like, oh, my fiancee
1:19:31fancy, isn't it? You're so fancy. You're such a fancy guy.
1:19:35What would you say? What would you say? My woman.
1:19:38No. It's I honestly
1:19:42I don't like the word either. It does sound a bit too
1:19:45It just sounds fancy. I it just sounds fancy. That's all I'm saying.
1:19:49What would you say? What did you say before you were married and the period where you weren't married where, you know That was a really short I had a very short period between getting engaged and getting Part of that was I didn't wanna have to deal, and this is very With same fiance.
1:20:06With same fiance. It's very Larry David of me. I hate the word.
1:20:11I truly hate the word. I'm like, I can't stand it anymore. I'm like, we need to get married today.
1:20:16Yeah.
1:20:18Oh, someone says our our microphones are noticeably different in volume. Who's louder?
1:20:25I think I might be louder, but, yeah, let us know in the chat. The second thing is, Ashley says, I swear to god every stream, JC is gonna bring up Eight Sleep, l o l.
1:20:37I'm either complaining about Eight Sleep, uh, or I'm saying it's great. It's it's one of the two.
1:20:43I I have a I have a emotional roller coaster relationship with that company, I tell you. Jonathan, you are louder.
1:20:51Oh, I'm louder. Well, that that's good. You don't need to really hear what Greg is saying.
1:20:54I'll turn mine I can turn mine down a bit here. Okay. We're gonna do one last take.
1:20:59And
1:21:00if you wanna keep the chat going and ask questions, please feel free on YouTube NX.
1:21:08I'll look at it, and maybe we can answer a few more. I don't know if John needs to leave and have a gin and tonic with his fiancee.
1:21:15So, feel free to leave. I have until 2PM Eastern.
1:21:20Laura, you start drinking now so that not only you can catch up now, then I'll come in, and I'll just take like three shots, and we'll be good.
1:21:28So this is just a cool thing I saw. Stanley says, I made an app store for iMessage agents. It's a directory of the best AI agents you can text on iOS.
1:21:41And, you know, what it is is, you know, there's a bunch of different these these assistants, Caddy, Poke, Tomo.
1:21:49And I thought what was interesting about it is you're starting to see just AI agents that are you just iMessage and which is so cool.
1:22:01So both from a consumer, like you're you're European, so I know you live in WhatsApp.
1:22:08What's iMessage?
1:22:10I'll I'll literally iMessage you, and you just, like, won't respond. But in WhatsApp, you'll just I'll just get a voice note, a four minute voice note from you. Like, that's that's our dynamic.
1:22:19But
1:22:20we love iMessage here in The United States, and it's a really easy way to just consume a product. So from a consumer, I love it.
1:22:31And then from a business perspective, I think that, you know, building iMessage agents, really interesting.
1:22:39Something I'm gonna be spending some more time thinking about, what do you think about this?
1:22:43Yeah. I mean, as you said, in Europe, we don't know what iMessage is.
1:22:48I don't know what an iPhone is. We don't know about these things. We we still use, like, a stick and, like, a just, like, you know, these types of objects.
1:22:57IMessage agents, honestly, I've never used one and never even interacted with something like this. The closest thing I've ever interacted with in terms of messaging is OpenClaw, you know, building that into, like, a a Telegram thing and and being able to interact with it.
1:23:15I don't I don't I honestly I honestly don't know. I think, for me, I imagine that that is just gonna be something similar to using OpenClaw or having it as, a front end interface to other objects that you have, you know, like, let's say it's a it's a smart home.
1:23:32This is now your interface instead of having to use the interface of the Philips Hue thing or Apple Home. Now you can just text and say, actually, hey. Could you just turn on all the lights?
1:23:41Or, hey. Could you just do this? And it just figures it out, and act it interfaces with all the other things.
1:23:46But I actually don't know do you know many different interesting use cases for these things? Because I imagine they're just like a front end interface to, like, a more main thing.
1:23:58Yes. Yes. So, I mean, like, over here, you can see Poke.
1:24:01If you haven't tried Poke No.
1:24:03I don't know what that is. It's basically like a personal assistant
1:24:07app. Just like It's it you should try it.
1:24:13And as, like, a UX person, you're gonna, like, be blown away. So Really?
1:24:18I'm You know? Yeah. Like, the onboarding experience, like, anyone who cares about UX, just download Poke, honestly, and just see how their onboarding experience is because it is world class and shows you a peek into what agent first onboarding experiences are gonna look like.
1:24:36I think it's poke.com.
1:24:38No affiliation. Yeah. So this this is the one, um, just to show you this is the European version of it.
1:24:45So that's what you're talking about. Right? Pokemon GO.
1:24:47Yeah?
1:24:49You can't get pokeinpoke.com.
1:24:51Doesn't look like it. No. It doesn't look like it.
1:24:53Damn. No AC and no poke. You guys are Yeah.
1:24:56We're You're struggling. Okay? Got we just got Pokemon over here.
1:24:59I'm just checking on poke.com directly. So what's the difference between poke.com and using, like, Claude?
1:25:09Well, I mean so Claude is wonderful, but you can do a million things with Claude.
1:25:17The these assistants that you're seeing, like, on this iMessage agent store are pretty verticalized. So they don't do everything. They do a few things, and they do a few things well.
1:25:31And that's what's cool about them. Once you actually, like it's sort of hard to explain, but once you actually start messaging with your poke or you start messaging with some of these other ones, you it just feels a little bit different.
1:25:47I invested in Lindy Lindy dot a I, which is, like, another one. Remember Lindy? Yeah.
1:25:51Yeah. Lindi is is really focused on, like, you know, you you work in a business, and you're you want a personal assistant.
1:26:01So I actually asked Flo, the founder of Lindi, like, what's the difference between Lindy and, say, working with OpenClaw or or Claude or whatever?
1:26:14And his his response was basically, like, what we're trying to build is, like, the Apple version. Like, it just Yeah. It's it just, like, works really well.
1:26:23Yes. And so I think there's that's what's that's that's why some of these iMessage agent apps are really interesting.
1:26:32So that's why I wanted to bring it up. I thought it would I thought it could get people's creative juices flowing. I thought whoever created this, Stanley, who created this app store for iMessage agents was really smart.
1:26:42And I'll be I'll be looking looking at that some more.
1:26:47I didn't get what you were talking about till I looked at their sales pitch here, and now I know what you mean. Like, whereas most people I know are still just using ChatGPT, still just using it as a search engine.
1:26:59And if you would try to present if I would try to present Cowork or ClaudeCode or any sort of automations to anyone I know who's just using it as some sort of chat engine, they would just be like, no. I'm not.
1:27:11That's just insane. I'm not gonna do that. Whereas poke is like a yeah.
1:27:15It's a it's essentially a polished front end for doing the complicated things you can do in these AI tools Yes. Without you having to know anything about these AI tools. Exactly.
1:27:26Yeah. That's that's cool. I I didn't even know this existed, to be honest.
1:27:31That's why I'm here. That's why I'm here. I This is cool.
1:27:34I wanna do one more take. This is this has sort of been, like, the article of the week, I would say.
1:27:45It's gone completely viral. You just hired a million bad employees. And, you know, I recommend people read it, but a 16 z quote retweeted and sort of showcased some of the most interesting points of the article.
1:28:02I'm gonna read it to you, and I'm curious your perspective like, what your your feedback is. So they say AI was supposed to replace human labor. It did the opposite.
1:28:10For the first time in history, humans are cheaper than software, and AI is creating more jobs than it eliminates. And then it shows these graphs.
1:28:20I don't know if you can see them. But basically can see. Basically, oh, basically, it's the token spend per per employee.
1:28:31So you can see on the y axis, you know, the annualized AI spend per employee.
1:28:39You know, people are spending 50 k, a 100 k, a 150 k, 200 k. So it's just you're you're starting to see people spend a lot of money on on on tokens per employee.
1:28:51And then the head count growth post AI adoption. So, basically, this graph shows once you've adopted AI, what happens to head count and what happens to, you know yeah.
1:29:07What happens to headcount? So what do you, you know, what do you think of this?
1:29:12What are your first reactions to this? And what do you think happens with human labor as people adopt AI?
1:29:21I wouldn't have expected this at all. I mean, I'm wondering, this, in Silicon Valley itself, is it related to the AI boom, or is this across companies generally in The US?
1:29:34Because I would obviously assume that within Silicon Valley and in tech companies, as the AI boom kind of keeps going, they wanna hire as many people as possible to to compete with these tools and and and, you know, build different things and and just be as fast as possible.
1:29:52But I'm just wondering, like, why are they I'm very surprised.
1:29:56I I would have assumed the head count would go down over well, I guess what I'm wondering is is this a small blip within the bigger picture of a more downward trend towards having smaller companies, which seems more logical?
1:30:14I I need to go deeper into this, but it does say it it it's pulling 21,559
1:30:24US firms. So I I I don't think it's just Silicon Valley. You know what I mean?
1:30:28No. Yeah. No.
1:30:29That wouldn't make sense. Yeah. My my take is I imagine that this is short term, and I can't imagine that I I just I just imagine that the bigger, longer term picture is that the head count might go down.
1:30:48If on an average, like, ten years from now, we're looking at, like, all of the companies in the s and p 500, I just imagine that the headcount would be lower due to these tools getting better and better and better, but not the expenses.
1:31:03Like, I can imagine that then those expenses you and I talked about this in person in San Francisco last year that these tools and these agents, etcetera, will essentially just, you know, replace the employee costs or be even more expensive than them.
1:31:18Mhmm. But I do I I'm just if I were betting, I would bet on head count overall slowly going down Interesting.
1:31:26Versus this trend continuing, but I don't I mean, I don't know. What do you think?
1:31:31So at
1:31:32stories by Will on YouTube has a really interesting comment. He says, is it going up because the companies are growing due to use AI?
1:31:42So that's the question, really. Like, these are Mhmm. So the companies that are becoming AI native and are actually doing the work to go and figure this out and doing workflows, having AI agents, doing all this stuff, is it because, like, these are the type of companies that are, you know, they're figuring out like, what we talked about around, like, marketing assets.
1:32:03Right? Like, maybe they're closing more deals. Maybe they're increasing margins.
1:32:09Maybe they're figuring out ways to have better Facebook ads, and that means more customers.
1:32:16Is that the reason why you're seeing this?
1:32:21Yeah. I mean but there's so I I guess the large corporates are still not really adopting it in a way that you'd see the result like that.
1:32:33So it's it's just strange. I mean, basically, I wanna see that graph in two years. Yeah.
1:32:39I mean,
1:32:40to be monitored, TBM. You know?
1:32:43Yeah. I think TBM.
1:32:45Absolutely. TBM.
1:32:48So is that it? Should we should we end end here?
1:32:57What do you think, Rashid? Like We're in the perfect TBM moment.
1:33:01To be monitored. We do If you Let's see.
1:33:04I'm gonna look at chat. See if there's any if there's any last few questions, let us know. I see one question.
1:33:11BlueberryMuffin on x says, what's a use case you've seen in application for a local LLM?
1:33:19How much compute do you actually need? And let me show you something here related to that.
1:33:30So I saw this really good let me go here.
1:33:37I saw this really good tweet by, Oll, another fellow German.
1:33:48T tldr, startups need their own intelligence. Why? You're currently paying OpenAI Anthropic twice, once in tokens and again in your proprietary knowledge.
1:33:58They feed the models to make them useful. The fix is to run open weight models where nothing you type gets stored or trashed on or trained on. Your your prompts are a live feed of how your business works, your pricing, your workflows, your data.
1:34:10The point is basically his point is he's he's he's he's like, alright.
1:34:17You you shouldn't be paying these big LLMs because they're looking at your data.
1:34:26And so there's that's, you know, why you should do local. And I'm curious what your perspective is on that because that that's been a big use case that people have had for local LMs.
1:34:40It's basically, like, they don't want the big labs to look at their data.
1:34:45Yeah. I think, I mean, the data privacy stuff. I'm I'm living here in Germany where that's a huge, huge topic, people wanting businesses, medium small to medium sized businesses even wanting to have their own local LLMs in house.
1:35:01You've probably seen also PewDiePie created his own self hosted tool where you can keep all of your own basically, all of the data stays with you no matter which of the the LLMs you're using as the, like, compute power.
1:35:18And, yeah, I think I think that or, well, a lot of people think, and I think, like, some of the big tech names on Twitter have also posted about this.
1:35:29A lot of businesses just will not want their proprietary proprietary information out there being worked on, being trained on by these LLMs, and especially if you're sort of, you know, working on something or inventing, you know, something that's very sensitive, maybe pharma companies as well, things like this, there's just going to have to be some form of there's just going to have to be a lot of really good systems and services and and tools for people doing running these things locally, but it's also going to be expensive because the compute required to actually run these things in house is gonna be quite crazy.
1:36:16If you look at PewDiePie's own personal home setup, it's like $25,000 to do some of the basic computing.
1:36:26And so, yeah, I mean, still, even at a company of my size, that's actually worth it because we're going be spending close to that anyway using these external tools.
1:36:37So, yeah, I do think once the Apple version of that comes along, once the version of local building, being able to build local compute and being able to run these LLMs locally comes along in a way that's plug in and play, then I think it's gonna be pretty successful because I think a lot of people are concerned about the privacy element of this.
1:36:59It just doesn't really exist right now.
1:37:03So blueberry buff muffin also says to run an 80,000,000,000 parameter model, you need to spend 5 k on compute. So I'm wondering where the right balance is for less powerful local models.
1:37:15So, yes, he or she agrees with you that, like, you do need to spend money for a rig essentially to run these things. I think that
1:37:28Yeah. But that's nothing compared to what even a small business like mine pays per month in for Claude, team, and open like, you know, that's not actually a lot if you're running a business. I think what's gonna happen is
1:37:41you're gonna you're gonna have your frontier models, and the frontier models are gonna have open source models also. So, like, you're seeing that with Google.
1:37:50You know? Google has Gemini, but they've also got Gemma. And Gemma's, you know, you is it is their open source.
1:37:57And your you know, you you'll have certain tasks that are for frontier models.
1:38:02You're gonna want the very, very best, the Ferrari. But there's certain tasks that you just won't care that they're sort of behind
1:38:09a And little
1:38:10there's certain tasks that also are privacy centric. So those you'll you'll want there. And I think the future is a world where you have some sort of hybrid of those.
1:38:21I I think that today, we're in a stage where it's fun to play with local models because we know that this is the direction, the arrow of progress of where things are going, but they're not here to run everything. They're not you know, you've you've played with Fable.
1:38:37You've seen how amazing they are, and it's just once you play with it, you're kinda like, well,
1:38:43I don't wanna go back in time. Yeah. Right?
1:38:46So Apple show Yeah. Showed a good paradigm for this. You know, they even showed that visual when they were presenting Siri the first time that they didn't launch, but they still had this idea of you've kinda got this bubble of privacy.
1:38:58This inside this bubble, everything is local on your phone, and when it wants to go out and pull something from outside that bubble of privacy, it asks for your permission.
1:39:08It asks for your permission to share data outside of that bubble. And Apple itself, then the the product itself or or the Siri product itself, helps you decide which things it can and can't do within that bubble.
1:39:21And so I think that is the kind of thing that is kind of required in terms of of UX for a local local setup to work where, you know, me running a small business, I know that, okay, when we're doing client proposals, none of that data can go outside of what we're doing here, But maybe when we're doing research, we want something to pop up on screen on screen and say, hey.
1:39:44Like, your your local system here isn't gonna be able to handle this. Can we take this element out of the local thing and plug it into Fable, whatever?
1:39:55But then here's what's gonna get shared. I think that I can't imagine it going a different way other than that for most businesses that won't be able to spend, you know, have a data center of their own.
1:40:10Alright. There you have it, everyone. That's the show.
1:40:14Sip live episode two. Hope this has been interesting. Hope this has been creative for you that's just gotten your brain scratched in a certain way.
1:40:23That's the goal.
1:40:26Keep honestly, let me know if you like this, what you didn't like about it, and and just reply to my tweets. Reply to Jonathan Courtney's tweets.
1:40:37He's at Jiscream, Jiscream on x. And as Stories by Will says, enjoy your gin and tonic, Jonathan Courtney.
1:40:49Thank you. And There will be many. Anything you wanna leave people with?
1:40:55I hope you guys have a great day. I really think the books my favorite moment of this is when we talked about the books thing. I think it's a huge I think if you're I think if you're an entrepreneur and if your goal is to if you want to be a successful entrepreneur, I do think reading, and whether fiction or nonfiction, I don't think that matters so much, but picking up a habit of reading and not just consuming information short form constantly, I do think it will be beneficial for you, your business, your life, your fiance.
1:41:28Your shrimps. That would be your shrimp. That's what I would sign off with.
1:41:33Yeah. The booksing is interesting. I think I do feel a little guilty in the sense, like, I feel like I don't re I'm reading.
1:41:41I don't feel even though I have got a bunch of books back there, I feel like I don't read just enough.
1:41:49I I or I read just enough, and I I could be reading more. So
1:41:55inspiring You're doing alright. You're doing alright, man. I mean, dude, the Instagram the reels are too good.
1:42:01They're too I can't they're too I had to delete Instagram again. I I I went off Instagram from 2020 until two weeks ago, and I was like, I'm gonna promote my podcast on Instagram, downloaded it, already got rid of it. I can't be on there.
1:42:13I can't believe that. Wow. That was quick.
1:42:16I can't do it. We were DMing.
1:42:18I know. I'm gone. I'm already gone.
1:42:20Like, fully gone. I well, maybe I'll I don't know. I I just can't do it.
1:42:25I I can't control myself. I'll just sit there and, like, go, like, you know, flipping through stupid crap that doesn't do anything good for me, especially the messages from you.
1:42:38Don't send me that shit. Yeah. Well, listen.
1:42:42You opened it. You opened it.
1:42:46Thank you so much, guys. Thanks for having me on the show as well, Greg. I really appreciate it having having European on here.
1:42:51You know? Thank you.
1:42:53Thank you for taking the time out of your out of your day. I really appreciate it and and being generous with your time and and everything. So thank you everyone for tuning in.
1:43:03I know you're all busy doing your whatever it is you're doing, you know, work, family, both.
1:43:10So appreciate you, and and I'll see you on the next one. Thank you, everyone. Bye, everyone.
1:43:15Thank you so much.
The Hook
The bait, then the rug-pull.
Two bootstrapped founders open the second episode of a live AMA-and-timeline-reaction show, working through audience questions on getting first customers before spending ninety minutes reacting live to the tweets that defined the week in AI and startups.
How they asked for the click.
MENTIONED ON CAMERA
