Everything is a wrapper. Until you understand this stack you don't understand the AI space.

VC money is only useful for meta and google ads nowadays (prove me wrong)

hey so, which country will be the first to give an AI agent real legal status? i want to register an agent as its own legal person. it would start companies under its own name, sign contracts, open a bank account... legally it would work like my child. it acts on its own and i carry the responsibility for what it does. so where will this be allowed first? the US, estonia, Cyprus, south korea, BVI, one of the gulf states??

which country will be the first to give an AI agent real legal status? I want to register an agent as its own legal person. it would sign contracts, open a bank account and start companies under its own name. legally it would work like my child. it acts on its own and i carry the responsibility for what it does. so where will this be allowed first? the US, estonia, south korea or one of the gulf states?

Who's still optimizing their homepage for humans? If you haven't already you should already be optimizing for agents and not just the homepage: the entire journey Because like it or not, agents are going to be picking software vendors...

agents will pick your next software vendors. GEO and AEO mattered the past 12 to 18 months. a human asked ChatGPT and got a recommendation. agents make very different ones. here is what i think sits underneath it. a lot of software pricing has been hidden behind complexity. tiers, seats, "contact sales", a demo call before you see a number. that worked when a human decided with emotions and relationships. an agent reads the price, the coverage and the quality and picks. no dinner, no golf, no logo on the wall. so i feel like we finally entered the real era of homo economicus. the rational buyer who only looks at the facts never really existed with humans. with agents it does. agent economicus. we're already doing this at looot. our agents choose between data providers for each enrichment call. they route on quality and cost and fall back to the next provider on a miss. the user never picks a vendor. if your price sits behind a demo call, an agent skips you. if your price is a clear number with clear coverage, it can pick you. so if you sell software, make your price and your coverage readable by a machine. let me know what you think.

We are the command line generation. potentially the biggest revolution in knowledge work is here... how we used to work = gtm, marketing, product and revops people used to live inside ten tabs/ten tools....We used to have to interface between different tools...A CRM here, a task management tool there + an ad platform here and an enrichment tool there, a prototyping tool here, a communication tool there, a lifecycle messaging interface etc etc.... Then export some data, play around with it in Excel, turn it into a report for management, build that quarterly business review (in 4 different formats for 4 different audiences). Most of the day went to moving between them. This ate up so much time and creativity. We were basically human interfaces for software...Juggling data across these interfaces. If you were smart/allowed you'd create a few automations on the backend. But still, juggling through different UIs Times have changed for the more advanced: How we work now = open the terminal and tell the agent what you need, often times it tells YOU what is worth your attention. ex: pull the accounts that changed this week, enrich the new ones, update the crm, draft the outreach, push the budget change to the ad platform. it calls each tool directly. the crm, the enrichment provider, the ad account and the lifecycle tool all sit behind it as data + logic. The excel step happens in the same session. The QBR gets written once from the same data, then the agent cuts it into the 4 versions for the 4 audiences. Your day goes to deciding + checking. Moving between tools is the agent's job now. "What about in the context of teams?" you mgith be asking? Well it runs on shared rules. The ontology lives in one shared repo (the business rules, the defintions) and any change to it gets reviewed like code. Each person's agent only sees what their role is allowed to see + can only push what their role is allowed to push. Writes back to the crm (the single source of truth) wait for a yes from the owner and land in a log anyone can audit. It's all about governance. But it's no longer about switching between all these interfaces. The interface stops mattering. what matters is the context behind it. clear business rules = ONTOLOGY (some call it 'context'). The productivity gains are mind blowing. The QUALITY of what we can focus our human brains on, too. You might say it's a bubble but there's way, way, way more than you think. This is spreading fast and i think it goes further than most people expect. Sure today this is builders (developers) first and in other areas it is considered a 'bubble'. Clients and most teams still need a normal UI. but... My bet for the next few years is that most gtm, marketing, product and revops work gets run from the command line by whole teams. And if you need a UI...you whip it up yourself. We are the command line generation. And a lot of you are reading this ;)

the future IS headless... sure you'll say it's a bubble...but... it's a bubble of excellence and over-performance. what I'm seeing in this so-called bubble: people wire a dozen amazing tools into their Claude, their Claude Code or their Codex through MCP or API and then they build and run GTM campaigns, dashboards and apps from the terminal. The future marketer, product person, revops, product marketer and builder all end up in the same place.. a command line and a conversation. “The future is headless”. Need a user interface? wire it up yourself. true stories (I'll call these people out by name in a separate post) A guy I met who runs an old consulting firm (big clients, family business), he's 27, he took a long look at what he had been selling for years to clients who could rarely afford him. He rewired the whole thing into a fully deployed SaaS product from his command line (we're not talking POC here, it was live and working)...and the clients who couldn't afford him (they have all the contacts) are signing up super fast. It was then hardened with a 'real dev'. He built a new revenue line from his command line. CEO who owns an agency decided to rebuild it from scratch around AI and a command line. The agency he runs today looks very little like the one he started with. AgencyOS. A CEO of a big/well-known SME (who I admire for his growth mindset), in his late 50s opened his Claude Code one weekend and whipped up a CRM for his own company, then added enrichment + application layer + sequencing. then showed it to the rest of the business, because he wanted everyone to internally to see (change management). It's being built now. A CMO + RevOps who work fully out of their command line, partner with a dev and sunset 500k of Saas..from the command line. Pipeline is through the roof. I know it's a bubble. But that bubble is winning. Period. Sure, most companies are nowhere near this, normal for a wave this young. Inside the bubble, though, people are doing the work of 10, 100? (btw: every board I hear about is asking for higher revenue per employee). They work the same way every time. Claude, Claude Code or Codex in front, a dozen tools wired in through MCP and campaigns, dashboards and apps coming out the other end. The stack I see most 👇 → Claude Code or Codex to work → looot for data, dozens of providers behind one MCP key, pay per use → Supabase as source of truth → Vercel for the app and pages → Resend for email → Composio for integrations → OpenRouter for models (or just use Vercel's gateway) → Sequencers to send The future is headless. (and again the future's already now. It's just not evenly distributed) shameless plug: that's all btw why we built looot 👇 → one MCP key (or API) for all your data and tools (Claude, Claude Code, Codex, Cursor, Gemini CLI) → emails, phones, LinkedIn profiles, company news, job postings, ads, SEO, local leads, influencers → no subscription, you pay as you go

Let’s test if you’re a financial genius. imagine France was under your control what do you go for first? - Stimulate growth? - increase taxes? - Reduce public spending (where)? … and after that add one last variable: take into account which group of the population votes the most (60+ yr olds) and gets you re-elected? is this what they call a catch 22?

🫠 When you have no idea what you’re talking about ….and your AI assistant crashes during the call C’mon you know who you are ;) (Context: Prince Harry Abruptly Ends Speech After Teleprompter Failure…)

It wasn't super simple to understand all of the power behind looot... so we built this cool thing.

subscriptions are dead. the future is headless…… here comes the rant... let me explain, but first, i think we need to talk about how we pay for software. here’s the history of software pricing in four lines.. 1/ years ago = one big license, paid upfront 2/ then SaaS = flat tiers. €29, $49, $99 3/ then per seat + credits = you pay for a credit and never know if it came back 4/ now headless = your agent calls the tool, you pay for the result just look at how a lot of us (and this is growing very very fast) work. people wire a dozen amazing tools into their Claude, their Claude Code or their Codex through MCP or API and then they build and run GTM campaigns, dashboards and apps from the terminal. The future marketer, product person, revops, product marketer and builder all end up in the same place.. a command line and a conversation. “The future is headless”. Need a user interface? wire it up yourself. the pain though?? you need a bunch of subscriptions AND a bunch of api keys for this to work. annoying. the even BIGGER pain = when your agent is the one using the tool, a flat monthly fee makes no sense! the agent never logs in. it wants a result. so you should pay for the result. what i believe = you pay per result. one key. the cheapest price for the job, the best quality for the job, the best coverage for the job. all three at once. BTW I think SaaS does survive (not a huge fan of the saaspocalypse talk). BUT the ones that keep per seat pricing to get wiped out. not overnight of course, but it is happening. I don't see a single new AI-native startup needing to buy a CRM for example. I've been saying this for a long time.. shameless plug: that's all btw why we built looot 👇 → one MCP key (or API) for all your data and tools (Claude, Claude Code, Codex, Cursor, Gemini CLI) → 100+ providers already wired in, we signed the deals so you don't open the accounts → no subscription, you pay as you go → you only pay when a call returns a result → the providers compete on price for every job (benchmarks coming soon) → best quality and best coverage per job, we benchmark them in the background and route accordingly → emails, phones, LinkedIn profiles, company news, job postings, ads, SEO, local leads, influencers → a spend cap per API key, OpenRouter style, so you control cost per client → nothing stored, we're a subprocessor (European guardrails possible) → 200+ recipes you paste into your CLI, build one and earn when others use it → setup in 10 seconds: connect an agent and paste one block → video and voice next, because it's all just wiring up tools i could be wrong. and it's a bit more nuanced. But I do think this is where we're headed. PS: the future is also BUILD (with a tiny bit of buy). build the engine. buy the data underneath, per result, on one key.

I used to (and still do) tell early stage entrepreneurs...Product-market fit is like love. When you find it, you'll just know. This screenshot is what it feels like and looks like. (1 week after launching) Next step, of course, is to ensure usage and retention. But when you find something that truly solves a pain, you'll feel it right away....

I used (and stilll do) tell early stage entrepreneurs...Product-market fit is like love. When you find it, you'll just know. This screenshot is what it feels like and looks like. (1 week after launching) Next step, of course, is to ensure usage and retention. But when you find something that truly solves a pain, you'll feel it right away....

I made our entire Product Hunt launch video with Claude Code and one GitHub repo. The sound too. repo is called: "awesome-opus5-5-videos" go find it!

What you think a modern GTM stack is... Claude Code + Clay + Apollo + a signal engine + a sequencer What a modern GTM stack ACTUALLY is...👇 1. Mindset: the future is headless. We work out of the command line or out of a conversational interface (Claude Code, Claude, Codex etc) 2. Looot gives your agent EVERY the cheapest data and enrichment output behind one MCP key and with no subscriptions = pay as you go (Firecrawl, Hunter, BuiltWith, Apify, Exa, Brave Search, Bright Data, Prospeo, LeadMagic, People Data Labs, TheirStack, ZeroBounce, FullEnrich and plenty more) 3. Supabase holds your data and stays your source of truth (who needs a CRM?) 4. Vercel hosts your app and your pages build out all your engines: Signal Engine, ABM Engine, Lifecycle messaging, SEO/GEO, Outreach, Ads optimization etc... 5. Resend sends your emails and routes your lifecycle messaging 6. OpenRouter routes every call to the right AI model...use cheap models for simple tasks (99% of your uses)... only use the big expensive models for deep thinking tasks 7. Duh: Meta Ads and Google Ads run your paid campaigns 8. Composio connects your agent to all your other apps 9. The usual suspects: HeyReach, Instantly and/or lemlist are the sequencers that send your outreach A CRM like HubSpot or Attio is optional. All you need is one source of truth. Claim = "the future is headless" Question = What is in your stack that I forgot?