@13yearoldvc

in essence i think agentic payments is a good asymmetric bet for stripe and big players: - it has limited downside (doesn't interfere with their main business) - and unlimited upside (ai narrative/hype for them, and in case if it actually does happen they could catch it) but for most startups, agentic payments currently is a reversed asymmetric bet: - it has limited upside (how to compete with incumbents, is the market even big enough) - but unlimited downside (too much uncertainty, no signal, opportunity cost to tackle other more meaningful and real markets)

https://t.co/DGdLu79krs

hot take: meetings are the most underrated productivity tool in tech. get gemini to transcribe google hangouts, feed the transcript into claude alongside slack, github, docs and calendar. you now have a living strategic brain that updates every time your team talks. + speaking is 4x faster than typing. every meeting is a live data harvest from every brain in the room

sf bro be working on his laptop while lifting I walked over and saw he’s coding with Claude like bro get yourself a hermes agent on ur phone instead https://t.co/CNhJn0wVdr

I went through the founding stories of the world's top 50 talent-dense companies right now and tried to figure out how they actually got to pmf 7 patterns kept showing up. The Practitioner's Revolt - founder hates their own tools, builds the replacement. Linear's three founders literally bonded over hating Jira at a bar. One of them had already built a Chrome extension to fix Jira at Airbnb and it went viral internally. They took Big Tech jobs on purpose to build pedigree and savings, then quit to build what they actually wanted. Profitable in two years. Still under 30 people at the time. The Platform Shift Surfer - a new platform shows up and you build the definitive tool for it before anyone else figures out what's happening. Cursor's founders were obsessive Copilot users who realized the whole IDE needed to be rebuilt around AI, not just have it bolted on. $100M ARR in 12 months with zero marketing spend. That's not a growth hack. That's just being right about what developers need at exactly the right moment. Wedge-and-Expand - you enter through something so narrow it's almost embarrassing, then you never stop expanding. Ramp started as a corporate card. Then expense management. Then bill pay. Then procurement. Then travel. Then treasury. They shipped 270 features in six months last year. Their bill payments volume is now bigger than their card business. The Talent Magnet - team is so stacked that the company's existence is the gtm. Thinking Machines Lab raised $2B before they had a product. Meta offered one of their engineers $1.5B to leave. He said no. (He eventually said yes, but still.) Unsexy Infrastructure - you build the boring plumbing everyone in a growing ecosystem needs. Modal Labs is like 120 people building serverless compute for AI workloads. Nobody writes breathless Twitter threads about it. Developers love it. Stripe bought Metronome for $1B and all Metronome did was usage-based billing metering. Community Marketplace - you build the community first and let the product emerge from what they actually need. Whatnot started selling Funko Pops. Their CTO coded livestreaming in six weeks. First stream sold $5K in a few hours. Now they do $6B in annual sales and users spend 80 minutes a day on the app. They talked to over 100 investors before anyone would give them $300K. Vertical AI - you take general-purpose AI and make it dramatically better for one specific industry. Harvey is the wildest story here. A securities litigator and a DeepMind researcher were roommates. They pulled 100 questions from Reddit's legal advice sub, ran them through GPT-3, and 86 passed review by practicing attorneys. Cold-emailed Sam Altman. Got the OpenAI Startup Fund's first check. Then four people operating out of an Airbnb signed Allen & Overy - one of the largest law firms on earth, 4,000 lawyers - as their first customer. Every single investor told them to start with small firms. They ignored all of them. $190M+ ARR now. The thing that surprised me most: the strongest companies hit three or four at once. Cursor is a Practitioner's Revolt and a Platform Shift Surfer and a Wedge-and-Expand all at the same time. They just wanted a better code editor and the rest fell into place because the timing was right and they were the right people. A few other things that kept coming up across all 50: Repeat founders won at a disproportionate rate. At least 20 of the 50 companies were started by people who'd already built and sold something. Ramp's founders sold Paribus to Capital One. Decagon's founders both had exits. Whatnot's CEO previously sold Kit to Patreon. The playbook knowledge compounds. "Go straight to the top" almost never happens cold. Harvey got to Allen & Overy through OpenAI's network after the investment. Mercor got to the top AI labs by first proving they could reliably place engineers at smaller companies. The shortcut is either a warm intro through your investors or a proof of concept so undeniable it functions as its own introduction. And in the AI era specifically - building speed is not the moat anymore. Everyone can build fast now. The companies surviving against the big labs aren't technically smarter. They're just deeper inside their customer's daily workflow than any lab wants to go. Harvey has ex-lawyers embedded inside law firms doing change management. That's almost a services business and OpenAI probably doesn't want to run a services business. The single clearest takeaway from all 50 stories: PMF comes from compression, not expansion. Make your world smaller. Own it completely. Then expand from strength.