We've backed 11 companies from the current YC P26 batch. None of them were Demo Day investments. Every check we wrote came weeks before the event — in some cases before the founders had even started their investor outreach.
This isn't a timing accident. It's a deliberate strategy built around five specific signals that tell us a company is worth backing before anyone else has made up their mind.
The companies we backed from P26: Prototyping.io (autonomous manufacturing), Astraea (clinical trial data automation), Manicule (AI growth for dev tools), TesterArmy (AI-native QA), Arzana (autonomous ERP for manufacturers), Standout (agentic hiring marketplace), Hub.xyz (AI training-data infrastructure), Arga Labs (digital twins for AI agent testing), Chronicle Labs (production simulation for enterprise agents), Cignara (AI agents for B2C support), and — as of yesterday — Project X, whose product Infinity hit 20,000 signups, 20M+ social impressions, and ran out of server capacity — all within 15 days of launch.
Here's what we were reading when we made these decisions.
Signal 1: Traction Before Marketing
The most reliable indicator of a breakout company isn't growth rate — it's the source of growth. The best companies we've backed have customers before they have a marketing strategy.
Prototyping.io came to us profitable, bootstrapped, and already trusted by Tesla, Zipline, Rivian, Generac, and Light Matter — a two-person team, before taking a single outside dollar. These clients didn't come from cold outreach. They came because Prototyping.io solved a genuine problem nobody else was solving: autonomous manufacturing validation that normally requires months of back-and-forth with engineering teams.
Arga Labs showed the same pattern from a different angle. Within weeks of launch, customers were arriving in volume. But the sharpest signal wasn't the volume — it was that a third of those customers liked the product so much they invested their own angel money into the company. Paul Graham wrote a bigger-than-usual check. When your customers invest in you, you don't have a marketing problem.
Signal 2: Revenue That Compounds
We look for early evidence of a referral or expansion dynamic. Not just revenue — revenue that begets more revenue.
Arzana's first customer, Milltown Paper, recovered over 100 hours per month and avoided a six-figure hiring decision. Within one week, they had referred two new customers. That's the shape of a flywheel at turn one — before any investor had written a check.
Manicule showed the content flywheel. Super Memory used Manicule to go from scattered documentation to 1M+ SEO impressions, with answer-success rate jumping from 60% to 90%. The growth was so obvious that Manicule's pipeline was full before their raise had formally begun. Naman and Shreyans were turning away inbounds.
Signal 3: The Timing Dagger — A Window That Won't Stay Open
Every great early-stage bet has a timing component. The market is ready, the technology is ready, but the incumbents aren't. That gap — usually measured in 12 to 24 months — is where the best returns are made.
Astraea is the clearest example from this batch. Clinical trial data is among the slowest, most painful data work in technology. The incumbents — SAS and its peers — won't have anything close to what Astraea is shipping until 2027–2028. Enterprise contracts are already signed. The window is open now, and Astraea is sprinting through it with 99.5% accuracy on data that hospitals typically spend months cleaning by hand.
TesterArmy occupies a similar position. It's the only truly developer-native QA product in a field where every other player hides behind enterprise sales cycles. No scripts, no selectors — a vision-based agent that understands your UI the way a human does. Meta, DirecTV, and GM were already using it. The window for a developer-first entrant is right now.
Signal 4: Founder-Market Fit So Deep It's Biographical
We pay less attention to pitch decks than to origin stories. The best founders don't choose their markets — they were made for them. You can feel it in the first 90 seconds of a conversation.
Naman Bansal of Manicule has been writing technical content since he was 13. He and co-founder Shreyans Jain first applied to YC at 15. Their entire adult lives have been building toward what Manicule is. The product isn't a pivot — it's the inevitable destination.
Ayman Saleh of Chronicle Labs spent 8 years at NASA JPL on the Mars rover landing system. His co-founder helped return the Boeing 737 Max to service after two fatal crashes. These are founders who understand failure modes that ordinary software engineers never encounter. Chronicle brings aerospace-grade simulation discipline to enterprise AI agents — and only a team like this would think to build it.
Akira, co-founder of Arga Labs, is 17 years old and a former Goldman Sachs quant. He built the digital twin infrastructure used to test AI agents the same way aerospace engineers test physical systems. The origin story is inseparable from the product.
Signal 5: The Viral Early Number You Can't Fake
The final signal — and the sharpest — is the number that appears before the company has even tried. It's not a growth metric from a campaign. It's demand that arrives uninvited.
Yesterday, we backed Project X, the company behind Infinity. Fifteen days into early preview, they had:
- 20,000+ builder signups
- 1,160% growth in 14 days
- 20M+ social impressions
- Server capacity exhausted — rate limiting required
- 130+ teams already building — including from Reliance Industries, CEMEX, and Cognizant
Infinity is Project X's product: a web-based OS where every application you open runs on its own independent compute and GPU inside a single shared workspace. CAD, simulation, ML training tools — all running simultaneously on any device, including a Chromebook or an iPad. A $5,000 workstation, accessible from anywhere, at a fraction of the cost.
That kind of pull doesn't come from marketing. It comes from building something people immediately understand they need. We saw the number. We moved before Demo Day.
The Pattern
What connects Prototyping.io's enterprise traction, Arga Labs' angel-investing customers, Arzana's referral flywheel, Astraea's timing window, and Project X's Infinity viral launch?
None of them needed Demo Day to validate their thesis. The signal was already there — in paying customers, compound revenue, a closing window, biographical founder fit, or demand that exhausted capacity before the founders could even respond.
Demo Day is useful. It creates a forcing function, sharpens positioning, and brings new investors to the table. But it also creates a pricing event. The best companies walk into Demo Day already oversubscribed. By the time the crowd arrives, the most interesting entry points are usually gone.
We'd rather be early and right than perfectly timed and expensive.
