On July 27, a few hours before Y Combinator's 8pm Pacific deadline, we pressed Submit on our application for the Fall 2026 batch. Two founders, one room in Nha Trang, five days after our public launch.
Then we sat there for a minute, because the moment was stranger than we expected.
We just entered the world's largest batch review
YC doesn't publish exact application counts, but outside estimates put a recent cycle at 15,000–20,000+ applications for roughly 150–250 spots — an acceptance rate somewhere between one and two percent. Every one of those applications gets read, sorted, and ranked against a bar.
EvalLens is deal screening for accelerators and VC funds. Structured first reads for large piles of applications: that's the product, that's the pitch, that's the company. So on July 27 we became a line item in the biggest version of the exact problem we sell software for.
When you build screening tools, applying to things stops being just applying. It's field research.
(Yes, the temptation to run our own application through our own panel was enormous.)
What we told them
An application form is a brutal editor. Fifty characters for what the company does. A few paragraphs for five months of work. Here's the version that survived the cut:
Feb 27 — first commit, as AI Jury (Amazon Nova hackathon)
April — pivot to EvalLens: same engine, new customer
July 10 — first live event: five startups scored end-to-end, ranked leaderboard
July 22 — public launch
July 27 — YC application submitted
Behind that timeline: 1,135 commits, 84 database migrations, and 1,000+ evaluation runs spent calibrating the judging panel. A VC fund that watched the July live event signed a letter of intent to pilot EvalLens on its own deal flow in September. Crypto Executives signed on as our first distribution partner.
And one line we were oddly proud of. The form asks what your company does, in fifty characters or less. Ours came in at forty-seven: AI deal screening for accelerators and VC funds.
The honest odds
This is a second attempt. Yaroslav applied solo in 2024 with a different company and got the polite no that most applicants get. Nothing about that experience made this application feel safer. What made it feel different was arithmetic: last time the application described plans. This one describes things that already happened.
We'd love to tell you we're confident. The honest version: we can count. One to two percent is one to two percent, and no traction paragraph changes the denominator. What we actually believe is narrower — that our application is easy to evaluate. Claims with evidence next to them. Dates that check out. Numbers a reviewer can verify without taking our word for it.
That's not a YC strategy. It's just how we think applications should be written, since we spend all day reading the other kind.
Why announce an application at all
Companies usually announce acceptances. Announcing an application looks a bit like framing a lottery ticket.
But we're building this company in public, and the application itself turned out to be work worth showing. The form forced us to compress five months into claims a stranger can check — which, incidentally, is the same discipline our product enforces on every deck it reads.
Decisions for on-time applications land by August 28. We'll publish the outcome either way. That's the deal with building in public: you don't get to report only the rankings you like.
What happens while we wait
Nothing changes. The September fund pilot is in preparation, the next live events are being scoped, and the product keeps shipping. If YC says yes, the batch runs October through December in San Francisco, and we'll be there. If it's a no, we'll still be reading decks the next morning.
If your inbox looks like a smaller version of YC's — hundreds of applications, one rubric, not enough weeks — that's the problem we build for. Book a demo, or try the live demo and run a deck through the lens yourself.
Twenty thousand decks, and somebody has to read them all. We have thoughts on that.
