TL;DR: The YC application is short and the one-minute video is shorter, which is exactly why they’re hard. You can’t fake clarity in sixty seconds. After running the Barn Sprint three times and sitting through a lot of draft reviews this fall, the pattern is consistent: people don’t fail the application, they fail the work that’s supposed to come before it. Talk to users, ship something, cut the hype, and say it plainly. Then record the video.
We just wrapped the third Barn Sprint, two weeks aimed at the Winter 2027 deadline with one deliverable each: a one-minute founder video. I don’t care much whether anyone gets in. I care that they do the work, because the video is a proxy for it. If you understand your customer and can explain the thing, the minute takes care of itself. If you can’t, no amount of retakes will save it, and it’ll always come in at one minute five.
Fill out the form early, badly
The single most useful thing you can do is open the application on day one and answer every question fast. Not well. Fast. What you know comes through, and so do the gaps you didn’t know you had. Those gaps are the to-do list.
Most people do the opposite: wait for the weekend before the deadline, write slowly, and polish answers to questions they haven’t lived yet. A weekend on a draft is too long. Get it out, show it to someone, and go fix the thing it exposed.
No users is the whole problem
The most common gap I saw this fall was the same one: nobody has used the thing yet. That makes every other answer speculative, and reviewers can tell. YC’s filter is still to build something people want, and the application is built to find out if anyone does.
The fix is smaller than people think. Get one person on it. TestFlight if the App Store is too slow. One real, enthusiastic user changes how you describe the product, because now you’re using their words instead of yours. Three or four is plenty for an application.
Deep tech and industrial are the exception. Nobody is going to TestFlight a new material or a grid component this week. There the weight shifts to the founders: why you? What do you know about this problem that almost no one else does, and how did you come to know it? Years inside the plant, the lab result nobody else has, the regulation you’ve watched trip up every incumbent. Say it concretely. And even here, a real conversation with a buyer who wants it beats another paragraph about the technology.
Same goes for conversations. At kickoff most of the group had talked to fewer than five customers in the previous week. The sharpest one-liners in the room came out of live conversations, never out of another round with an AI. One founder found his pitch, CPG companies going out of compliance, mid-call with a buyer. Another swapped a technical label nobody understood for a plain description of what the product does, and it landed immediately. Drop the ICP language. Talk to people.
Undersell less, hype less
Two failure modes, often in the same draft. The first is underselling: burying the hard thing you built or the real result you got, usually out of a Canadian reflex to not seem pushy. Paul Graham’s own How to Apply to Y Combinator calls the question about what impressive things each founder has done the most important one on the application. Pick the hardest thing you’ve done, and not the startup. They already know about that.
The second is hype. One draft this fall claimed the product mattered more than cancer and climate change. Claims like that read as noise to someone going through a hundred applications a day. The same guide is blunt about it: put the point in the first sentence in the simplest terms, skip the marketing language, and say concretely what you’re building. Here’s what it is, here’s how it works, here’s why it’s good. That copy takes longer to write than you expect. Start now.
The best line in that same draft wasn’t the hype. It was the founder saying, offhand, that the phone was using him. That’s the story. Your own words, about a real problem, are almost always better than the ones you write for the application.
The competitor question is about insight
If your competitor section reads as “same thing, but with an LLM,” you’ve told them you don’t have an insight. PG’s guide makes the point that “well designed and easy to use” isn’t one. Differentiate on purpose, not mechanism: what do you believe about the problem that the incumbents don’t? Name the obstacles too. A flaw you disclose looks a lot better than one they find.
Small answers that matter
When the form asks where you’ll be located, say San Francisco. If you say anything else they’ll ask why, and “staying in Canada for the grants” is a red flag. Alex Peats-Bond’s team at Senpilot is the counterexample worth knowing: they filled out the form, chose not to record the video, and stayed local because every early customer was in Ontario. That’s a real reason, and they knew it early. YC is a fit question, not a status question.
Cap table logic gets the same scrutiny. Unusual splits aren’t disqualifying; unexplained ones are. Know why the numbers are what they are.
The video is a tone test
YC looks for clarity and cadence in those sixty seconds. Spend enough time around the Valley and you’ll notice YC founders tend to talk the same way: short, specific, confident, no throat-clearing. Nobody in Canada does this well by default. We lead with qualifications, hedge, and apologize for not being in SF.
Record a rough take in week one and keep it. Then record again after a week of customer conversations and compare. At Demo Day we judged on distance traveled, not polish, and the gap between the two takes was the best evidence of the work.
If you get the interview
Read How to Convince Investors first. The point of the essay is that you don’t win by pitching. Partners are trying to figure out whether you’re formidable: will you actually do this, keep moving, and tell the truth. Repeat the question before you answer; it slows you down and shows you listened. Don’t get dragged into a market size debate; go back to why now. If you don’t know, say so. Don’t brainstorm a pivot on the spot. Have a real answer for why you two are working together, not “we met at a hackathon.”
What YC actually teaches
Mazin Al-Ani said it at our first dinner this fall: office hours never challenged their features. Every conversation went to who are you selling to, and how. If you shorten YC to one line, it’s teaching technical founders to sell, fast. So the best application prep is the same as the best company prep. Sell something, then write it down.
And keep perspective. Getting an interview often takes three or four tries. YC runs four batches a year now, so a no today is cheap, and it often reflects timing, the partner, or a competing company rather than you. Applications compound. The Waterloo Mafia directory counts more than 140 Waterloo-connected companies that have been through YC; most of them didn’t get in the first time either.
Key takeaways
- —Open the form on day one. Answer everything fast and badly. The questions you can’t answer are your plan for the next two weeks.
- —One real user beats a better paragraph. Ship to someone, even through TestFlight, and borrow their words. Speculation is the most common weakness in early applications.
- —Cut the hype and stop underselling. Specific beats grand. Lead with the hardest thing you’ve actually done and describe the product the way you’d explain it to a buyer.
- —The video is a proxy for the work. If you can’t say it in a minute, you don’t understand it yet. Record early, record again, and compare.
- —Be formidable, not convincing. In the interview, tell the truth, stay on what you’re building, and let them figure you out.
- —Treat it as a fit question. YC is right for a lot of companies and wrong for some. Know which you are before you spend the dilution.