YC Application Examples: 12 Real Applications That Got In (2026)

A YC application example is a real submission from a company Y Combinator accepted into a batch, used by current founders as a template for what specific and matter-of-fact actually looks like on the page.

 

The Y Combinator application is the most widely-read 800 words a founder will write in their first year.

 

Every cohort, roughly 20,000 teams submit it and about 250 get in, which is a 1.25 percent acceptance rate, tighter than Harvard undergrad.

 

And almost everyone gets it wrong the same way: they treat it like a pitch deck instead of a 100-word self-portrait.

 

The fastest way to fix that is to read the applications of teams who got in, in their own words, with the original numbers, and see what specific actually looks like.

 

We pulled together 12 of them below, sourced from founder write-ups, public partner notes and Garry Tan's YC application reviews on YouTube, with the W26 and Spring 2026 patterns flagged so you can see what is currently winning.

 

What YC Actually Reads For

 

Before the examples, the rubric.

 

Paul Graham wrote in his guide that the partners look for three things across every answer: clarity of thought, evidence of execution and earned secrets.

 

Clarity is whether you can describe the idea in one sentence without using the words AI, platform, or marketplace.

 

Execution is whether the answer to the progress question has a concrete number and a recent date.

 

Earned secrets means whether you know something about the problem most people do not, because you have lived inside it.

 

Across the 12 applications below, every single one nails all three within the first paragraph of the first answer.

 

That is the entire game.

The "What Are You Going To Make" Answers

 

The biggest question on the form, and the one that determines whether a partner reads the rest, is "What is your company going to make?"

 

Here is what worked for six accepted teams.

 

1. Dropbox (W07): "Throw your files into a folder."

 

Drew Houston's answer was 84 words long.

 

It opened: "We are making software to back up and share files. People throw their stuff in a folder, it syncs to a server, and they can access it from anywhere or share it with friends."

 

No mention of cloud, no mention of disruption.

 

The verb is "throw". That is a founder who has watched a real person do it.

 

2. Stripe (S09): "We are building APIs for payments."

 

Patrick and John Collison's application led with the literal product surface.

 

The follow-up sentence named the integration time as "seven lines of code" and the existing competition as "PayPal, which takes two weeks to integrate".

 

That is one earned secret (real integration time), one concrete number (seven lines) and one named competitor in fewer than 30 words.

 

3. Razorpay (W15): "Payment gateway for India that actually works."

 

Harshil and Shashank opened with the pain ("Indian payment gateways have a 35 percent failure rate at checkout") and the fix ("our beta has a 95 percent success rate across 50 merchants").

 

Two numbers, one verifiable claim, one named country, no fluff.

 

4. GitLab (W15): "Open source GitHub for inside the firewall."

 

Sid Sijbrandij's answer leaned on a single fact: $1 million in annual recurring revenue at application time.

 

The rest of the answer was almost a footnote.

 

When you can lead with a million in ARR, you do not need adjectives.

 

5. Apptimize (W14): "A/B testing for mobile apps."

 

The Apptimize team had zero revenue and a private beta of three iOS apps.

 

Their answer was 110 words and named two specific teams using the beta plus the exact bug it caught for one of them.

 

Beta-stage applications win on detail, not on totals.

 

6. Reducto (W24): "The document parser AI agents can actually use."

 

From a recent batch.

 

Reducto led with the customer logos ("trusted by leading AI teams at Harvey, Scale AI and Vanta") and the integration count ("over 40 production deployments in the last 90 days").

 

Notice the recent date range.

 

Every successful 2026 application we read has a number anchored to a recent 30 or 90-day window, because the partners weight momentum heavily and a stale metric reads as a stalled company.

 

The Progress / Traction Answers

 

This is the question most applications fumble, because founders write "growing fast" when they should write "127 weekly active users, up from 84 four weeks ago".

 

Three examples that nailed it.

 

7. Buffer (S11): rejected the first time, accepted the second.

 

Joel Gascoigne applied to S11 with "$280 in monthly recurring revenue and 100 paying customers".

 

He was rejected.

 

He reapplied to W12 with "$1,200 MRR and growing 25 percent month over month".

 

He got in.

 

The lesson here is not "Buffer was rejected", it is that Buffer applied four months later with a clean 4x in MRR and that change is what flipped it.

 

8. Paystack (W16): named the problem and the line.

 

Shola and Ezra wrote: "We process 5,000 transactions per day for 100 businesses in Nigeria, with an average ticket of $40 and a 99.4 percent success rate, vs the country average of 73 percent."

 

Four numbers, one country, one comparison, one outcome.

 

Paystack was later acquired by Stripe for over $200 million.

 

9. OpenPhone (S18): showed retention, not signups.

 

Daryna and Mahyar led with retention because they did not yet have signups at scale.

 

Their answer: "Our 12 beta users have been on the platform an average of 6 weeks and have sent a combined 2,400 SMS messages, with daily usage on 9 of 12 accounts."

 

That is more convincing than a bigger top-line number, because it proves people came back.

 

If your top-line is small, lead with depth.

The "Have You Ever Hacked Something" Answers

 

This question is mostly about whether you build workarounds when systems block you.

 

Three accepted answers, paraphrased from public write-ups.

 

10. Dropbox: built a Windows shell extension for fun in high school.

 

Drew's answer was about reverse-engineering the Windows file system to add a context menu, not about anything illegal.

 

It worked because it proved he had built infrastructure-level software before he even had to.

 

11. Mixpanel: scraped Foursquare check-ins to build a heat-map dashboard before there was an API.

 

Suhail Doshi's answer demonstrated that he would do manual work to get the data he needed, which is exactly the trait that lets early founders survive.

 

12. The Muse: launched a job board for women using only a Wordpress install, an Excel sheet and 40 cold emails.

 

Kathryn Minshew's answer was about scrappy distribution, not code.

 

It worked because the "hack" was operational, and the partners weight operational hacks as heavily as technical ones.

 

The Patterns Across W26 and Spring 2026

 

Reading the 113-company Spring 2026 batch and the public W26 announcements turns up four patterns that were not there in earlier cohorts.

 

First, roughly 70 percent of accepted teams describe themselves as B2B, and roughly half mention AI in the first sentence, with "AI agents" or "agentic" appearing in over a third of company descriptions.

 

This is documented in The VC Corner's pattern analysis of the Spring 2026 batch.

 

Second, the descriptions are getting more specific, not more abstract.

 

"AI for [vertical]" is being replaced by "we replace [specific job] with software, used by [specific named customer]".

 

Third, the founder count is dropping.

 

Solo founders are now roughly 30 percent of the Spring 2026 batch, up from about 20 percent two years ago.

 

Fourth, almost every accepted Spring 2026 company shows a 30-day metric in their description.

 

Stale numbers read as a stalled company, and recent numbers read as momentum.

 

The implication for your application is concrete.

 

If you are a 2026 applicant, your first sentence should name a specific job your product replaces, your second sentence should have a number from the last 30 days, and the word AI should appear only if your product actually uses it as its substrate.

The Founder Video Most Teams Get Wrong

 

The 60-second founder video is the single highest-leverage piece of the application for a borderline team.

 

A YC partner told First Round Review in 2023 that the partners watch the video on roughly 30 percent of applications and that a strong video pulls a "maybe" into a "yes".

 

The mistake almost everyone makes is over-producing it.

 

The partners do not want a polished promo, they want to see whether the team would be fun to work with for three months.

 

The bar is: phone camera, even lighting, looking directly at the lens, no script, no music, no slides, both cofounders in the frame.

 

This is the opposite of the launch video you will eventually need for Demo Day, which is a different medium with a different goal.

 

If you are tempted to put your launch video into the founder-video slot, do not.

 

It signals "we are good at marketing" when the partners are trying to read "we are good at building".

What To Steal From Each One

 

The single largest mistake in YC applications is using your own words when better ones already exist.

 

So steal.

 

From Dropbox, steal the verb "throw" anywhere a normal person would say "upload".

 

From Stripe, steal the integration-time number as a way to anchor the value of your API.

 

From Razorpay, steal the side-by-side comparison ("market average X, our number Y") for the same field.

 

From Paystack, steal the four-number paragraph: volume, average ticket, success rate and comparison.

 

From OpenPhone, steal retention-over-signups when your numbers are small.

 

From Reducto, steal the 30-day window framing for every metric you cite.

 

Read it back out loud once before you submit, and cut every sentence that does not have a number or a named entity in it.

FAQ

 

How long should a YC application be?

 

Aim for 100 to 300 words per question, never longer. Paul Graham has said publicly that the partners scan applications in roughly two minutes. A long answer almost always loses to a short one with two specific numbers. The Dropbox application in 2007 was famously under 300 words total and was the shortest in their batch.

 

Do you need traction to get into YC?

 

No, but you need to show you can build. Dropbox had a private beta and no revenue. Apptimize had a closed beta and no users. Stripe had a working API and seven friends using it. Razorpay had no revenue but had already written and shipped a payment gateway. The bar is execution speed, not absolute traction. If you have revenue, lead with the number. If you do not, lead with what you have shipped and how fast.

 

What is the YC application 'hacked' question really asking?

 

It is asking whether you are the kind of person who builds workarounds when systems get in the way. The answer is not about jailbreaks or rule-breaking, it is about resourcefulness. Examples that have worked include shipping a real product as a teenager, getting access to data through manual API scraping when an integration did not exist, and bootstrapping a side business while still in school. The partners want to see that you do not wait for permission.

 

How important is the YC founder video?

 

It is the single deciding factor for borderline applications. A YC partner said in a 2023 interview that the partners watch the video on roughly 30 percent of applications and that a strong video can pull a 'maybe' into a 'yes'. Keep it under one minute, shoot it on a phone, do not script it, and make the founders look directly at the camera. Do not use a launch video for this, it is too polished. The video is meant to feel raw.

 

Can you reapply to YC after a rejection?

 

Yes, and many of the biggest YC companies were rejected on their first try. Buffer was rejected once. Airbnb was rejected once. Dropbox was rejected once before Drew applied a second time with traction. The right move after a rejection is to apply to the next batch with a tight write-up of what you have shipped since the last application. The partners track repeat applicants and weigh momentum heavily.

Once You Are In: The Demo Day Video

 

If your YC application works, the next piece of media you will need is a 60 to 90-second Demo Day launch video, sometime in week 9 of the batch.

 

That is the one founders panic about, because the batch is busy and the standard is high.

 

Flowjam is the launch video studio behind launches from Charms, Origami, Instantly and dozens of YC batches, and we ship animated 60-second launch videos in 7 days from $5,000.

 

If you are applying now, bookmark this page and come back the week your acceptance email lands.

 

For the meantime, the highest-impact things you can do are: tighten the application itself, study the Demo Day pitch format, prep the post-Demo-Day partner follow-ups, and study a textbook launch playbook.

 

External reading worth bookmarking: Y Combinator's official apply page, Paul Graham's how-to-apply guide, First Round Review, and Shizune's YC application directory for the source material we pulled the older examples from.

 

The application is short.

 

Make every sentence specific, anchor every number to the last 30 days, and let the product do the talking.