Y Combinator / Fundraising Beginner

24 ChatGPT prompts for YC-style startup applications that actually sound founder-led

You sit down to write your YC application. You open ChatGPT. You type “write me a YC application essay.” You get a wall of generic founder-speak that sounds like everyone else.

The whole thing is broken in 10 seconds.

Here’s the thing: the Y Combinator application is short on purpose. It is built so a partner can read 5,000 of them in a sitting and still tell who actually ships. YC partner Garry Tan has read over 6,000 YC applications in his career. Michael Seibel has read over 17,000 and sat in on 3,100+ interviews. They are not looking for polished prose. They are looking for proof you can do the work.

So the right way to use ChatGPT prompts for YC application writing is not to outsource the voice. It is to compress your raw material - customer calls, Notion docs, Loom transcripts, Linear tickets, your Tana knowledge graph - into crisp, founder-led answers. Think of ChatGPT as your editor, not your ghostwriter.

In this guide I’ll hand you 24 prompts, grouped by the six real sections of the YC form, plus a 14-day sprint, a comparison table, an interview-prep section, and a People Also Ask block tuned for answer engines. Every stat is sourced from a YC page or a YC partner profile I pulled on June 11, 2026. If a claim isn’t on this page, I cut it.

Pull quote: YC invests $500,000 for 7% ($125K on a post-money safe + $375K uncapped MFN safe), then hands you a 10-minute Zoom with 2–3 partners who decide your batch the same day. Your application is the only thing standing between you and that room - so every sentence has to sound like you on your best day, not like a consultant on a generic day. (YC Deal, YC Interview Guide)

TL;DR - the 30-second version

  • Use AI to compress, not to write. Paste raw context (call notes, founder bios, metrics screenshots) and ask for tighter drafts. Keep your voice.
  • YC reads for signal, not polish. 40% of the batch is “just an idea.” YC FAQ. They want clarity, traction, and a team that can move.
  • Match the 2026 RFS themes. The Summer 2026 RFS leans hard into AI-native services, defense, biotech, and “Software for Agents.” (YC RFS).
  • The deal is fixed. $500K standard, 7% on the $125K safe, $375K MFN, US/Canada/Singapore/Cayman. (YC Deal).
  • The interview is 10 minutes. Don’t rehearse. Show progress since you applied. (YC Interview Guide).
  • Use the YC AI Stack. Students who attended a YC event from Fall 2025 onward can claim over $25,000 in AI devtool credits, including AWS, Azure, OpenAI, Anthropic, Cursor, Vapi, Firecrawl and more. (YC AI Stack blog, deals page).

What YC actually reads first (2026 reality check)

A YC application is deceptively short. It’s about 12 short-answer fields plus a few optional video links. The first 90 seconds decide whether a partner keeps reading.

Here’s what the YC interview guide says about how they read:

“We only do two things at interviews: we ask you questions, and we look at what you’ve built so far.”

That same logic applies to the application. The three things partners scan for in order:

  1. The “What is your company working on?” line. If you can’t say it in two sentences, you can’t do YC.
  2. The progress section. MRR, users, growth rate, design partners, or a shipped product. YC FAQ confirms 7% of recent batches had more than $50K MRR - but the majority of the batch is still pre-revenue.
  3. The founder bios. Domain experience, prior startups, technical depth, and the “why us, why now” thread.

The rest of the form exists to confirm what those three signals imply. ChatGPT can help you draft faster, but you have to feed it real material. Vague in = vague out.

A useful 2026 framing: YC’s deal hasn’t changed materially ($500K standard per the deal page), but the themes have. The Summer 2026 Requests for Startups opens with this line: “AI has stopped being a feature and started being the foundation.” If your company is in the 90% of AI wrappers, you have to show a moat, a distribution wedge, or a vertical-deep advantage.

Anatomy of a founder-led YC application

A founder-led application has a voice you can hear in your head. It’s not corporate. It’s not performative. It is one founder explaining a thing to another founder over coffee.

That voice has three properties:

  • Specificity over abstraction. “We sell accounting automation to 12-person dental practices in Texas” beats “we automate SMB back-office workflows.”
  • Numbers that anchor the story. “$11,400 MRR, growing 18% week over week, 34% of new signups from Product Hunt” beats “strong early traction.”
  • Honest friction. “Our biggest obstacle is onboarding time. It takes 4 hours for a new clinic. We’re shipping a one-click import in 14 days.” That line beats “we’re executing well.”

The 24 prompts below are built around that voice. I’ll group them by the actual YC form fields and then give you prompts for the 10-minute partner interview at the end.

A quick note on the “founder-led” qualifier in the title. There’s a real difference between a founder writing and a hired consultant writing. The 2015 Jessica Livingston tips post is the classic: “Don’t use buzzwords. Speak in plain language. Have a conversation with us. Don’t deliver a prewritten pitch.” The prompts below are designed so the output sounds like a founder who has done the work - not a content agency that has read 50 essays.

Section 1: Company basics & clarity prompts (Prompts 1–4)

These four prompts cover the short fields that show up first on the YC form: company name, one-liner, website, and the “describe what you do” field. Most founders over-write here. The job is the opposite: be sharp.

Prompt 1 - The “two-sentence company” test

Purpose: The YC form opens with “What is your company working on?” and partners read this in under 30 seconds. The goal is a one-sentence company description that a smart 10-year-old could repeat back, plus a “for whom” line.

The prompt:

You're a YC partner who reads 5,000 applications per batch. Read the 
context below and write three versions of a one-sentence company 
description. Each version must:
1. Be 25 words or fewer
2. Name a specific customer segment (industry, company size, geography)
3. State the painful job-to-be-done in plain English
4. Hint at the mechanism (not just "we use AI")
5. Avoid buzzwords: no "platform," "ecosystem," "leverage," "synergy," 
   "revolutionize," "disrupt," or "next-gen"

For each version, add a one-line "stress test" note: how would a 
skeptical partner push back on this claim?

CONTEXT (replace with your raw material):
- Product: [name + 1-line description]
- Customer: [ICP, e.g., "US-based accounting firms with 5-20 staff"]
- Pain: [the specific workflow that breaks today, with a real quote 
  from a user if you have one]
- Mechanism: [the technical or distribution wedge that makes you 
  different, e.g., "we read every W-2 in 90 seconds via a fine-tuned 
  document model"]
- Stage: [idea / prototype / launched / revenue / etc.]

Example output:

V1: “TaxGPT reads a stack of W-2s and 1099s in 90 seconds and drafts a Schedule C for solo US tax filers, so they stop paying $300 to a chain preparer.” Stress test: “How do you handle crypto income? What’s your CAC?”

V2: “Ledgerline is a back-office AI for 10- to 50-person dental practices in the Sun Belt, cutting month-end close from 9 days to 1.” Stress test: “Why dental and not legal? Why now, with [competitor] in market?”

V3: “Buildr is a code review agent for hardware teams shipping RTL, catching timing bugs 4x faster than human reviewers.” Stress test: “Is the wedge the model or the dataset? Who pays - IC designers or EDA vendors?”

Pro tips:

  • Run this prompt after you have a customer interview transcript. The pain will write itself.
  • If the AI spits out something you wouldn’t say out loud at a dinner party, throw it out and write the sentence yourself.
  • Save the final version in a Notion “YC source of truth” page. Every other answer should be consistent with it.

Prompt 2 - The “what is broken today” short answer

Purpose: YC partners want to know you understand the failure mode of the status quo, not just that you’re building something. The application has a short field that asks for the problem in concrete terms.

The prompt:

You're writing a 50-word problem statement for a YC application in 
the [industry] space. 

Inputs:
- Status quo today: [what people currently do, with the cost/time/anger]
- Why the status quo still exists: [legacy software, regulation, network 
  effects, switching cost, etc.]
- A real user quote: [paste one verbatim from a call, Slack, or email]
- The trigger that made this urgent in the last 12-18 months: [a new 
  model, a new regulation, a new buyer, a new entrant]

Output:
- A 50-word problem statement in plain English
- A "3rd-grade explanation" version of the same problem
- The single most painful 30-second moment in the user's week that this 
  problem creates

Example output:

Problem (50 words): US tax preparers still re-key W-2s and 1099s by hand into desktop software like Drake or ProConnect, losing 3 hours per return. 60% of preparers we interviewed said this is the #1 reason they can’t take on new clients. One said: “I’d rather do laundry than type in another W-2.”

3rd-grade version: Accountants copy numbers from PDFs into a spreadsheet, and they hate it.

The 30-second hell moment: A solo preparer with 80 returns due Monday, a 200-page stack of paper W-2s, and a coffee that’s gone cold at 11pm.

Pro tips:

  • The “3rd-grade explanation” forces the AI to drop jargon. If the AI can’t explain it to a 3rd grader, you don’t understand it.
  • The “30-second hell moment” is what you’ll actually use in the interview when Garry Tan asks “tell me about your users.”

Prompt 3 - The “why we, why now” short answer

Purpose: The YC form asks for a 50-character “what’s the insight” type field and a longer “why now” essay. Founders confuse this with a vision statement. It’s not. It’s a “what changed in the world that lets us win in the next 24 months” argument.

The prompt:

Write a 100-word "why we, why now" paragraph for our YC application.

The job of this paragraph is NOT to explain our vision. It's to 
defend three claims in order:
1. WHY US: Why are we - specifically these two founders - the team 
   to win this? Use past evidence: prior projects, domain scars, 
   unique access (a hospital, a regulator, a dataset).
2. WHY NOW: What changed in 2024-2026 that makes this winnable now 
   and not 2 years ago? Name the real shift: a model capability, a 
   regulatory change, a buyer behavior shift, a labor cost shift.
3. WHY THIS IDEA: Of the 10 ideas we could pick, why this one? 
   Be specific about what we saw (a real user moment) that made this 
   the right bet.

Avoid: "passionate," "mission-driven," "best-in-class," "industry-leading."

Inputs:
- Founder 1 background: [paste resume + 2 lines of why this problem 
  matters to them personally]
- Founder 2 background: [same]
- Macro shift: [the 2024-2026 change]
- The user moment: [a real call or DM that crystallized the idea]

Example output:

Priya spent 6 years at Epic implementing EHRs for safety-net hospitals. She saw the same workflow break in 14 of 14 deployments: nurses transcribing medication lists into structured fields by hand. Jake led the team at Hugging Face that built the first open-weight clinical NER model. Together, they ran a 4-week pilot with Denver Health and shaved nurse med-rec time from 22 minutes to 4. We didn’t pick this idea because AI is hot. We picked it because the model finally works on noisy clinical notes, and CMS just changed its documentation rules in January 2026, which is forcing every safety-net hospital to rebuild this workflow in the next 18 months.

Pro tips:

  • Three sentences, three jobs. The model can do this in one pass if you give it the three claims clearly.
  • “Why now” is the most under-written section. Most founders talk about their team and their market, but skip the timing argument. Don’t.

Prompt 4 - The “competition is obvious” line

Purpose: YC partners do not want to hear that you have “no competitors.” That’s a red flag. They want a one-line take on the existing landscape, plus a single sentence on why you win.

The prompt:

Write a 75-word "competition" section for our YC application.

Rules:
1. Name at least 2 named competitors (real companies, real products).
2. Name the legacy default (the thing people use today even though 
   it sucks).
3. State your unfair wedge in one sentence: data, distribution, speed, 
   design, cost, regulatory access, or something else.
4. Acknowledge one competitor's strength honestly. Don't trash-talk.
5. End with: "In 24 months, the market will look like [X], and we'll 
   be positioned at [Y]."

Inputs:
- Our product: [name + 1-line]
- Named competitors: [list 2-4 real companies with their product names]
- Legacy default: [Excel / Salesforce / a phone call / an agency / etc.]
- Our unfair wedge: [be honest - what's actually defensible?]
- Competitor we respect: [one]

Example output:

Today, mid-market tax preparers use Drake Tax (legacy desktop) or TaxDome (modern SaaS). The legacy default is a $200/hr bookkeeper. The legacy tools are slow because they were designed for the 2010 paper workflow. We win on (1) data - we trained on 1.4M anonymized 1040s with preparer consent, and (2) speed - our 1040 draft is 4x faster than Drake’s. TaxDome is well-designed but has no AI depth. In 24 months, we believe every preparer will run an AI co-pilot, and we’ll own the one with the best data moat.

Pro tips:

  • Naming a real competitor shows you’ve done the work. Not naming any is a yellow flag.
  • “Unfair wedge” is the most important phrase. If you can’t say it in one sentence, you don’t have one yet.

Section 2: “What do you make” prompts (Prompts 5–8)

This is where the YC application gets technical. The “describe your product” fields are where founders either win or lose a partner’s attention. Vague pitches die here.

Prompt 5 - The “show, don’t tell” product paragraph

Purpose: The YC form has a 250-character field that asks “what do you make?” Most founders answer in abstractions. This prompt forces concrete, specific output.

The prompt:

Write a 250-character "what do you make" answer for our YC application.

Rules:
- It must describe a tangible artifact the user gets, not a category.
- Use a concrete noun ("a draft Schedule C," "a 90-second call summary," 
  "a 3D model of a chipped tooth").
- Mention the input and the output. (Input = "10 PDFs." Output = 
  "a single tax return.")
- Avoid the words "platform," "solution," "tool," "suite," "experience."
- If a sentence starts with "we," rewrite it.

Inputs:
- Our product: [name]
- The artifact: [what the user literally receives]
- The input: [what the user gives us]
- The metric: [the speedup or quality gain]

Example output:

We turn a stack of paper W-2s and brokerage 1099s into a signed, e-filed 1040 in 8 minutes. The user uploads photos; we return a draft return. Median draft-to-e-file time across 4,200 returns last quarter: 7m 41s.

Pro tips:

  • Numbers are your friend. “8 minutes” beats “fast.”
  • The artifact phrase (“a signed, e-filed 1040”) is the kind of concrete noun partners can picture.

Prompt 6 - The “demo in 60 seconds” script

Purpose: YC’s interview guide says: “We sometimes will ask to see a demo of what you’re building.” This prompt writes the 60-second script you’ll use in that moment.

The prompt:

Write a 60-second demo script for a YC interview.

The interview is 10 minutes total. We have 60 seconds to show the 
product. The partner has 0 context.

The script must:
- Open with a one-sentence setup ("We make X for Y, here's what it 
  looks like.")
- Show the input (a real-looking user prompt, screenshot, or paste).
- Show the AI working in real time (with one moment of waiting - this 
  builds trust).
- Show the output (a polished, specific result).
- End with the user taking a real action (clicking, paying, sending) 
  that proves value.

Inputs:
- Product: [name + URL]
- The 1 most impressive output: [describe a real result]
- The input that's most relatable: [what would a YC partner recognize?]
- A real user reaction: [a quote]

Example output:

“We make an AI tax preparer for US individuals. Here’s a real return. [Open return #4291, a freelancer with $87K income, 4 W-2s, 2 1099-NECs, 1 K-1.] She uploaded these on Sunday night. Watch the screen - we’re going to OCR each form, normalize them, and draft a 1040 with a Schedule C. [Live wait: 14 seconds.] Here’s the draft. [Show the form on screen, line by line.] The user accepted it in 32 seconds and paid $59. Total time: 11 minutes, not 11 hours. She said, quote, ‘this is the first time I’ve filed without crying.’ [End.]”

Pro tips:

  • One moment of real waiting is more persuasive than a perfectly edited video.
  • The closing quote is what people remember. Make it real.

Prompt 7 - The technical “how it works” 100 words

Purpose: YC partners want to know you can build. The application has a 200-word “how does it work” optional field. Most founders skip it. That’s a mistake - it’s where the technical founders get to shine.

The prompt:

Write a 100-word "how it works" technical paragraph for our YC 
application.

Structure:
- Sentence 1: The data inputs (what we read, what we don't)
- Sentence 2: The core model or system (fine-tuned Llama, classical 
  ML, deterministic code, retrieval, agent loop, etc.)
- Sentence 3: The output (a structured artifact, an action, a 
  recommendation)
- Sentence 4: The novel thing (what's actually new here, in 1 sentence)
- Sentence 5: The known weakness (be honest about one limitation)

Inputs:
- Inputs: [list]
- Models: [list, with rough size if relevant]
- Output: [artifact]
- Novelty: [the one thing no one else has]
- Weakness: [the honest one]

Example output:

We ingest PDFs and XLSX exports from QuickBooks, Xero, and Sage via a fine-tuned layout-aware OCR (Surya + a custom W-2 head). A 7B open-weight model, fine-tuned on 1.4M anonymized preparer-reviewed 1040s, drafts the form line-by-line with a deterministic validation pass that flags Schedule C miscategorizations. The novel thing is the validation pass - it catches 31% of the errors the pure-LLM version misses. The known weakness: we still struggle with multi-state returns, which we plan to ship in September.

Pro tips:

  • Naming a model by size and family (“7B fine-tuned Llama”) signals real technical depth. Don’t say “we use AI.”
  • Admitting a known weakness in writing earns trust. Don’t fake perfection.

Prompt 8 - The “we use [AI Stack tool]” product slide

Purpose: The 2026 YC AI Stack (announced January 16, 2026) includes credits for tools like Firecrawl, Vapi, Browser Use, Exa, Reducto, Supabase, Unsloth AI, Deepgram, Tavus, and Gumloop. This prompt turns your product into a slide that names 4-5 of those tools, showing you ship in the current AI native stack.

The prompt:

Write a 100-word "tech stack" answer for our YC application that 
shows we're building on the 2026 AI native stack.

Tools to consider mentioning (pick 4-5 that are real in our stack):
- Firecrawl (web scraping)
- Vapi (voice agents)
- Browser Use (browser automation)
- Exa (semantic search)
- Reducto (document parsing)
- Supabase (database + auth)
- Unsloth AI (fast fine-tuning)
- Deepgram (speech-to-text)
- Tavus (video generation)
- Gumloop (workflow automation)
- Cursor (code editor)
- OpenAI, Anthropic, Grok APIs

Format:
- Sentence 1: The 1-2 things we build on top of (not from scratch).
- Sentence 2: The 1-2 things we build ourselves and why.
- Sentence 3: A line about why this combination is unique to us.

Inputs:
- The tools we actually use: [list]
- What we build in-house: [list]
- What's unique: [the one wedge]

Example output:

We use Firecrawl to ingest 14,000 state-court dockets a day, Exa for semantic precedent search, and Supabase for the case-graph store. We fine-tune our own 8B model on 280K attorney-cited cases using Unsloth. The thing we build ourselves is the citation-graph reasoning loop - the rest of the stack is best-in-class off-the-shelf. The YC AI Stack credits let us run this at zero infra cost through Demo Day.

Pro tips:

  • Naming real, current tools signals you actually ship. Hand-wavy “we use the latest LLMs” does not.
  • You don’t have to use the AI Stack credits. The point is to show fluency with the actual 2026 ecosystem.

Section 3: Progress & traction prompts (Prompts 9–12)

This is the section that decides if you’re a “real startup” or “just an idea.” YC explicitly says 40% of the batch is just an idea, so the bar is lower than you think - but specificity still wins.

Prompt 9 - The “honest metrics” snapshot

Purpose: The YC form has a 100-word “progress” field. The AI tends to inflate. This prompt forces brutal honesty, which is what partners actually trust.

The prompt:

Draft a 100-word "progress" section for our YC application.

Rules:
- Lead with the single most-impressive number, with context.
- Then 2-3 secondary numbers (users, retention, growth rate, design 
  partners, LOIs).
- Then one line of "not yet" honesty (what we haven't hit yet but 
  expect to).
- Avoid vanity metrics (downloads, signups without activation).
- No adjectives. Just numbers and one fact.

Inputs:
- The single best number: [e.g., "$11,400 MRR, up 6x in 90 days"]
- Secondary numbers: [DAU, WAU, retention %, NPS, etc.]
- Pipeline: [LOIs, pilots, design partners]
- The honest gap: [what we don't have yet]

Example output:

$11,400 MRR, 6.0x growth in 90 days, 78% net revenue retention. 42 paying customers across 7 US states, median ARPU $271, 4.1% monthly churn. 6 LOIs from regional tax franchises totaling ~$3.8K MRR if converted. We do not yet have a full-time sales hire; today all sales are founder-led. We are at 6.2% of the $184K MRR we need to be default-profitable for a Series A in 18 months.

Pro tips:

  • “Net revenue retention” is a SaaS metric that partners recognize. If you have it, lead with it.
  • Showing the gap to “default-profitable” is a sign of a founder who can plan a P&L.

Prompt 10 - The “growth loop” explanation

Purpose: YC partners want to know how growth happens, not just that it is happening. This prompt forces you to articulate the actual loop.

The prompt:

Write a 100-word "growth loop" paragraph for our YC application.

The growth loop must be a real closed loop, not a "we post on Twitter" 
list. Format: "We get users from [X] → they create [Y] → that drives 
[Z] new users."

Rules:
- Name the trigger event (a new regulation, a viral artifact, a 
  partnership, an API change).
- Name the artifact the user creates that pulls in the next user.
- Name the constraint that limits this loop's growth.
- One line on what's still hand-holding (paid ads, founder outreach, 
  SEO, etc.) and what we'd need to do to remove the hand-holding.

Inputs:
- Trigger event: [the change that creates demand]
- User artifact: [what they create that pulls in the next user]
- Constraint: [what's actually rate-limiting us]
- Hand-holding: [what's not yet a real loop]

Example output:

Each new 1040 we file is a one-page summary the user shares with their accountant and 3 friends (“I did my taxes in 8 minutes”). That summary contains a discreet “prepared with [us]” footer. Each share drives ~0.4 new signups. The rate limiter is the volume of filed returns, not paid acquisition. The hand-holding today: our summary template is still manually designed by a contractor. In Q3 we ship a self-serve summary editor, which should 4x our share-driven signups.

Pro tips:

  • “X → Y → Z” forces linearity. If your growth is messy, the AI will help you find the dominant loop.
  • “Rate limiter” is the magic word. If you don’t know what’s rate-limiting you, partners will.

Prompt 11 - The “design partner” or “pilot” paragraph

Purpose: Many early-stage companies win YC interviews because they have 3-5 design partners. This prompt writes the paragraph that makes those partners feel real.

The prompt:

Write a 100-word "design partners" paragraph for our YC application.

Rules:
- Name 2-3 specific design partners (real companies, real humans, 
  real roles).
- For each, state what they gave us (weekly call, data, co-design 
  time, letters of intent, paid pilot).
- One line on what we built specifically because of them.
- One line on what we still owe them (the honest "we don't have 
  feature X yet").

Inputs:
- Partner 1: [company, contact name + role, what they did]
- Partner 2: [same]
- Partner 3: [same]
- Feature we built for them: [the one we're proud of]
- Feature we still owe: [the one we don't have]

Example output:

Denver Health (Sarah Chen, CMIO) gave us 6 weekly calls and 4,200 de-identified discharge summaries. We built a structured med-rec list specifically for their cardiology unit. UCHealth (David Park, Director of Pharmacy) signed an LOI to pilot in 3 hospitals for $48K/year, contingent on us shipping the drug-interaction module in October - which we will.

Pro tips:

  • Naming a real human with a real role is more convincing than “we work with 3 health systems.”
  • The “feature we still owe” is what makes this paragraph trustworthy. YC partners read 5,000 of these; they can smell fluff.

Prompt 12 - The “what would you build with $500K” answer

Purpose: YC has a short “if we fund you, what will you do” field. This prompt pre-empts the most common wrong answer (“we’ll hire engineers”) and forces a milestone-based plan.

The prompt:

Write a 100-word "what we'll do with the $500K" paragraph for our 
YC application.

Rules:
- The $500K is split: $125K for 7% on a post-money safe + $375K on an 
  uncapped MFN safe (per the YC deal page).
- Do NOT mention valuations or dilution math.
- List 3-4 concrete milestones, each with a target date.
- Milestone 1: a product milestone (ship feature X).
- Milestone 2: a distribution milestone (close channel Y, hit Z users).
- Milestone 3: a revenue or capital milestone ($X MRR or close round Y).
- Milestone 4: a hiring milestone (one key first hire).
- One honest line about what could go wrong.

Inputs:
- Product milestone: [e.g., "ship voice agent by Aug 15"]
- Distribution milestone: [e.g., "300 paying practices by Dec 1"]
- Revenue/capital milestone: [e.g., "$25K MRR by Demo Day"]
- Hiring milestone: [e.g., "first AE hire by week 6"]
- The risk: [e.g., "we depend on one integration partner"]

Example output:

With the $500K, we will: (1) ship voice agent v2 by Aug 15; (2) close 3 regional tax franchises as channel partners by Oct 1; (3) hit $25K MRR by Demo Day; (4) make our first AE hire by week 6 of the batch. The honest risk: we depend on Intuit’s API for one workflow. If Intuit changes access terms in Q4, our Q1 plan slips by 60 days.

Pro tips:

  • “Demo Day” is a YC-specific term. If you reference it, you sound like you’re already in the family.
  • One honest risk beats five confident claims.

Section 4: Founder story & team prompts (Prompts 13–17)

YC partners read thousands of founder bios. The winning ones have a specific texture: a person who was unusually exposed to a problem, unusually skilled to solve it, and unusually stubborn to keep at it. These prompts help you write that.

Prompt 13 - The “scars + skills” founder bio

Purpose: The YC form has a 150-character bio field per founder. The job is to compress a resume into a line that says “I have been marinating in this exact problem for years.”

The prompt:

Write a 150-character founder bio for our YC application.

Rules:
- Lead with the specific prior experience that made you the right 
  person to start THIS company (not "passionate about X").
- Mention one credential or signal (school, company, role, paper) 
  that a YC partner would recognize.
- End with the one line of personal stake ("my mom runs a 12-person 
  clinic," "I spent 5 years on the inside of this industry").

Inputs:
- Founder 1: [name, role, prior companies, education]
- The personal stake: [the line that ties you to the problem]
- The one signal: [a credential that YC partners would respect]

Example output:

Priya (CEO): ex-Epic implementation lead for safety-net hospitals, deployed EHRs at 14 health systems, BS Comp Sci UT Austin. Her mom is a nurse manager at a 9-bed rural ER - the target buyer.

Pro tips:

  • “Ex-Epic” + a number is a stronger line than “5 years in healthcare.”
  • The personal stake should be true and specific. “Mom is a nurse manager at a 9-bed rural ER” beats “from a healthcare family.”

Prompt 14 - The “co-founder fit” paragraph

Purpose: YC funds teams more than ideas. The 100-word “co-founder relationship” field is where you prove the team works. This prompt forces the AI to articulate the actual division of labor and the reason the team sticks.

The prompt:

Write a 100-word "co-founder relationship" paragraph for our YC 
application.

Rules:
- State how you met (one line, with a real detail).
- State the explicit division of labor (CEO does X, CTO does Y, with 
  no overlap).
- State one decision you made together that proves you actually 
  work as a team (a real example from the last 90 days).
- End with one line about why the team is hard to replicate 
  (a skill combination, a shared history, a unique trust signal).

Inputs:
- How we met: [one line]
- Division of labor: [who does what]
- A real decision: [from the last 90 days]
- The irreplaceable thing: [the trust signal]

Example output:

Jake and Priya met at a hackathon in 2019, lost touch, then reconnected in 2024 when Jake’s clinical NER model was the only one that handled Priya’s worst hospital’s discharge notes. Priya runs sales, customer, and clinical research. Jake runs product, model, and infra. Two months ago we cut our entire go-to-market in Texas because the data showed our highest-LTV segment was in Florida. We made that pivot in 36 hours. We’ve known each other long enough to disagree on architecture and still ship the same week.

Pro tips:

  • A real recent decision beats any platitude about teamwork.
  • “Disagree on architecture and still ship the same week” is a specific texture YC partners love.

Prompt 15 - The “long-term commitment” line

Purpose: YC has a single-line “are you all in?” question. This prompt writes a 50-word answer that doesn’t sound like a cover letter.

The prompt:

Write a 50-word "long-term commitment" answer for our YC application.

Rules:
- One line on what each founder gave up to do this.
- One line on what each founder has explicitly committed to (rent 
  paid by, savings runway, family situation).
- One line on what would make us quit (the honest "we wouldn't quit 
  unless X").

Inputs:
- What we gave up: [salary, equity, status]
- Runway: [months of personal runway, partners covered]
- The "would quit" trigger: [be honest]

Example output:

Priya left her $185K Epic role with 14 months of savings. Jake turned down a staff eng offer at a public company. We have 9 months of personal runway if we make $0 revenue. We will not quit unless one of us has a health issue or we run out of money for 60+ days.

Pro tips:

  • Naming the salary you left is a strong signal. Don’t be coy.
  • “9 months of personal runway” is what partners actually want to read.

Prompt 16 - The “weakness” honesty answer

Purpose: The YC form has a single-line “biggest weakness” field. Founders who skip it or sugarcoat it lose points. This prompt writes a real weakness with a real mitigation.

The prompt:

Write a 50-word "biggest weakness" answer for our YC application.

Rules:
- Pick a real, specific weakness (not "we work too hard").
- Show you understand why it's a weakness.
- Show the concrete thing you're doing in the next 60 days to fix it.

Inputs:
- The weakness: [be honest]
- Why it's a weakness: [be specific]
- The fix: [the action with a date]

Example output:

Neither of us has sold into a hospital system before. We’re 9 months into this company and have closed 3 of 14 target accounts, well behind plan. We’re hiring a fractional VP Sales with 12 years at Epic in week 2 of the batch.

Pro tips:

  • “Neither of us has sold into a hospital system before” is more credible than “we need to improve our enterprise sales motion.”
  • Naming a specific hire and a specific week is the kind of concrete action YC partners reward.

Prompt 17 - The “domain credentials” line

Purpose: The YC form has an optional “anything else we should know” field. Use it for one or two lines of domain credentials, prior art, or unfair access. This prompt writes that.

The prompt:

Write a 75-word "domain credentials" optional field for our YC 
application.

Rules:
- 1-2 lines of any prior research, papers, patents, or published work 
  in the field.
- 1-2 lines of any unusual data, partnership, or distribution access 
  ("we have a data sharing agreement with X," "we have an LOI from Y").
- 1 line of any social proof that doesn't sound braggy (e.g., a real 
  user quote, a press mention, an award).

Inputs:
- Prior art: [papers, patents, etc.]
- Unfair access: [data, partnerships, LOIs]
- Social proof: [real quotes, press, awards]

Example output:

Priya co-authored the 2023 JAMIA paper on EHR discharge summary error rates. Jake’s clinical NER model is the open-weight baseline on 3 of 4 MedNLI leaderboards. We have an LOI from Denver Health for 4,200 de-identified records, and a 30-minute call with the CMIO of HCA on the books for next Tuesday.

Pro tips:

  • “Co-authored the 2023 JAMIA paper” is more credible than “expert in clinical workflows.”
  • A real upcoming call (“on the books for next Tuesday”) is a great live signal.

Section 5: Vision & “why now” prompts (Prompts 18–21)

These prompts help you write the optional long-form fields: vision, “anything else,” and the “why YC” question. They also help you align with the 2026 RFS themes if you happen to be in those spaces.

Prompt 18 - The 2026 RFS alignment paragraph

Purpose: The Summer 2026 Requests for Startups page has 15 named RFS topics, from “AI for Low-Pesticide Agriculture” (Garry Tan) to “SaaS Challengers” (Jared Friedman) to “Software for Agents” (Aaron Epstein). If your company maps to one, this prompt writes a short paragraph that signals that alignment.

The prompt:

Write a 75-word "alignment with YC's 2026 RFS" paragraph for our YC 
application.

Pick the single closest RFS theme from the Summer 2026 list:
- AI for Low-Pesticide Agriculture (Garry Tan)
- AI-Native Discovery Engines (Jon Xu)
- AI-Native Service Companies (Gustaf Alströmer)
- AI Personalized Medicine (Ankit Gupta)
- Company Brain (Tom Blomfield)
- Counter-Swarm Defense (Tyler Bosmeny)
- Dynamic Software Interfaces (Ankit Gupta)
- Electronics in Space (Philip Johnston)
- Hardware Supply Chain (Nicolas Dessaigne)
- Industrial Capabilities in Space (Adi Oltean)
- Inference Chips for Agent Workflows (Diana Hu)
- SaaS Challengers (Jared Friedman)
- Software for Agents (Aaron Epstein)
- Startups Selling to Huge Companies (Harshita Arora + Brad Flora)
- Supply Chain 2.0 for Semiconductors (Diana Hu)
- The AI Operating System for Companies (Diana Hu)

Rules:
- Quote one line from the RFS post that we directly align with.
- Name the YC partner who wrote it.
- State our specific wedge inside that theme.

Inputs:
- Our product: [name + 1-line]
- The RFS theme: [which one]
- The specific line that resonates: [quote]
- Our wedge: [the unfair thing]

Example output:

We’re building a Company Brain - what Tom Blomfield describes as “a system that pulls knowledge out of fragmented sources, structures it, and turns it into an executable skills file for AI.” Our wedge: we’re starting with one vertical (US tax), where we already have 1.4M preparer-reviewed 1040s as training data. We believe the Company Brain is going to win category by category, not all at once.

Pro tips:

  • Citing the partner by name shows you read the RFS page, not just the headlines.
  • “We believe the X is going to win category by category” is the kind of opinionated framing partners respect.

Prompt 19 - The “why YC, not a competitor” answer

Purpose: YC has a short “why YC” field. The honest answer is usually “YC is the only accelerator where we already have 3 friends who went through it.” This prompt writes that version.

The prompt:

Write a 75-word "why YC, not another accelerator" answer for our YC 
application.

Rules:
- Name 2-3 specific YC resources you actually plan to use (the YC 
  AI Stack credits, a specific YC partner's office hours, Bookface, 
  Demo Day intros, a specific YC community).
- Name 2-3 specific YC alumni you've already talked to (real names, 
  real companies).
- One line on why a different accelerator would be a worse fit.

Inputs:
- Specific YC resources: [the AI Stack, partner office hours, etc.]
- YC alumni contacts: [names, companies]
- The competitor accelerator: [a real one you considered]

Example output:

We plan to use the YC AI Stack credits (announced Jan 16, 2026) for our model fine-tuning, book office hours with Diana Hu on inference chips, and tap the Bookface hardware-supply-chain channel for sourcing. We have 4 close YC alumni (Figma, Linear, Mercury, Vapi) we’ve already shared product roadmaps with. We considered a16z’s speedrun; it doesn’t have the operator density we need.

Pro tips:

  • Naming a specific partner and a specific topic (“office hours with Diana Hu on inference chips”) is more credible than “we want YC’s network.”
  • Naming 4 specific YC alumni companies is a strong signal you’ve done the work.

Prompt 20 - The “10-year vision” answer

Purpose: YC’s optional “where do you see this in 10 years” field is where founders either under-write or over-write. The right answer is one paragraph, opinionated, with one specific milestone that scares you a little.

The prompt:

Write a 100-word "10-year vision" answer for our YC application.

Rules:
- One sentence on the company in year 10.
- One sentence on the market in year 10.
- One sentence on what we'll have to do in years 1-3 to make year 10 
  possible.
- One honest sentence on the one bet that has to be true for this 
  to work.

Inputs:
- Year 10 company: [the one-line vision]
- Year 10 market: [the market's shape]
- Years 1-3 actions: [the 1-3 big moves]
- The bet: [the one thing that has to be true]

Example output:

In 10 years, Ledgerline is the default back-office AI for the 60,000 US dental practices that don’t have a CFO. The market will look like: every mid-market services business has a vertical AI co-pilot. To get there, in years 1-3 we need to ship the one-click data import, expand from dental into 2 adjacent specialties, and build the in-app practice-network effects. The bet: vertical AI wins over horizontal AI in regulated services.

Pro tips:

  • A specific market number (“60,000 US dental practices”) is more credible than “the SMB market.”
  • A specific bet (“vertical AI wins over horizontal AI in regulated services”) signals a founder who has opinions.

Prompt 21 - The “what scares you” honesty line

Purpose: The most under-used line in the YC form is the optional “anything else.” This prompt writes a one-line confession that builds trust. Use it in the interview, not the form, if you prefer.

The prompt:

Write a 50-word "what scares me about this company" honesty line for 
our YC application (optional field).

Rules:
- One real, specific fear.
- One thing we're already doing to address it.
- No "I'm worried about competition" boilerplate.

Inputs:
- The fear: [be honest]
- The action: [the thing you're already doing]

Example output:

I’m scared that the dental market is too small for our burn rate. We’re addressing it by capping our seed round at $3M and shipping the legal-vertical MVP in month 9 - we will not raise a Series A without $1.2M MRR.

Pro tips:

  • Naming a dollar figure (“$3M seed, $1.2M MRR Series A bar”) is a real founder answer.
  • “I’m scared” is a stronger opening than “we recognize the risk that.”

Section 6: Interview prep & practice prompts (Prompts 22–24)

The YC interview is 10 minutes, over Zoom, with 2-3 partners, with your application open in front of them. They will not let you present. They will not let you monologue. They will ask you questions. Most of them are predictable. These three prompts prep you for the actual call.

Prompt 22 - The “10 questions YC will ask” practice set

Purpose: Run this prompt the night before the interview. It generates the 10 most likely questions, the bad answers, and the better answers, tuned to your company.

The prompt:

You're a YC partner who has read 17,000 applications and done 3,100 
interviews (loosely modeled on Michael Seibel's bio). I'm interviewing 
for [company] in 3 days. Generate the 10 most likely questions you'll 
ask, in priority order.

For each question:
- The question, in 1 sentence.
- The "bad" answer (a 1-line version that would lose the room).
- The "better" answer (a 3-line version in founder voice, with one 
  concrete number).
- The "follow-up trap" question (the 2nd question they'd ask if you 
  answered well).

Context about us:
- Company: [name, 1-line]
- Stage: [MRR, users, growth]
- Founders: [1-line each]
- Market: [the wedge]

Output format: a numbered list of 10 questions with the 3 sub-points 
under each.

Example output (truncated):

1. What is your company working on? Bad: “We’re a B2B SaaS for dental practices.” Better: “Ledgerline is a back-office AI for 10- to 50-person dental practices in the Sun Belt. We cut month-end close from 9 days to 1. We’re at $11,400 MRR, 6.0x in 90 days.” Follow-up: “Why dental and not legal?”

2. How do you acquire users? Bad: “We do content marketing and SEO.” Better: “Each filed return is shared by the user with their accountant and 3 friends; that share drives 0.4 new signups. Our rate limiter is filed-return volume, not paid acquisition.” Follow-up: “What’s your CAC by channel?”

(… 8 more)

Pro tips:

  • Practice answering out loud, not in your head. The interview is verbal.
  • Time yourself. If your first answer takes 90 seconds, cut it in half.

Prompt 23 - The “demo the product” script (for the moment they say “show me”)

Purpose: YC says in their interview guide: “We sometimes will ask to see a demo of what you’re building. By demo, we mean a working version of whatever you’ve built or plan to build.” This prompt writes the 90-second demo and the 3 follow-up questions you’ll get.

The prompt:

Write a 90-second live demo script for my YC interview.

I will screenshare. The partner is on a Mac with Chrome. I have 90 
seconds, and one moment of real waiting is fine.

The script must include:
- 1 sentence of context.
- The exact input I will paste or click.
- The exact 1-2 clicks or commands I will run.
- The exact 5-10 second wait (with a verbal "this takes about 12 
  seconds" cue).
- The exact output I will highlight.
- The exact user action that proves value (a click, a payment, a 
  send).

After the script, list the 3 most likely follow-up questions the 
partner will ask, with 1-line answers to each.

Inputs:
- The most-impressive real result from the last 7 days: [describe]
- The most relatable input: [the one a partner recognizes]
- The user reaction: [a real quote]

Example output:

“We make an AI tax preparer for US individuals. Here’s a real 1040. [Open case #4291.] The user uploaded these on Sunday. I’m going to run the draft. [Click ‘Draft 1040.’] Takes about 12 seconds - that’s our OCR + fine-tuned Llama + the validation pass. [12-second wait.] Here’s the draft. [Show line 1, line 2, line 7, Schedule C.] The user accepted in 32 seconds, paid $59, and e-filed. Total time: 11 minutes. She said, quote, ‘this is the first time I’ve filed without crying.’”

Likely follow-ups:

  1. “What happens if the input is a crypto 1099?” - “We currently route those to a human preparer. We’re shipping crypto support in September.”
  2. “What does this cost to run per return?” - “$0.31 in OpenAI + inference.”
  3. “Why would H&R Block not just clone this?” - “Their distribution is 10,000 retail stores. We’re targeting the 33M filers who don’t use a preparer at all.”

Pro tips:

  • Rehearse the demo 5 times. The interview is too short for fumbles.
  • The follow-up answers should be memorized, not improvised.

Prompt 24 - The “they just said no, what now” reflection prompt

Purpose: The YC FAQ says: “In a typical YC batch, about half the companies applied multiple times before being accepted.” This prompt writes the post-rejection email and the 14-day improvement plan.

The prompt:

I just got rejected from YC after an interview. Write me:

1. A 100-word thank-you + reflection email to the partner who 
   interviewed me.
2. A 14-day "ship one more thing" plan with 3 concrete product or 
   traction moves.
3. A 100-word reapplication paragraph for the next batch that 
   references the specific feedback they gave me.
4. A "feedback interpretation" - re-state the most likely reason they 
   said no, in 1 sentence, and what to do about it.

Inputs:
- The interview: [what was actually asked + what you answered]
- The feedback (if any): [paste the email]
- The one product thing we're closest to shipping: [name it]
- The one traction thing we could realistically hit in 14 days: [name it]

Example output:

Email: “Sarah - thank you for the time and the candid feedback. The ‘you need to ship the multi-state engine before reapplying’ note is exactly what we needed to hear. We’re going to use the next 90 days to do exactly that, and we will reapply for [next batch]. If it’s useful, we’ll send you a Loom when the multi-state engine is in production. - Priya & Jake”

14-day plan:

  • Day 1-4: Close 2 design partner LOIs in [state] (specific target: Acme Tax in Austin, owner Sarah Lee, on the books for Tuesday).
  • Day 5-10: Ship the multi-state engine behind a flag for 3 states.
  • Day 11-14: Re-run last quarter’s returns through the new engine; record the time savings as a before/after.

Reapplication paragraph: “Last cycle, we interviewed for [batch] and the feedback was ‘ship the multi-state engine before reapplying.’ We did. We also closed [N] new practices in 90 days and hit [$X] MRR. The product now handles [Y] states in [Z] minutes. The team is the same. The company is meaningfully further.”

Feedback interpretation: “The product wasn’t fully solved for the obvious next state. The fix is shipping the obvious thing and showing the result.”

Pro tips:

  • Send the thank-you email within 24 hours. It’s a relationship.
  • The “feedback interpretation” line forces you to actually hear what they said, not what you wanted to hear.

Comparison table: 24 prompts mapped to the YC form

I built the table below by mapping every prompt to the actual YC application field, the model to use, and what good output looks like. I use ChatGPT (GPT-4o or GPT-5) for most of these, Claude Sonnet 4.5 for the “voice” prompts (4, 11, 16, 17, 21, 23) where tone matters most, and Grok-3 for the “competitive” prompts (4, 18) where YC partner opinions leak in.

#PromptYC form field it answersBest modelOutput lengthVoice
1Two-sentence company”What is your company working on?”GPT-4o / 53 × 25 wordsSharp, plain
2What is broken todayProblem short answerGPT-4o / 550 wordsEmpathetic
3Why we, why nowWhy now essayClaude Sonnet 4.5100 wordsFounder-led
4Competition is obviousCompetitors short fieldClaude Sonnet 4.575 wordsHonest
5Show, don’t tell”What do you make?”GPT-4o / 5250 charsSpecific
6Demo in 60 secondsOptional demo linkGPT-4o / 560-second scriptVerbal
7How it works (technical)Optional technical fieldClaude Sonnet 4.5100 wordsBuilder
8AI Stack product slideOptional “tech stack” mentionGPT-4o / 5100 wordsNative
9Honest metrics snapshotProgress fieldGPT-4o / 5100 wordsNumbers-first
10Growth loop explanationOptional growth fieldClaude Sonnet 4.5100 wordsAnalytical
11Design partner paragraphProgress / LOIsClaude Sonnet 4.5100 wordsTrustworthy
12What we’d build with $500KFunding-use fieldGPT-4o / 5100 wordsMilestone
13Scars + skills bioFounder bio (per founder)Claude Sonnet 4.5150 charsSpecific
14Co-founder fitCo-founder fieldClaude Sonnet 4.5100 wordsRelational
15Long-term commitment”Are you all in?”GPT-4o / 550 wordsConcrete
16Weakness honestyOptional weaknessClaude Sonnet 4.550 wordsVulnerable
17Domain credentialsOptional “anything else”Claude Sonnet 4.575 wordsCredible
182026 RFS alignmentOptional “why now”GPT-4o / 575 wordsAligned
19Why YCWhy YC fieldGPT-4o / 575 wordsSpecific
2010-year visionOptional long-formClaude Sonnet 4.5100 wordsOpinionated
21What scares youOptional honestyClaude Sonnet 4.550 wordsBrave
2210 likely questionsInterview prepGPT-4o / 510 × 3 sub-pointsRehearsed
2390-second demo scriptInterview live demoClaude Sonnet 4.590 sec + 3 follow-upsVerbal
24Post-rejection planReapplicationClaude Sonnet 4.54 partsResilient

People Also Ask: 10 questions, 1-line answers

These are the questions a smart founder would type into Google before applying. I’m writing them as answer-engine-friendly Q&A blocks so they get cited in AI Overviews, Perplexity, and Google SGE.

1. What does a YC application actually look like in 2026? A 12-field online form on apply.ycombinator.com with short-answer essays (50–250 chars), 2 founder bios, a 50-word “what’s broken today,” a 100-word “what’s your insight,” a “progress” field, and 2 optional long-form fields. Most founders finish it in 2–4 hours.

2. How much does YC invest, and on what terms? YC’s standard deal is $500,000: $125,000 for 7% on a post-money safe + $375,000 on an uncapped MFN safe. Investment is made the day you’re accepted, with no milestones.

3. How long is the YC interview? 10 minutes, over Zoom, with 2-3 YC partners, and your application open in front of them. Decisions typically come the same day.

4. What is the YC AI Stack? A January 2026 program where YC partnered with 20+ companies to offer over $25,000 in free AI devtool credits to students who attend a YC university event. Includes AWS ($10K), Azure ($10K), OpenAI ($2,500), Cursor ($720), Anthropic ($500), and credits for Firecrawl, Vapi, Browser Use, Exa, Reducto, Supabase, Unsloth, Deepgram, Tavus, and Gumloop.

5. What percentage of YC companies are pre-revenue? About 40% of the companies funded in each batch are “just an idea”. 7% of recent batches had $50K+ MRR at acceptance. Both pre-revenue and post-revenue companies get funded.

6. What is the YC acceptance rate? YC does not publish an official number, but Garry Tan has said he has read over 6,000 applications and Michael Seibel has read over 17,000 in his career, with batches ranging from ~200 to ~250 companies per cycle. The widely-cited estimate is in the 1–2% range. (Garry Tan bio, Michael Seibel bio)

7. Can solo founders apply? Yes. YC regularly accepts solo founders, though they recommend a co-founder. If you don’t have one, YC’s Co-Founder Matching is a free platform with 100,000+ matches made. (YC FAQ)

8. How many times should I apply to YC? The FAQ says “in a typical YC batch, about half the companies applied multiple times before being accepted.” Reapplying with progress is a strong signal.

9. What are the 2026 YC RFS themes? The Summer 2026 RFS lists 15 themes including AI for Low-Pesticide Agriculture, AI-Native Discovery Engines, AI-Native Service Companies, AI Personalized Medicine, Company Brain, Counter-Swarm Defense, Dynamic Software Interfaces, Electronics in Space, Hardware Supply Chain, Inference Chips for Agent Workflows, SaaS Challengers, Software for Agents, Startups Selling to Huge Companies, Supply Chain 2.0 for Semiconductors, and The AI Operating System for Companies.

10. Can I use AI to write my YC application? Yes - and most top applicants do, in some form. The best practice is to use AI to compress your own raw material (call transcripts, founder bios, metrics dashboards) into tight drafts, then edit for founder voice. The 24 prompts in this guide are designed for exactly that workflow.

A 14-day YC application sprint

Here’s the calendar I’d run if I were applying to a YC batch today. Each day has 90 minutes of work. Adjust the dates to your batch deadline - the Fall 2026 deadline is July 27, 2026 at 8pm PT, and on-time applications get a decision by August 28.

  • Day 1 - Raw material dump (3 hours). Export your last 30 days of customer call notes from Fathom or Read AI, your last 90 days of Linear tickets, your Notion strategy doc, and your most recent investor update. Save to a single folder.
  • Day 2 - Voice baseline. Write your “what is your company” answer in 25 words by hand, no AI. This is your north star.
  • Day 3 - Run Prompts 1, 2, 3. Use Claude for the longer ones. Compare to your baseline.
  • Day 4 - Run Prompts 4, 5, 7. Get the competitive, product, and technical answers.
  • Day 5 - Run Prompts 9, 10, 11. Get the metrics, growth, and design partner answers. Pull real numbers from your dashboard.
  • Day 6 - Run Prompts 13, 14, 15. Get the founder bios, co-founder paragraph, and commitment line.
  • Day 7 - Run Prompts 6, 8. Rehearse the demo script and the AI Stack product slide. Record a Loom.
  • Day 8 - Run Prompts 18, 19, 20. Read the full Summer 2026 RFS page before this. Quote 1 partner.
  • Day 9 - Run Prompts 22, 23. Practice the 10-question set and the 90-second demo out loud, on Zoom, with one friend.
  • Day 10 - Run Prompt 24 (pre-mortem). Have a friend pretend to be a YC partner and reject you. Write the reapplication paragraph in advance.
  • Day 11 - Founder-led edit pass. Cut every word that doesn’t sound like you on your best day. Read each answer out loud.
  • Day 12 - External review. Send the full application to 2 YC alumni you trust. Get 1 round of edits.
  • Day 13 - Polish + submit checklist. Verify all dates, links, founder LinkedIn URLs, and that the demo link doesn’t require a login.
  • Day 14 - Submit. Submit 48–72 hours before the deadline. Do not submit at 7:59pm PT.

A small but important tip from the YC interview guide: “If you really want to improve your chances of getting into YC, the best way to get an edge is to work hard and have your startup improve between the time that you applied and the time that you interview.” So between Day 14 and the interview (which is typically 2–4 weeks later for on-time applicants), keep shipping.

Common mistakes to avoid (and what to do instead)

These are the patterns I see over and over in weak YC applications. None of them are deal-breakers alone, but they stack.

  • Mistake 1: Buzzwords. “Leverage,” “synergy,” “platform,” “best-in-class,” “disrupt,” “next-gen.” Replace each with the specific thing it stands for.
  • Mistake 2: Talking about your TAM first. Founders lead with “$50B market opportunity.” Partners care about your wedge in a $50M wedge. Lead with the wedge.
  • Mistake 3: Over-rotating on AI. “We’re an AI company” is no longer a sentence. Say what the AI does, for whom, and how it works.
  • Mistake 4: Burying the number. If you have $10K MRR, say it in the first sentence of the progress field. If you have 0 MRR, say “we have 8 design partners, 4 LOIs, and 1 paid pilot” instead.
  • Mistake 5: Skipping the technical “how it works” field. A surprising number of strong teams skip it. Don’t.
  • Mistake 6: Treating the interview like a pitch. Garry Tan’s interview video and the YC interview guide both say: “Have a conversation with us. Don’t deliver a prewritten pitch.”
  • Mistake 7: Not mentioning progress since applying. If you apply on Monday and your product changes by Friday, submit an update through the edit page. This is the single highest-leverage move most founders miss.
  • Mistake 8: Misreading the deal. The deal is $500K total - $125K for 7% and $375K on an uncapped MFN. Don’t talk about it like it’s $500K for 7%. Read the deal page once.
  • Mistake 9: Ignoring the YC AI Stack. If you’re a student and you attended a YC event from Fall 2025, you have over $25,000 in credits waiting. That’s free runway.
  • Mistake 10: Forgetting that YC is selective but not random. About half of every batch applied multiple times. Rejection is a data point, not a verdict.

Final word

A YC application is a 12-field document that decides whether you get a 10-minute Zoom with a partner who’s read 17,000 of these.

The right way to use AI here is to compress. The wrong way is to outsource. Every prompt in this guide is built to make your founder voice louder, not to replace it. The partners are not looking for beautiful prose. They are looking for proof you can do the work, with a team that can take the hits, on a problem that’s worth the next 10 years of your life.

Run the 14-day sprint. Paste your real material into the prompts. Read the output out loud. Cut everything that doesn’t sound like you. Then submit 72 hours before the deadline and keep shipping.

If you get the interview, watch Dalton Caldwell’s interview video, re-read the YC interview guide, and practice Prompts 22 and 23 until you can deliver them cold at 2am.

Good luck. We’re rooting for you.