The pre-seed round is the strangest fundraise you'll ever do.
You have no revenue. Possibly no product. Maybe just a working prototype and a belief — deeply held, supported by real evidence, but still a belief — that you've identified a problem worth solving and that you're the person to solve it.
And you're asking someone to give you a meaningful amount of money based primarily on that.
I've been there. I've sat across from investors who asked me to explain, in two sentences, why I would succeed where every comparable attempt had failed. I've left meetings knowing I didn't have the answer they needed yet. And I've eventually raised a pre-seed round from people who believed in what I was building before anyone else could see it clearly.
Here's what I've learned about how that process actually works for AI founders in 2026.
What Pre-Seed Actually Means in 2026
Pre-seed is the first institutional money. It comes before seed, before Series A, often before you have a fully built product or paying customers.
For AI startups, the numbers look different than the general market:
- Typical raise: $500K to $2M (AI companies raise significantly more at pre-seed than non-AI, because compute costs are real before revenue exists)
- Check sizes from individual angels: $25K to $300K
- Pre-money valuation: $3M to $8M median (AI skews lower at pre-seed because there's less to anchor on)
- Investment instrument: SAFE notes in 90%+ of pre-seed deals — no interest, no maturity date, converts at your next priced round
- Timeline: 60 to 120 days from first outreach to money in the bank
The hardest thing to accept about pre-seed: you are raising on the lowest possible information. The investor is making a decision based largely on you and the quality of your thinking — not metrics, not revenue, not a proven product. This is simultaneously intimidating and freeing.
What Pre-Seed Investors Actually Look For
Team First, Almost Everything Else Second
I mean this literally. At pre-seed, before you have meaningful traction or revenue, investors are primarily underwriting the team.
What that means specifically for AI:
Technical depth: At least one person on the founding team who understands how AI systems actually work — not just how to call an API, but what happens when a model hallucinates, how to evaluate AI performance on specific tasks, what the architecture decisions are that will matter in 18 months. This person doesn't need to have worked at OpenAI. But they need to be credible when a technical investor asks hard questions.
Domain expertise: You should understand the problem you're solving at a level that most people in that industry don't reach. If you're building AI for legal discovery, you should understand what actually makes discovery painful, who in the organization controls the budget for this tool, and what "good" looks like in a way that a generalist could never articulate. This knowledge is what distinguishes a product that solves the real problem from one that solves the demo.
Prior evidence of execution: Have you shipped something before? Sold something? Recruited people to work with you on hard problems? Pre-seed investors aren't just betting on your vision — they're betting that you can convert vision into action under pressure. Evidence from earlier in your career matters.
If you're a solo founder, the question you'll face in every meeting is: "Who else is going to build this with you?" It's not a disqualifying question, but you need a credible answer.
The Hypothesis Needs to Be Testable
At pre-seed, you don't need to have proven your thesis. You need a thesis that is specific enough to be tested.
"AI will transform healthcare" is not a thesis. "AI-powered clinical documentation for small primary care practices reduces note-writing time from 20 minutes to 4 minutes per patient, and we can sell this to independent practices at $500/month because they can't afford the enterprise solutions" is a thesis. It's specific, it's falsifiable, and it tells an investor exactly what you're going to do with the money.
The best pre-seed theses I've seen have three components:
- A specific user with a specific pain (not "all doctors," not "the healthcare industry")
- A specific way AI makes that pain better than everything that currently exists
- A reason why this company, at this time, has a real shot at capturing that opportunity
Some Evidence of Demand — Not Proof, But Evidence
Most pre-seed investors understand that you haven't proven product-market fit. They're not expecting it. But they want some indication that someone other than you thinks this problem is real.
What "evidence of demand" looks like at pre-seed:
- Customer discovery interviews with specific quotes from potential users describing the pain in their own words
- A waitlist of people who gave you their email without being asked to
- A prototype (even rough) that real users have interacted with
- Letters of intent or expressions of interest from potential customers
- Prior industry experience that gives you pre-existing relationships with potential buyers
If you have even one paying customer — even at a deep discount, even at a price that doesn't work long-term — that signal is worth more than anything in your deck. Pre-seed investors know the difference between someone who has talked to potential users and someone who has actually put a product in front of them and watched them respond.
Who Writes Pre-Seed Checks Into AI Startups
Y Combinator
YC remains the gold standard for pre-seed. Their current terms: $500,000 for approximately 7% equity. Acceptance rate: around 1% from 15,000+ applications per cohort.
In the Spring 2025 batch, over 50% of accepted companies were building with AI. YC is not just capital — the batch network, the alumni community, and the Demo Day investor access are genuinely valuable in ways that compound over time.
Apply even if you're not sure you'll get in. The application forces you to articulate your thesis with precision, and the interview process (if you get there) is one of the better stress tests for your narrative.
Antler
Antler has become the most active early-stage investor globally by deal volume — 128 investments in 2025. Their model is unusual: they invest before co-founders are even paired. If you're a strong individual without a co-founder, Antler runs structured programs to help you find the right founding partner, then invests in the company you form together.
They write £80K to £200K for approximately 10% equity, depending on the cohort structure. Their London cohort invested £1.7M in 14 AI startups in Spring 2025. They also run cohorts in New York, Berlin, Singapore, and other major startup cities.
For first-time founders who haven't yet assembled a team, Antler is one of the few legitimate paths to pre-seed capital without a warm investor network.
Angels: The Fastest Pre-Seed Money
For many AI founders, the first check comes from an individual angel investor — not a fund. Angels move faster, require less formal process, and can write checks from $25K to $300K without a committee decision.
Prominent angels active in AI right now:
Naval Ravikant — AngelList founder, prolific AI investor, known for backing contrarian technical theses early. Writes smaller checks but brings network credibility.
Elad Gil — Backed Stripe, Airbnb, Square, Gusto at early stages. Wrote the High Growth Handbook. Increasingly focused on AI infrastructure and AI-native companies. Checks typically $100K to $500K.
Fei-Fei Li — Stanford AI professor, co-director of the Stanford Human-Centered AI Institute. Founded ImageNet, former Google Cloud Chief AI Scientist. Invests personally in AI companies where domain expertise and academic credibility intersect. Extremely selective.
Nat Friedman / Daniel Gross — Former GitHub CEO and Apple AI lead respectively. Run the AI Grant program, which gives $250K in compute credits plus cash to AI researchers doing foundational work. More research-oriented but worth understanding if your work is at that frontier.
The angel network that matters most is often more local and specific than the famous names. Investors who have built companies in your specific domain — a healthcare entrepreneur who sold a startup and now angels in digital health AI, for example — can be the most valuable pre-seed capital because they bring operational knowledge alongside the check.
Micro VCs and Pre-Seed Funds
Beyond the large-name accelerators, there is an active ecosystem of smaller funds that specifically write pre-seed checks:
Precursor Ventures: Invests up to $500K at pre-seed, 30-40 investments per year, known for founder-friendliness and quick decisions.
Hustle Fund: Pre-seed focused, writes $25K to $200K checks, particularly active with first-time founders. The "hustle" thesis — betting on velocity of learning and iteration rather than credentials — resonates for AI founders moving fast.
Pioneer Fund: Applications-based, equity-free at the contest stage, then small checks for winners. Accessible to founders without warm networks.
AI2 Incubator (Allen Institute for AI): Incubates AI companies from scratch, provides funding, compute, and technical resources. Primarily for founders coming from research backgrounds.
Do You Need a Working Product Before You Raise Pre-Seed?
Honest answer: it depends on what you're raising on.
If you have exceptional team credentials — you led ML research at a top lab, you've shipped AI products that reached significant scale before, or you're a domain expert with 15 years in the industry you're disrupting — you can raise pre-seed without a working product. Investors are buying the thesis and the capability.
If you're a first-time founder without that track record, a working prototype matters enormously. It's not that you need product-market fit. You need to demonstrate that you can build. A rough, imperfect AI prototype that does one thing and does it somewhat well is worth a hundred slides about what you're going to build.
The bar for "working prototype" in AI specifically: the AI should actually be running inference on real inputs, not a mock-up with canned outputs. It doesn't need to be fast, cheap, or scalable. It needs to demonstrate that the core technical challenge — making the AI do the thing — is solved.
I've watched founders lose pre-seed conversations because they couldn't answer "can I try the product right now?" with a yes. Get something live before you start pitching.
The Practical Pre-Seed Process
Build Relationships Before You Need Them
The worst time to start building investor relationships is when you're running out of runway. Pre-seed fundraising is dramatically easier when investors know who you are before you show up with an ask.
Six months before you intend to raise, start sharing what you're working on publicly. Write about what you're learning. Share your thinking on the problem space you're working in. Follow the investors you eventually want to approach and engage genuinely with their content. Attend relevant events.
When you eventually reach out, the best case is that they already know your name.
Use Warm Introductions Where You Can
Even at pre-seed, warm introductions dramatically increase your response rate. Other founders — especially YC alumni or people who've gone through accelerators with the investors you're targeting — are often willing to make introductions if they know your work.
Ask specifically. "I'm raising my pre-seed and I'd love an intro to [partner] at [fund]. Would you be willing to make that connection?" is a more successful ask than hoping someone will think of you.
Prepare for the Questions, Not Just the Deck
The deck is a prop. The conversation is the pitch.
Pre-seed investors are evaluating whether you think clearly, respond well under pressure, and genuinely understand the problem you're working on at a level of depth that would survive contact with reality. Rehearse answers to hard questions:
- "What does this look like in 3 years if it works?"
- "What have you learned from talking to potential customers that surprised you?"
- "Why hasn't someone already built this?"
- "What's the single biggest risk to this company?"
- "Why are you the person who should build this?"
The best answers are specific. "Because we have a unique distribution advantage through [specific partnership]" beats "because we're the best team." "Because the regulatory moat takes 2 years to build and we've started" beats "because we'll outexecute."
Move Fast Once You Get a Yes
Pre-seed rounds can close fast when momentum is real. If an angel offers you a term sheet, don't let weeks pass before closing. Momentum in fundraising is fragile — an investor who said yes in a meeting can talk themselves out of it in the time you spend deliberating.
Get a startup lawyer (not a generalist). Have your SAFE documents ready before you get to a yes. Have your cap table prepared and clean. The administrative friction of closing a round can be eliminated before the first meeting.
Once You Have Capital, Move Fast
Pre-seed funding buys you time. That time is almost always less than you think it is. The math is simple and merciless: most pre-seed rounds give you 12 to 18 months of runway to prove the thesis that justified the investment.
The founders who use that time well are the ones who get to product in users' hands as quickly as possible. The founders who struggle are the ones who spend it debating architecture decisions, hiring slowly, and iterating on features before anyone has validated the core value.
For non-technical founders who've raised pre-seed capital and need to move fast, this is exactly the moment where the choice of how to build matters. A big agency will take $150K and six months before they've shipped anything. A freelancer from a marketplace might cost $15K but comes with real execution risk and no accountability structure. The sweet spot — a professional product studio with AI-native expertise, a structured process, and accountability for delivery — typically costs $30K to $60K and gets you from funded to live product in 8 to 12 weeks.
That's the window that matters. FeatherFlow works with pre-seed AI founders at exactly this stage, helping turn early capital into a real, working product — and that product is often what makes the seed raise possible.
The dream here is real: a functioning AI product, in front of real users, within your pre-seed runway, with enough signal to raise the next round from a position of strength rather than desperation.
That's the goal. Use your first capital to get there.
Frequently Asked Questions
What's the difference between pre-seed and seed for AI startups?
Pre-seed is the first institutional check — typically before you have revenue, often before you have a fully built product. Seed is the next round, where investors expect a live product, early evidence of user engagement, and ideally some paying customers. The median AI seed raise is $4.6M at a $17.9M pre-money valuation. The median pre-seed is closer to $500K to $2M at a $3M to $8M valuation. The investor profile, the diligence process, and what you need to demonstrate are all different.
Should I raise on a SAFE or a priced equity round at pre-seed?
Almost always a SAFE. They're faster, simpler, and standard at this stage — 90% of pre-seed deals use them. You don't need to spend six weeks negotiating a priced equity round before you've proven anything. The key SAFE terms: the valuation cap (lower is better for investors; higher is better for you) and the discount rate (typically 15-20%). Work with a startup lawyer to understand what these mean for your cap table at your next round.
Can a non-technical founder raise AI pre-seed?
Yes, but with more difficulty and usually at lower valuations. Non-technical founders need to compensate with strong domain expertise, a credible technical advisor or team member who can speak to the AI architecture, and evidence that they've already made real product progress. If you haven't yet found a technical co-founder, consider Antler's co-founder matching program or EF (Entrepreneur First) before raising externally.
How much equity should I give up at pre-seed?
Target less than 15% to accelerators (YC is 7%; most others are 5-10%). For angels, expect 10-20% collectively across your pre-seed round. Above 25% total dilution at pre-seed leaves you with a difficult cap table for subsequent rounds and makes it harder to attract the team and investors you'll need later. The goal is to keep enough of the company that you're genuinely motivated by the outcome and that future investors see a cap table that can support a real company.
What should I do if pre-seed investors pass?
First, understand the reason. Investors who pass and tell you why are giving you information — take it seriously. "We don't invest at this stage" is different from "we don't believe the market is real" which is different from "we need to see more traction." If the pattern across five or more meetings is "not enough traction," build more traction before your next round of conversations. If the pattern is "we need a technical co-founder," find one before the next meeting. Use the passes as signal, not just rejection.
The First Check Changes Everything
The first check into your company is different from every check that comes after it.
It's the moment someone else decided, with their own money, that your thesis is worth betting on. That changes how you think about the idea, how you talk about it in public, and how you recruit.
The path to that check is simpler than it sounds: a specific thesis about a real problem, early evidence that someone other than you believes the problem is real, a team that can credibly execute, and a process that puts you in front of investors who might actually care.
Start building the relationships before you need them. Get something live before you pitch. Know exactly why your AI moat survives competition.
And when you have the first yes — move fast. The market rewards the founders who turn early capital into early proof.