FounderFold: the layer on top
AI models: they do the analysis, and they’re built in
FounderFold is the layer on top of AI models for startup idea validation: it finds and generates ideas, keeps score of your five half-started ones and, once one works, generates the strategy to improve it.
Early access is by invitation.
FounderFold for founders is a layer on top of AI models for people working alone or in a very small team: it finds and generates business ideas, keeps the decision on each one with its reason and, once the business is running, generates the strategy to optimize it. The AI is already built in. You decide.
FounderFold: the layer on top
Tell it what you’re after and what you’d never do. The Explorer searches the web with that and shows each fact with its source. If nothing fits, it says “Nothing found” instead of padding the list.
Your business comes in as a running project, with its starting figures and its goal. FounderFold’s CEO, its AI agent, gives you the next step each day and a review every Monday.
AI models: they do the analysis, and they’re built in
Startup idea validation is testing whether a business idea deserves your time and money before you build it: checking the problem, the market and the willingness to pay with evidence and small experiments, and agreeing in advance which result would make you stop.
“I’ve got five half-started ideas and I can’t drop any.”
They all stay alive because none has a number that would kill it.
“The site’s live. Now what?”
Building something proves nothing anymore. It feels like progress, and nobody asks for evidence.
“Why did I drop that one? It’s in a chat somewhere.”
Six months later you’re back on the same idea, researching it from scratch.
“Let’s give it one more month.”
With no number agreed up front, every weak month earns another chance.
| AI models | AI models + FounderFold | |
|---|---|---|
| Researching an idea | Multi-step research across many sources, with citations, and scheduled tasks that repeat it if you set them up. | The Explorer shows the funnel of each round: queries, candidates, those ruled out by your reasons, and finalists. Every fact has its source, or “Not checked”. |
| Remembering what was said | Memory of your chats, plus search across past ones. | One record per idea, with its number, its status and its history: what happened, who decided and why. |
| The idea you drop | You find it by searching your chats, if that’s where you talked it through. | It stays in the portfolio as “Ruled out”, with its reason. Records are never deleted, and the Explorer only brings it back if a new source appears. |
| Five half-started ideas | They summarize where each one stands, if you tell them. | Each with its status, and the Summary counts how many await your decision. Your criteria live in the portfolio (“How it’s scored” and “Reasons to discard”). |
| Follow-up | Scheduled tasks with your connectors, which you describe and maintain. | Every Monday, the review by FounderFold’s CEO, its AI agent, of your records with nothing to set up: 3 to 5 lines and 3 actions at most. |
| Documents and spreadsheets | Documents, spreadsheets, presentations and work on your files. | When you add the record, you note what comes out of them under “What you already know”, with its link. |
Indicative comparison with features offered by some AI models and general assistants as of October 30, 2026; they vary by provider and plan, and change quickly. The column with FounderFold only lists what already works in the panel.
Early access is by invitation.
Think of the last idea you dropped. Where is it written down why?
The Explorer searches open sources for problems that fit what you want to build, so your first idea competes with others before it gets your savings.
Each idea is scored on the same rubric, and every score keeps its source and date. Anything that could not be verified says so.
A landing page, a price test, a paid pilot: each experiment has a numeric success criterion and a deadline written before it starts.
Kill criteria are set before each stage and measured from the figures you record, so the decision to stop is not left to a bad week.
Today lines up what needs doing, one step at a time, and the CEO prepares each close for you to confirm.
When an idea or a project ends, its retrospective and its lessons stay in your portfolio for the next attempt.
Illustrative data — Fictional company, ideas and figures, shown for illustration only.
A fictional founder wants to leave her job to build software for independent physiotherapy clinics. She gives herself twelve weeks.
The Explorer brings back 14 opportunities in clinic software, each with its sources. She adds two ideas of her own.
All sixteen are scored with her rubric. Three stand out on market and channel; the scheduling idea she started with comes fourth.
One experiment on the strongest idea: at least 5 of 20 clinics must agree to a paid pilot within 45 days.
Six clinics agree. She chooses the idea and turns it into a project at its Validation stage.
The MVP stage opens with milestones for the first version, the price and the first ten paying clinics, and a kill criterion on retention.
Twelve weeks later she has a project with paid pilots, a written reason for every idea she dropped and a number that will tell her when to stop.
Validate your business idea before you build it.
Early access is by invitation.
Tell FounderFold the kind of business, market and size you are after, and what you never want to do. It proposes your rubric and your reasons to discard.
Run the Explorer, add the ideas you already have or import them from a spreadsheet.
Study the best ideas in depth and design one experiment for the strongest, with its criterion written first.
Choose the idea that met its criterion, turn it into a project and follow Today. A report you can share with a partner or an investor shows where it stands.
Weighted value per year rose to €3.1M; two decisions need the board.
Decisions requested
Cutoff Sep 30, 2026 · Portfolio New lines 2027 · Page 3 of 24
Illustrative data — Fictional company, ideas and figures, shown for illustration only.
Experiment E-0131 on idea I-0412: the hypothesis, the metric “paid pilots signed”, the success criterion “at least 5 of 20 clinics” and a 45-day deadline.
The hard part is rarely building. It is choosing what to build and knowing when to stop.
The first idea gets all the attention, so nobody compares it with the five others that might have been better.
Weeks of reading about the market produce opinions, not a decision, and none of it is written down where it can be checked later.
Without a number agreed in advance, every disappointing month turns into “let’s give it one more”.
A reference lifecycle for software built by one person. Each stage has its milestones, its gate and its kill criteria, and you can rename or add stages.
Projects by stage
Lifecycle: new business line in an existing company · Halden Mutual
Illustrative data — Fictional company, ideas and figures, shown for illustration only.Run the Explorer on a thesis and see which problems keep coming up, with their sources, before you commit to one.
Score the ideas you already have on the same rubric as the ones you found, and see which ones hold up.
Study an idea in depth: a score per dimension, market and competition included, each with its evidence and date.
Turn the chosen idea into a project with stages, milestones and the experiments that test price and channel.
Today shows the next step, and the figures you record show whether the plan is working.
Illustrative data — Fictional company, ideas and figures, shown for illustration only.
Today panel docked beside project P-0107: step 1 of 4, labeled “Overdue experiment”, with the buttons “Yes, I have the result”, “Let the CEO prepare it” and “I need more time”.
You commit time and money to the idea with the most evidence behind it, not to the first one you had.
To a partner, an investor or yourself in a year’s time: every score, discard and gate keeps its reason.
Before you start, you know which result would end the project, so a bad week does not decide for you.
Your ideas, your decisions and the reasons behind them live in one record instead of notes, spreadsheets and memory.
Score it on a rubric with evidence for each dimension, then test what the evidence cannot settle with an experiment that has a numeric success criterion and a deadline. If it meets the criterion, choose it and turn it into a project.
No. A project can follow the lifecycle for software, services, commerce or physical products, among others, and you can add stages of your own.
Yes. A solo founder works with the same portfolio, rubric, experiments and CEO as a team, and can invite a partner or a mentor as a reader.
The CEO studies the idea and tells you what the evidence says, with sources. It can propose a decision, marked as AI-generated, but the decision is yours.
No. The AI comes with your FounderFold membership, with a monthly AI limit in plain sight, and there’s nothing to connect. If you already use a model, keep using it for your own work.
If what you need is to think and write, you have it. FounderFold earns its place when you’re carrying several ideas and want to know which continues, which stops and why, without relying on what you remember.
A validator gives you an opinion on one idea at one moment. FounderFold keeps all your ideas against the same yardstick, with the reason for each discard, and stays with you when the chosen idea becomes a project.
You start with a short form: “Add an idea” uses no AI and asks for nothing you don’t know. The method shows up when you need it, one step at a time, in “Today”.
They stay in your portfolio with the reason they were discarded, so you never test the same idea twice without knowing it.
For founders, FounderFold is in early access by invitation, and pricing is not published. You can join the waiting list for the Founder membership.
FounderFold helps you decide what to build and when to stop, with the evidence on record. The agents research and score; you decide.
A founder validates one business and builds it. A venture studio does it again and again with a shared team, comparing many opportunities at once.
Early access is by invitation.
Tell us who you are and what you want to solve. If it is a fit, we create your organization and send the owner an invitation.Leave your details and we’ll email you when the Founder membership opens. Nothing to pay now.Leave your details if you are interested in deal analysis and we will email you about it. Signing up does not give you access to any service or commit you to anything.
Check your email.Check your email.Check your email.
We have sent you an email to confirm your address. Until you confirm it, your request does not count and we do not handle it; if you do not confirm it within 7 days, we delete it. The link lasts 72 hours: if it does not arrive within a few minutes, check your spam folder or send the form again.We have sent you an email to confirm your address. Until you confirm it, your request does not count and we do not handle it; if you do not confirm it within 7 days, we delete it. The link lasts 72 hours: if it does not arrive within a few minutes, check your spam folder or send the form again. Once you confirm, we will only email you when the Founder membership opens, unless you ticked the news box or the design partner program box.We have sent you an email to confirm your address. Until you confirm it, your request does not count and we do not handle it; if you do not confirm it within 7 days, we delete it. The link lasts 72 hours: if it does not arrive within a few minutes, check your spam folder or send the form again. Once you confirm, we will only email you about deal analysis, and about your case only if you described it, unless you ticked the news box.