Evidence auditor for founders
What does your evidence actually justify?
Foundry is an evidence auditor for founders. It tells you what your current evidence actually justifies — and what it does not — before you commit more time or money.
Customer enthusiasm is not the same as customer behaviour. Foundry grades what people actually did, how independent the evidence is, what would falsify the assumption, and what remains unresolved. Then it applies the same deterministic policy every time.
One assumption, one review, one stored result. For founders and small product teams.
How far has the evidence climbed?
PAIDWhat they paid
Money changed hands, or a contract commits them to pay.
COMMITTEDWhat they committed
They put real time, people or budget behind it.
DIDWhat they did
They acted: built a workaround, shared real data or access, changed how they work.
SAIDWhat they said
Opinions, stated intent, and a problem described in their own words.
What Foundry inspects
You bring one clearly stated assumption and the evidence you have for it. For each piece of evidence, Foundry asks:
- What did they do? An opinion, a stated intent, a confirmed problem, a workaround, shared data, a changed workflow, a committed resource, or a payment.
- Who is the source? Independent of you, related to you, or controlled by you.
- Is it still true? Current, or stale.
- What would prove you wrong? And did you write that down before or after you went looking?
What you get back
One of four outcomes, with the reasons, the conditions still unresolved, the limits of your evidence, and the single next thing to go and find out. “Not yet” is the most common answer, and a useful one.
Iterate
Not yet. There is signal here, but not enough to build on, and not enough to stop.
Abstain
A responsible evaluation isn't possible yet.
Build
Current evidence justifies a bounded next step.
Kill
This assumption, as tested, should stop.
What Foundry refuses to do
- It does not score your idea or rank it out of 100.
- It does not predict whether your startup will succeed.
- It does not let a model choose the outcome. A fixed, versioned policy does, and the same input always gives the same result.
- It does not interview synthetic customers or invent evidence you do not have.
- It does not always answer. When the question or the evidence is not usable, it abstains and says why.
How to start
A Decision Review costs €199, once. You pay, describe one assumption and its evidence in a structured form, and receive a result you can come back to. After a first review you can keep that assumption under review with Foundry Continuous at €99 a month.