AI REHEARSAL FOR FIRST-TIME FOUNDERS

Your idea deserves better questions.

Your canvas is a list of guesses. Unassumed turns them into assumptions, rehearses the conversations that could break them, then sends you out with questions a real customer will actually answer.

NO IDEA SCORE. NO “GREAT IDEA!” MACHINE. BUILT ON LEAN PRINCIPLES AND THE MOM TEST. ONE EMAIL WHEN YOUR SEAT OPENS.

QUESTION REVIEW DRAFT · INTERVIEW 01

YOU WROTE

“Would you use an app that automatically organises all your receipts?”

LEADING HYPOTHETICAL SOLUTION-BIASED

QUESTION QUALITY

41/100

3 PROBLEMS
IN ONE SENTENCE

you have asked this one ↓

THE QUESTION THAT RUINS EVERYTHING

Compliments feel useful. Only past behaviour is.

BAD QUESTION Would you use this?

Asks a stranger to predict their own future behaviour, and hands them the polite answer for free. You will hear yes. You will learn nothing.

BETTER QUESTION Tell me about the last time this happened.

Asks for a memory instead of a forecast. You get a date, a workaround, a cost and a person: the raw material of a real business.

An attractive canvas is not a validated business. ↙

One leading question, taken apart and rebuilt

LIVE TEARDOWN DRAFT · INTERVIEW 01

YOU WROTE

Would you use an app that automatically organises all your receipts?

ASK INSTEAD

Tell me about the last time you had to find a receipt.

  • HYPOTHETICALAsks them to forecast their own behaviour.
  • SOLUTION-BIASEDYour answer is already inside the question.
  • LEADINGOnly one polite reply is left available.

QUESTION QUALITY

78/100

NOW ASK A REAL PERSON

Scribble. Guess. Rehearse. Evidence.

Every stage earns the next one. The destination is not a product. It is a canvas that changed because of something a real person did.

  1. SCRIBBLE

    Get it out of your head

    Paste the note, the voice memo, the photo of the napkin. Unassumed drafts a first-pass canvas you can argue with.

  2. ASSUMPTION

    Name the risky guess

    Every block becomes a claim with a status. Yellow means we believe it and cannot yet prove it.

  3. REHEARSAL

    Practise the pushback

    Run the interview against a synthetic customer who objects, digresses and refuses to be led.

  4. EVIDENCE

    Leave the building

    Log what a real person actually said, attach the source, and watch which beliefs stop surviving.

AND THE CANVAS CHANGES. THEN YOU DO IT AGAIN.

Rehearse here. Learn out there.

A synthetic person is not your market. They are a sparring partner: a cheap place to find your leading questions, your vague claims and your missing evidence before you spend a real conversation on them.

And we score you, not your idea. A language model cannot know whether your company will work. It can absolutely tell when you are fishing for a compliment.

  • ASSUMPTION we think this, we cannot prove it
  • CORRECTION bias, leading question, contradiction
  • EVIDENCE a real person did or said this

INTERVIEW SCORECARD

REHEARSAL 03 · SYNTHETIC FREELANCE DESIGNER

62/100
Past-behaviour questions76
Specific examples elicited58
Leading language42
Solution leakage31
MARGIN NOTE

You pitched the solution in question 2. They agreed with you for the rest of the call.

REWRITE 3 QUESTIONS →

AI IS REHEARSAL. CUSTOMERS ARE EVIDENCE.

No synthetic persona ever says validated.

Every other tool in this category will happily tell you your idea scored 78 out of 100. We think that is the exact false confidence you came here to escape.

So an assumption in Unassumed never carries a green tick. It carries a provenance: what has actually happened to this belief, who said it, and when.

SIMULATION · PRACTICE ONLY
Busy freelance designers lose invoices every month.
STILL A GUESS

Rehearsed with 3 synthetic personas
0 real interviews attached

✓ FIELD EVIDENCE
They don’t lose invoices. They lose weeks chasing late payment.
SUPPORTED

4 customer interviews · 3 described it unprompted
1 forwarded their chase email as proof

VIEW THE 4 SOURCES →

OUR OWN ASSUMPTIONS, IN PUBLIC

We haven’t got testimonials yet. So here are our open guesses.

It would take ten minutes to invent six glowing quotes from founders who do not exist. Instead: the beliefs this company is currently running on, and their status. We update them as the evidence arrives.

TESTING

First-time founders avoid customer interviews because they are frightened, not lazy.

11 conversations so far · 9 agree
UNKNOWN

They understand “The Mom Test” without us explaining it first.

No evidence attached yet
CONTRADICTED

Founders want an investor pitch simulator first.

7 of 9 asked for customer practice instead

For first-time founders. Not for startup theatre.

  • You have a rough idea

    Even if it currently lives in a note called “startup thing FINAL v4”. Rough is the correct starting state.

  • You want honest resistance

    You would rather hear the weak argument this week than after six months of building the wrong thing.

  • You want to be told yes

    No synthetic persona can prove demand exists. If you want a certificate, this is the wrong tool.

Less than your Thursday coffee-and-croissant meeting. And considerably less agreeable.

FIRST COHORT PRICING HELD FOR TWELVE MONTHS · CANCEL WHENEVER

FOUNDER PLAN [$6.99] /mo [5] REHEARSALS · UNLIMITED ASSUMPTIONS

THE FIELD TEST STARTS SOON

Bring a scribble. Leave with a question worth asking.

Join the first cohort for early access, the founder field exercises, and our build notes, including the assumptions we get wrong.

NO SPAM. NO DRIP SEQUENCE. ONE EMAIL WHEN YOUR SEAT OPENS.