AI Idea Validator.

Twelve questions that separate AI products that ship from AI demos that die. Run before you commit a quarter of engineering.

0 / 12 ANSWERED

Q1Problem clarity — is there a named workflow this replaces?

Yes, I can name the user and the daily workflowRoughly — I know the use case, not the workflowIt's a category, not a workflowStill hand-wavy

Q2How often does the user hit this pain?

Daily — it interrupts real workWeeklyMonthlyRarely / hypothetical

Q3Willingness to pay — proven by existing spend?

Users already pay competitors $100+/mo for adjacent toolsUsers pay something for the problem todayNo paid market yet, but high painNo evidence of spend on this

Q4Can you reach the data the model needs?

Yes — I own it or it's in a queryable APIMostly — some scraping or integration workScattered across systems and silosBehind a moat I don't control

Q5How much hallucination can the user tolerate?

High — output is a draft, user reviewsMedium — wrong outputs are annoying, not dangerousLow — errors damage trust quicklyZero — regulated, safety-critical

Q6Can you tell when the output is wrong without manual review?

Yes — clean automated eval signalPartially — need some samplingOnly with manual reviewNo reliable way to grade outputs

Q7Build path — what does v1 look like?

Thin wrapper on a hosted API + prompt engAPI + light fine-tuning on examplesFine-tune or distill a small modelPretrain or train-from-scratch

Q8Time-to-first-real-output (not a demo)?

Hours — wired tomorrowDaysWeeksMonths

Q9Distribution — do you have a wedge?

Yes — audience, channel, or integrationPartial — one warm channel to startCold start, but a known beachheadNo distribution thesis yet

Q10Defensibility — what stops a copy?

Proprietary data, workflow lock-in, or network effectsBrand + ICP focus + speedJust first-mover on a generic promptNothing — anyone with the same API can ship it

Q11Founder fit — do you (or your team) have the technical depth?

Yes — shipped AI features in production beforeStrong engineering, new to LLMsWill need a technical co-founder or contractorNo technical depth available

Q12Runway — how many months of cash can you spend on a slow ramp?

18+ months12 – 18 months6 – 12 monthsUnder 6 months

RESULT · YOUR VERDICT0/12

VERDICT

—

Answer all 12 questions to see your verdict.

WHAT THIS MEANS

Answer each question on the left and the verdict will update.

NOTES · WEEKLY