Find the assumption to test before you build.
Give your AI your idea, evidence and next building commitment. Leave with one assumption to test and a small experiment you can run.
Use this promptHow can AI help you test a business idea before you build it?
A week spent building feels productive. It helps to know what could make that work worth doing before you start.
You might need to learn whether someone will pay for the offer, whether you can reach the buyer or whether you can deliver what you promised. Each needs a different test.
Use this prompt when you have a next piece in mind but little evidence behind it. Your AI helps you choose the unknown that matters most and prepare a small way to learn about it.
Before you paste it
Give your AI this context:
- Your idea, the person it could help and the result you would offer.
- What you are about to build or commit to, including the time or money involved.
- What you have actually observed. Keep compliments, stated interest and completed purchases separate.
- Who you can reach for a test and how much time or money you can use. Remove private details from customer notes.
PASTE INTO YOUR AI
Help me choose a test before I build
Help me decide what to test before my next building commitment. My idea and intended buyer: [who I could help and with what] What I am about to build or commit to: [include time and cost] What I have observed: [real behavior or results; “nothing yet” is fine] What I am assuming: [my guesses] People I can reach and how: [known access or an untested route] My limits for a test: [time, budget and what I can deliver now] Use what I have given you. If a missing answer could change which assumption or test to choose, ask me one question at a time. Use each answer to decide what you still need. When you have enough, prepare the test; don't repeat settled questions. Work backward from the commitment. What would need to be true for it to be worth making? Separate observations, choices and untested assumptions. Choose one assumption whose failure would most change my next move. Explain why it matters more than the other unknowns. Keep it specific enough to test. Propose the smallest credible test within my limits. Match it to the assumption: a compliment is not a purchase, and interest from friends may say little about the intended buyer. If access is unverified, establish how I could reach relevant people first. If a test cannot fit my limits, explain the gap and suggest a smaller commitment or a way to defer it. Don't invent participants, responses or demand. Give me a short test note: - The assumption and what we currently know. - What I will do, with whom and how I will reach them. - A time and cost cap. - The behavior or result I will observe. - A proposed working rule for continuing, changing or stopping, set before the test. Explain why it fits this decision; do not treat a small sample as statistical proof. - What would make the result inconclusive and what to check then. Prepare the material I need to run it, such as an invitation, conversation question or manual demonstration outline. Keep the offer honest about what exists and can be delivered. Finish with one action for me to take. Prepare only; do not send messages, take payments or spend money.
What you should leave with
A test note you can use: one assumption, the people and action involved, a time and cost cap, the result to observe and a working rule for the next decision. You should also have the invitation, question or demonstration outline needed to start.
Illustrative example
You bring: You want to run a Saturday drawing workshop and plan to spend eight evenings building a booking site. Two friends like the idea. You can contact five local people who already attend art classes, but nobody has agreed to pay for your workshop.
A useful result: Test whether those people will book this particular class. Prepare an honest invitation with the date, $25 price and what a beginner will make. Cap preparation and outreach at one hour. A possible rule: two confirmed paid bookings justify trying one class; no bookings means learn why people declined before building the site. Vague interest leaves the payment question open. These numbers are a chosen pilot rule, not proof of a wider market.
Then do this
Check that the proposed test measures the assumption and fits your limits. Choose the working rule before you start. Run the first action, record what people actually did and bring that result back to your AI before making the building commitment.
AI can help prepare the test and interpret your notes. Evidence comes from the test itself. A weak response could reflect the offer, timing or access; it does not automatically settle whether the whole idea can work.