How to Test a Business Idea with AI Before You Spend Anything

How to test a business idea with AI before you spend anything

Updated 27 October 2026 · 6 steps

To test a business idea with AI, write it as one sentence, have AI find who would pay and who already sells it, check the numbers at a price people already accept, then talk to five real buyers before you build anything. AI does the research in an afternoon; the go, change or stop decision stays yours.

Most new businesses do not fail because the founder could not build the thing. They fail because nobody wanted to pay for it. The good news is that finding out costs almost nothing now: an AI co-founder can do the desk research in an afternoon, and five honest conversations do the rest. Here is the order that works, with Corey doing the legwork.

1. Write the idea as one sentence

Say who it is for, what it does for them and roughly what it costs: “A dog-walking service for busy owners in Reading, from £15 a walk.” If you cannot fit it in one sentence, the idea is not ready to test yet - and Corey will help you narrow it down first.

2. Find the people who would pay

Corey describes your most likely buyer, where they already spend time, and what they do about the problem today. That last part matters most. A real buyer is already paying something to solve the problem - money, time or hassle. If nobody is doing anything about it, it may not be a problem worth paying for.

3. See who already sells it

Corey lists the competitors near you or online, what they charge and what their customers complain about in reviews. Competition is good news: it proves people pay. The complaints are your opening - the thing you can do better.

4. Check the numbers

Corey works out your cost per sale and the margin at the price competitors already charge, then how many customers a month you need to pay yourself. If the numbers only work at a price nobody charges, change the idea before you spend anything on it.

5. Talk to five real buyers

This is the step people skip, and it is the one that matters. Corey drafts five short, honest messages to people who fit your buyer - asking about the problem, not pitching the product. You send them yourself. Five real answers beat any report, including one written by an AI.

6. Decide go, change or stop

Corey puts it together on one page - the buyer, the competitors, the numbers and what the five people said - and recommends go, change or stop, with the reason. The decision is yours. If it is go, the next step is choosing a structure and registering.

Test your idea with Corey

Corey is a free AI co-founder for solopreneurs and startups, inside Claude. Get Corey free, then use the ready-made Test my idea prompt: fill in your idea, who it is for and where, and Corey runs the steps above with you. You need a paid plan on your AI app, such as Claude Pro.

Questions, answered

It can tell you whether the evidence supports it - who would pay, who already sells it, and whether the numbers work at a realistic price. It cannot know your appetite for risk or what you are willing to give up, so treat its verdict as a recommendation, not a decision.

The desk research - buyers, competitors and numbers - takes an afternoon with an AI co-founder doing the legwork. Talking to five real buyers takes a few days, and it is the step that matters most.

No. Test first, register once you know people will pay. The registration step is quick in most countries and there is no point paying for it before you know the idea works.

A plain chat answers from what it remembers and forgets your idea between chats. Corey runs a set method - buyer, competitors, numbers, real conversations - remembers your business, and checks the rules where you are. The research steps are the same; the method and the memory are what change.