The job: You have an idea for a product or service, but you don’t know whether anyone would pay for it, and building it properly would take months.
Effort
Low to medium: a weekend to write the offer and questions, then the real work of talking to potential customers.
Cost to start
Very low. AI is used for thinking and drafting; the test itself is conversations and, if you choose, a simple sign-up page.
No income promises. We don’t promise or imply any level of earnings. Any of these ideas can earn nothing at all, and some cost money to start. What you make, if anything, depends on your skills, time, prices, market and luck. We don’t publish income figures because we can’t back them up. This is general information, not financial, tax or legal advice. If you start selling, check GOV.UK on whether you need to tell HMRC.
What AI actually does here
- Argues against your idea, listing the reasons it might fail so you can test the biggest one first.
- Drafts neutral interview questions that don’t lead people to the answer you want.
- Summarises your interview notes into patterns, which you check against the notes themselves.
Tools to look at
- General assistants (ChatGPT, Claude, Gemini)
- A simple website or form builder if you want a sign-up page
Named tools are examples, not endorsements. We have no paid relationships with them at the moment; see our disclosure.
What you need first
- Real conversations with potential customers. AI can’t tell you whether people will pay; only people can.
- Honest wording on any sign-up page: don’t promise results, and make clear what is free and what isn’t (the ASA’s guidance on money-making ads covers both).
- If you collect email addresses, a clear privacy notice and consent, in line with UK GDPR.
How to tell if it’s working
- Number of real conversations held, and how many people described the problem without prompting.
- Commitments, not compliments: pre-orders, deposits, introductions or time given.
- Money and hours spent before deciding to continue, change or stop.
What the evidence says
No statistics are cited. This is a method for learning cheaply, not a prediction of success.
Ready-to-use prompt
Replace the [BRACKETS]. Only paste data you’re allowed to share with the tool you’re using.
View promptHere is my business idea:
Here is my business idea:
"""[describe the idea, who it's for and what you'd charge]"""
Act as a sceptical but fair adviser.
1. List the five assumptions this idea depends on, ranked by how risky they are.
2. For the riskiest one, design the cheapest test I could run in two weeks without building the product.
3. Write 8 neutral customer interview questions about their current problem (not about my idea). No leading questions.
4. Tell me what result would make you advise me to stop.
Be direct. Use British English. Don't estimate revenue or market size.Sources
- GOV.UK, "Working for yourself" Government guidance
- ASA/CAP, "Money making courses: stick to the facts" (April 2026) Regulator guidance
- ICO, Guidance on AI and data protection Regulator guidance
Last updated . Spotted something out of date? Email support@aiforthisjob.com. How we research and check playbooks.