Home › Guides

How to validate a business idea with ChatGPT or Claude in 24 hours

By George Batchelor · Updated October 31, 2026

Before you spend a month building something, spend a day trying to break the idea. AI is an excellent sparring partner for this, because it's fast, tireless and willing to be brutally honest if you ask. But it can't tell you whether real people will pay. That part is on you. This sprint combines both.

Important: AI can make up facts, numbers and competitors. Treat everything below as hypotheses to check, not answers.

Hour 0–1: Write the one-sentence idea

I help [specific person] get [specific result] without [specific pain], by [what you'll sell].

If this sentence is fuzzy, have AI help:

ROLE: You are a blunt startup advisor.
CONTEXT: Here is my rough idea: [describe it]. My skills and experience: [list].
TASK: Give me 5 sharper versions of this idea, each aimed at a narrower, more specific customer.
OUTPUT: A table with: customer | problem | offer | why they'd pay now.

Hour 1–3: Steelman the "no"

ROLE: You are my most skeptical potential customer: [describe them].
TASK: Give me the 10 strongest reasons you would NOT buy [offer] at [price]. Then tell me what would change your mind for each.
STANDARDS: Be honest and specific. Don't be polite.

Look for reasons that repeat or feel uncomfortably true. Those are your real risks.

Hour 3–5: Map the alternatives

TASK: List the ways [customer] solves [problem] today, including doing nothing, DIY, free tools, hiring someone and existing paid products.
OUTPUT: Table: alternative | cost | what's good | what's frustrating | how I'd verify this.

Then verify. Search for the alternatives yourself and read real reviews, forum threads and comments. Complaints about existing options are gold, because they tell you what to do better.

Hour 5–8: Find where your customers already talk

Ask AI for communities, hashtags, newsletters, events and search phrases your customer would use. Then go look. Spend an hour reading, not posting. Save exact phrases people use to describe the problem. Those phrases become your headlines.

Hour 8–16: Talk to 5 real people

This is the step that actually validates. Message five people who match your customer (friends of friends, LinkedIn, local businesses) and ask for 15 minutes. Use AI to prepare a short, non-salesy script:

TASK: Write 6 open-ended interview questions to learn how [customer] currently deals with [problem], what they've tried, and what it costs them. No leading questions and no pitching.

Listen for: Have they tried to solve it? Have they spent money on it? Do they bring it up without prompting?

Hour 16–22: Test with a real offer

Words are cheap, so ask for something small and real: a pre-order, a deposit, a booked call or at least an email sign-up for a specific offer. A simple one-page description and a checkout or form link is enough. "That sounds cool" is not validation. A payment or a calendar booking is.

Hour 22–24: Decide

ROLE: You are an objective advisor.
CONTEXT: Here are my notes from 5 customer conversations and my offer test results: [paste].
TASK: Summarize the evidence for and against this idea, then recommend: build, change the offer, or move on.
STANDARDS: Base it only on the evidence I gave you. Point out where the evidence is thin.

Green and red flags