Vet the idea in an hour

⏱ 60 min8 steps

This one runs in the . If you use , see Check the idea first.

Most owners have a new idea every few weeks: a class, say, or a second van. The hard part is deciding before it eats a season. You'll give one idea a single hour with Claude, and the most important line gets written before you open the chat: the result that would make you stop. Claude looks up what's already out there and helps you plan a small test. If the hour runs long the first time, that's normal.

Claude app Claude on your phone or computer, where you chat with it by typing. claude.ai is the same thing in a web browser. All wordsClaude Code Claude, working inside Terminal. It can make files and run things on your Mac, with a safety checker watching what it does. All words

Steps

Step 1 of 80 done · about 60 min left

Step 1 of 8, not done yet
Step 2 of 8, not done yet
Step 3 of 8, not done yet
Step 4 of 8, not done yet
Step 5 of 8, not done yet
Step 6 of 8, not done yet
Step 7 of 8, not done yet
Step 8 of 8, not done yet

Write the idea in one sentence
Pick one idea and write it so a stranger would understand it. A bakery might try a Saturday cake-decorating class. A plumber might offer a yearly check-up plan. A bookkeeper might add payroll for her smallest clients. A dog groomer might buy a van and go mobile.
1. Set a timer for one hour.
2. Write the idea in one sentence, in a note or on paper.
3. Under it, write who would pay for it, as narrowly as you can, like regulars who already buy birthday cakes.
That's not what I see
I have three ideas and can't pick one.
Pick the one a customer asked you about most recently. Put the other two on a list for later. You can them the same way next month.vet Check an idea quickly and fairly, the same way every time, to decide whether it's worth more of your time. All words
Write the result that would make you stop
This is the step people skip, and the one that saves them. Decide now, before Claude says anything, what result means you drop the idea. Use your own number. Nobody else can pick it, Claude included.
1. Under your idea, write your in one line, like If fewer than [your number] people sign up, book or pay a deposit by [date], I stop.
2. Write your budget for testing it: the hours and the money you'll spend before you decide.
3. Leave this note open. You'll paste from it later.
stop result The result, written down before a test starts, that means you drop the idea or change it. All words
That's not what I see
I don't know what number to pick.
Pick one you'd be disappointed by, not one you'd be crushed by. Ask yourself: if only this many signed up, would I still want to spend my weekends on it? If the answer is no, that's your number.
Start a new chat and set the rules
This sets the rules before Claude says a word about your idea. The big one: Claude never invents numbers about how many people would buy.

drafts and sorts; a person approves anything with a price, a date or your name on it. Whether the idea goes ahead is yours to decide.
1. Open the Claude app or claude.ai and start a new chat.
2. Copy the prompt below.
```
I own a small business, [what you do], and I want to vet one new idea before I spend money on it. You'll help me in this chat. The rules: never make up numbers about demand, market size, how many people would buy, or what they'd pay. If you look something up, give the link for every fact. If you can't find something, write "couldn't find" and move on, rather than guessing. Mark your own opinions as "my take". I've already written the result that would make me stop. You never change it or talk me out of it. My idea: [your idea in one sentence]
```
1. Paste it into the chat. Replace the parts in square brackets with what you do and your idea in one sentence.
2. Send it.
prompt What you type to Claude. A good one says what you want, how to work and what never to do. All wordsAI Short for artificial intelligence. Here it means a program like Claude that reads what you write and writes back in plain words. All words
You should see:
Claude says it understands the rules, then says something back about your idea or asks a question about it.
That's not what I see
Claude started telling me how big the market is.
Reply: No numbers you didn't find with a link. Take that out. Then carry on.
Ask Claude for the one thing that would sink it
Every idea rests on something you're counting on without having checked. This prompt asks Claude to name the one that matters most, so your test can go straight at it.
1. Copy the prompt below.
```
What's the single assumption that would sink this idea if it turned out to be wrong? Pick one, and say why in two or three lines. Then list two or three more assumptions I'm making without knowing it, one line each. Don't tell me whether the idea is good. Mark all of it "my take".
```
1. Paste it into the same chat and send it.
2. Write the one at the top of your note.
assumption Something you're counting on being true without having checked it yet. All words
You should see:
One assumption with a short reason, then a short list of others, marked as Claude's take.
That's not what I see
The assumption Claude picked isn't the one that worries me.
Trust your gut on this one. Reply with the one that worries you, and ask Claude to use it for the rest of the chat.
Have Claude look up who already sells something like it
Claude uses to look things up. It's built in. If you see a Web search switch under + in the message box, turn it on. When it doesn't search, Claude works from what it already knows, and that can be out of date. This prompt asks for links on everything, and a plain "couldn't find" when there's nothing.
1. Copy the prompt below.
```
Search the web for businesses that already sell something like my idea, near [your town] or online. For each one, give the name, what they sell, any price they publish on their own website or listing, and the link. Then find where people talk about this problem online, like local groups, forums or trade associations, with links. Say "couldn't find" for anything you can't find. Don't estimate prices, and don't count or size the demand. If you can't search the web in this chat, say so and stop. Don't list anything from memory.
```
1. Paste it into the same chat. Replace the part in square brackets with your town.
2. Send it.
3. Open each link and check it says what Claude says it does.
web search Claude looking things up on the internet during a chat. Its answer comes with links to the pages it used. All words
You should see:
A short list of businesses with links and any published prices, then a list of places people talk about it, or a note that Claude couldn't find some of it.
That's not what I see
Claude answered without looking anything up.
Web search is built in. If you see a Web search switch under +, turn it on. Then reply: Search the web for this, and give me a link for each thing you found. On a Team or Enterprise plan, an Owner has to turn web search on for the workspace first. If it still can't look things up, do the search yourself and paste what you find into the chat.
Claude gave something without a link.
Reply: Show me where you found that, or take it off the list. Then open every link and check it says what Claude says it does.
Someone already sells exactly my idea.
That can be good news. People may already pay for it. Ask Claude: What do customers complain about in their reviews? Quote them and link each one. That's where you could do better.
Have Claude write questions to ask real customers
A search shows you what exists. Only people tell you whether they'd change what they do. "Would you buy this?" gets a polite yes that means nothing, so these questions ask about real life instead.
1. Copy the prompt below.
```
Write five questions I can ask five of my customers about how they handle this problem today. Ask about what they did last time, what it cost them in time or money, and what they've already tried. Never ask "would you buy this?" or describe my idea in the questions. Then write a short, friendly message, under 60 words, asking a customer for ten minutes. Say plainly that I'm not selling anything.
```
1. Paste it into the same chat and send it.
2. Pick five customers who'd want this most, and send the message yourself, in your own words if you like.
You should see:
Five questions about what customers do now, then a short message asking for ten minutes.
That's not what I see
I don't want to bother my customers.
Most people like being asked about their own day. Start with two or three you know well. If nobody will spare ten minutes on it, write that down. It tells you something too.
Have Claude help you plan the smallest test
The test should answer one question: is that assumption true? A sign-up sheet on the counter can count, or a phone call to ten regulars. This prompt keeps the test inside your budget and leaves the stop result to you.
1. Copy the prompt below.
```
Suggest the smallest test I could run in the next two weeks that answers whether the assumption you named is true. It has to fit inside my budget: [your time and money budget]. Give me two options, each with what I'd do, what I'd count, and the date I'd count it by. Don't suggest a stop result. I've written my own.
```
1. Paste it into the same chat. Replace the part in square brackets with your budget from Step 2.
2. Send it.
3. Pick one test, and put the end date in your calendar.
4. Check the test counts the same thing as your stop result. If it doesn't, change the test, not the stop result.
You should see:
Two small tests, each with what you'd do, what you'd count and a date, and no stop result.
That's not what I see
Both tests cost more than my budget.
Reply: Smaller. What could I do with a sign-up sheet on the counter, one post, or a phone call? A test you can afford beats one you put off.
Put it on one page as a worksheet
This prompt turns the hour into a worksheet, with your stop result in your own words. When the test ends, you fill in what happened and compare it with the line you wrote before you started.
1. Copy the prompt below.
```
Turn this chat into an idea-vet worksheet I can fill in again for the next idea. Put these sections in order, each with blanks: the idea in one sentence; who it's for; my stop result, in my words; my time and money budget; the one assumption that would sink it; what's already out there, with links; five questions for customers; the test I picked, what I count and the date. Fill in today's answers from this chat, and write FILL IN where I haven't decided yet. Leave a section at the end called "What happened" for me to fill in after the test. At the top, write these rules: I write my stop result before I start. Claude never makes up demand numbers. Every fact has a link or says "couldn't find". I decide.
```
1. Paste it into the same chat and send it.
2. Check the stop result matches your note word for word.
3. Copy the worksheet into a note or a document called Idea-vet worksheet.
If the idea passes and you ever want to build a site or tool for it, start with the Build something new section's lesson "What to build, and what you need first". For a deeper vet, with several agents checking each other, see "Agents as a team: , expand, vet" in the Connect and automate section.
brainstorm Opening an idea up wide before judging it. Here, it starts with questions, not answers. All words
You should see:
A worksheet with the rules at the top, today's answers filled in, and a blank "What happened" section at the end.
That's not what I see
Claude reworded my stop result.
Copy your stop result from your note and paste it over Claude's version. It has to stay in your words.

Your challenge

Take the new idea you keep coming back to and give it one hour this week, with a timer. Write your stop result before you open the chat. Then run the smallest test, and when it ends, read your stop result out loud before you decide anything.

Last checked 2026-10-21

Build something new › 1. Grow, in the app