The fastest way to use AI to find a business idea is to feed it what you already know how to do at work, how many hours you have, and how much money you can risk, and ask for ideas that sell that knowledge to a specific group of people. Then pick one idea and ask three real potential customers about the problem before you build anything. The prompts below walk through each step, and the validation section shows how to run those three conversations without fooling yourself.
Generic "business ideas for nurses" lists are a fine place to browse. The trouble is that they were written for everyone with that job title, so they ignore the things that decide whether an idea works for you: your specialty, your schedule, your savings, and the people you already know. An AI chat can fill that gap, but only if you give it those facts.
Key takeaways
- Your job is your unfair advantage. The best first business usually sells something you already do well to people you already understand.
- Constraints make better ideas. Tell the AI your hours, budget and deal-breakers up front, or it will suggest things you can't start.
- Pick the customer before the product. "Help new grad nurses pass their first-year review" is testable. "A wellness app" is not.
- Validate with three real people before you spend money. Ask about their past behavior, not whether they like your idea.
- Check your employment contract and license rules. Some jobs restrict outside work or what you can advise on.
Step 1: Describe what you actually know
Paste this into a fresh chat and fill in every line. Be specific. "Nurse" is less useful than "eight years in a cardiac step-down unit, precepting new grads."
I want help finding a small business or side income idea that uses what I already know from my job. Don't suggest anything yet. Read my situation and ask me up to 5 questions about anything important I left out.
- My job, specialty and years of experience: [ ]
- Tasks people at work regularly ask me for help with: [ ]
- Things I do faster or better than most colleagues: [ ]
- Groups of people I understand well (patients, parents, small business owners, new hires, etc.): [ ]
- Hours per week I can give this: [ ]
- Money I can risk in the first 3 months: [ ]
- Things I won't do (cold calling, social media on camera, weekends, inventory): [ ]
- Anything my employer or license restricts: [ ]
Step 2: Turn skills into sellable problems
From what I told you, list 10 problems that people outside my workplace have and would pay to solve, where my experience gives me a real edge. For each problem: who has it (be specific), how they solve it today, and why my experience matters. Skip anything that needs more hours or money than I said I have.
This prompt forces the model to start with a customer and a problem. It is also where you'll see whether it actually used your details or fell back to a generic list. If the answers could apply to anyone in your field, push back:
These are too generic. Rewrite them using the specific details I gave you about my specialty and the people I understand. If you don't have enough detail to do that, ask me.
Step 3: Shape three ideas you could start small
Pick the 3 strongest problems from your list. For each, suggest the simplest offer I could sell within 30 days without building software: a service, a workshop, a template, a consulting call, or a small paid community. For each offer include: who buys it, a rough price range I should test, how I'd find my first 3 customers, and the biggest reason it might fail. Mark any price as a guess I need to check.
Then ask for the uncomfortable part:
Now argue against each of these 3 ideas like a skeptical friend who has run a small business. What will I underestimate? Which one is most likely to eat my evenings without making money?
Step 4: Check the practical limits
For the idea I'm leaning toward, [describe it], list the practical questions I need answered before I charge anyone: licensing, liability, whether my employment contract might restrict it, taxes on side income where I live ([your location]), and anything about giving advice in my field that I should be careful with. Don't give me legal conclusions. Tell me what to check and who usually answers each question.
AI is useful for making the checklist. It is not a lawyer or an accountant, and the rules vary by state and country. For anything touching clinical, legal or financial advice, check with your licensing board or a professional.
Step 5: Validate with three real customers
This is where most ideas die, and that is the point. A day of conversations can save you months building something nobody pays for.
The most useful guide to these conversations is Rob Fitzpatrick's short book The Mom Test. Its central idea: people will be polite about your idea, so ask about their lives and past behavior instead of asking whether they like what you plan to build. Use the AI to prepare, then go talk to humans.
I'm testing this idea: [describe it]. The customer is [specific person]. Write 8 questions I can ask in a 15-minute conversation that ask about their past behavior and current workarounds. Don't mention my idea in the questions. No leading questions like "would you pay for" or "would you use." Include one question about what they've already spent money or time on to solve this.
Write a short, honest message I can send to [type of person] I know or can reach through [community or network], asking for 15 minutes to learn how they handle [problem]. Say clearly I'm not selling anything yet. Under 70 words.
After each conversation, write down what they said, as close to word-for-word as you can, and run this:
Here are my notes from 3 customer conversations: [paste]. Separate what they actually did or paid for in the past from what they said they might do. Based only on past behavior, is this problem painful enough that someone would pay to solve it? What would I need to hear in the next 3 conversations to be confident? Don't be encouraging for the sake of it.
A good sign is someone who has already spent money or real time trying to fix the problem. A weak sign is "that sounds great, I'd totally use that." If two of your three people have hacked together their own fix, you have something to test with a small paid offer.
Where a chatbot falls short
These prompts work in any chatbot. You will notice a few gaps.
It doesn't know you between sessions. Every new chat forgets your hours, your budget and the fact that your hospital has a moonlighting policy. You end up re-pasting the whole story and it drifts.
It rarely says "wait, why?" Ask for ideas and you'll get ideas. A good coach asks why you want side income in the first place, because "pay off a loan in a year" and "test a way out of my job" lead to different plans.
It can't help you practice the conversation. Asking a stranger for fifteen minutes, or naming a price out loud for the first time, is the hard part. Reading a script is different from saying it.
That is why we built Carmen. She is an AI career coach who remembers your situation, asks follow-up questions before she suggests anything, and pulls in specialists when needed: Nora to research a market or a type of customer, and Chama to turn a chosen idea into a week-by-week plan that fits around your shifts. See the guide to building side income with Carmen, or start from your job with pages for nurses and call center agents.
If the side project is really the first step out of your current job, how to use AI to change careers covers the bigger move.
Carmen is free to start with 30 voice minutes. Plus is $12 a month and Pro is $29 a month.
FAQ
What is the best AI business idea generator?
Any capable chatbot (ChatGPT, Claude, Gemini) can generate business ideas. The difference comes from the input. A tool that asks about your job, skills, hours and budget will give better ideas than one that takes a single keyword. If you use a general chatbot, paste your full situation first and ask it to question you before suggesting anything.
What are good business ideas for someone with a full-time job?
Ideas that sell what you already know to a specific group of people, and that you can start as a service before building anything. Examples include tutoring or coaching people a few years behind you in your field, running paid workshops, or creating templates and guides for a problem you solve at work every week. Check your employment contract for restrictions on outside work first.
How do I validate a business idea before spending money?
Talk to at least three people who have the problem you want to solve. Ask how they handle it today and what they have already spent to fix it, without pitching your idea. If people have already paid for workarounds, try a small paid offer next. If they only say it sounds nice, keep looking.
Can nurses or teachers start a side business?
Many do, but check two things first: your employer's policy on outside work, and any licensing rules about what advice you can give outside your job. Your state board or professional body can tell you what is allowed. Education, content and coaching for people entering your field are common starting points.
Should I trust AI's price suggestions for my business?
Treat them as guesses to test. AI does not know what your specific customers pay today. Ask real customers what they currently spend on the problem, look at what similar offers charge, and adjust after your first few sales.