Prompt Playbook: AI Business Ideas PART 4

You've had a business idea. A good one. This is the big one.

You kept it to yourself - didn't want anyone stealing it. Didn't move on it for months, maybe years. "I'll build it when I have time."

Finally, you built something. Took months of nights and weekends. Hard work. Late nights. You were proud. And rightly so.

Released it to the world and... nothing. Nada. Zilch.

Maybe three signups from friends being polite.

All that time wasted.

Here's what actually happened

• Keeping it to yourself = zero external feedback. You built what YOU thought people wanted.

• Not moving on it = missed months of learning. Could have been collecting real data, building an audience, understanding the problem better.

• Building in secret = you were your only feedback loop. No course correction. No validation.

And the failure hits hard because you'd invested so much. It’s psychologically devastating - this whole building a business thing maybe ain’t for you.

Now flip it: quick idea selection, fast build, public feedback, early users, rapid iteration. Lower effort, completely different outcome. That is our goal.

Let’s get started

We Can't Trust Ourselves

Here's the uncomfortable truth: we're biased toward our worst ideas.

We pick the complex ones because they feel more "legitimate." We choose the technically challenging ones because they seem more impressive. We select our personal favourites because... they're our favourites.

And we systematically get it wrong. 🤣 

Meanwhile, the “simple” solution that actually solves a burning problem? We dismiss it as "too basic." Oopsie daisy.

This is why most founders build what nobody wants. Not because they can't identify good ideas, but because they can't stop themselves from picking the wrong ones. We all do it. Let’s look at how we can save ourselves from …ourselves!

The Human Feedback Reality

The best way to pick a winner is to get human feedback. Talk to real potential customers. Test willingness to pay. Get actual validation.

Hands down this is what you need to do. BUT it’s hard.

I know if I tell you to do that now, you'll hit a brick wall. You'll freeze up. You'll overthink it. You'll delay. And you won’t move forward.

So…I’m going to go easy. For now.

Instead we’re using AI as our objective MVP analyst first. Get a winner selected. Build a basic version next week. THEN get human feedback on something real.

We are going to have to talk to humans (our audience) but you’re getting a slight stay of execution!

Using AI for now is not perfect, but it keeps you moving. Movement beats perfection every time.

The Selection Framework

Here’s the basic framework we’ll be using with our AI. Good MVP ideas share three characteristics:

Problem Severity: How much pain does this solve? A problem that costs someone 10 hours weekly beats one that causes mild annoyance. Look for problems people actively try to solve with duct-tape solutions.

Technical Simplicity: Can you build this with no-code tools in a week? If you're thinking about custom development, authentication systems, or complex workflows, you're overcomplicating. Simple tools ship. Complex tools die in development.

Revenue Clarity: Can you explain the pricing in one sentence? "£30/month for unlimited invoice processing" is clear. "AI-powered business intelligence platform with tiered pricing based on usage" is not. Most people go with the latter and then act surprised when people don’t understand what their product does.

The sweet spot? High pain, low complexity, obvious pricing. Cool?

Your AI Business Analyst

OK let’s get rolling with a prompt. For this particular prompt I recommend a reasoning model. Or any model that can give a detailed, “thought-out” response.

identify the most viable option based on data, not emotion.

Analyse these MVP concepts:
[List all ideas from yesterday with their Input → Process → Output format]

For EACH idea, score 1-10 on:

1. Problem Severity
- How painful is this problem?
- How much time/money does it cost?
- How desperately do people want it solved?

2. Technical Simplicity  
- How quickly can this be built with no-code?
- How many edge cases exist?
- How straightforward is the AI implementation?

3. Market Evidence
- How many people complained about this?
- How clear is the problem from research?
- How obviously does this solution fit?

4. Revenue Potential
- How clear is the value proposition?
- How easy to explain pricing?
- Would people pay £20-100/month?

5. Competition Landscape
- Is someone solving this perfectly already?
- Are current solutions too complex/expensive?
- Is there room for a simple alternative?

Provide:
- Score table for all ideas
- Top 3 recommendations with reasoning
- The ONE to build first and why
- Main risk for the chosen idea

Be harsh. Look for flaws. I want truth, not encouragement.

Why This Works

Models are agreeable by default. Ask "is this a good idea?" and they'll find ways to say yes. But ask them to compare and score objectively? They become useful critics.

Too many people fall into the trap of just asking AI if their idea is good. And guess what? It will always say yes.

So they add to their custom instructions “don’t always agree with me”.

And what happens then? They’ll ask if their idea is good and the model will say no.

What’s happening here? The model is agreeing to disagree! It’s still being agreeable…and just complying with our request for it to disagree with us. Useless!

The framework forces comparison. It's not "is this good?" but "which is best?" That relative analysis reveals strengths and weaknesses you'd miss otherwise.

Making the Choice

After running the analysis, you'll likely see a clear winner emerge. It probably won't be your favourite. It might even be the one you thought was "too simple."

That's exactly the point.

The market doesn't care about your impressive technical skills. It cares about problems getting solved. The simplest solution that effectively solves a real problem will beat the complex solution every time. And we’ve used this prompt to force us towards that solution.

Your Day 14 Deliverable

By end of today

• Run the AI analyst prompt on all your MVPs.

• Get the complete scoring table.

• Identify your top 3 with clear reasoning.

• Choose your winner based on data, not gut feeling.

Trust the process. Your winner might surprise you.

Build in Public Component

Share your selection process

"Day 14 of AI Summer Camp: Let AI analyse my ideas objectively.

My favourite idea scored: 28/50 The winner scored: 42/50. Huh.

Lesson learned: Simple solutions to severe problems beat complex solutions to mild problems. Maybe I need to simplify.

So I’m pushing ahead and building: [Your winning idea]"

What's Next?

Tomorrow, we create a detailed build specification for your chosen MVP. Every screen, every flow, every technical decision - documented and ready for Week 4's build.

Keep Prompting,

Kyle