Before you pitch anything, you need to know it's real. This week is split into two stages — Research (is the problem real?) and Development (is the concept good?) — with step-by-step ways to use ChatGPT or Claude to move through both faster.
Most rough ideas fail because people jump straight to building. These two stages happen in order — research tells you whether the problem is worth solving, development shapes the idea into something you could actually pitch.
Find out if the problem is real, who actually has it, what they currently do about it, and whether they'd pay to solve it better.
Shape the validated problem into a concept — a one-line pitch, a smallest testable version, and a rough sense of how it could make money.
Copy any of these prompts into ChatGPT or Claude, swap in your own idea, and work through them one at a time — don't skip ahead until each one gives you a real answer, not a vague one.
Most ideas start as a solution looking for a problem. Flip it: write one sentence describing who hurts, and why, before you describe what you'd build.
I have a rough idea: [describe your idea]. Help me find the underlying problem it solves. Ask me clarifying questions one at a time until we can write a single, specific problem statement — not a solution.
"Everyone" is not an audience. Push for specifics: context, behavior, what they're already doing instead of your idea.
Based on this problem statement: [paste your problem statement], generate 3 distinct audience personas who experience it differently. For each, describe their current workaround and what would actually make them switch to something new.
Check whether people are already talking about this problem or paying to solve it. If you're using a tool with web search (Claude or ChatGPT with browsing on), point it at the open web.
Search for how people currently solve this problem: [describe the problem]. List the existing tools, products, or workarounds they use today, then summarize the most common complaints people have about those options.
Every idea has competition, even if it's "doing nothing." Get an unflattering comparison, not a flattering one.
Here are the closest existing alternatives to my idea: [list 2-4 competitors or workarounds]. Compare them across price, audience, and their biggest weakness. Then tell me honestly where my idea doesn't yet have a strong enough edge.
Once the problem is validated, turn it into something you could actually explain to a stranger in one breath — and the smallest version you could test before building anything real.
A clear concept fits a formula. If you can't fill in the blanks, the idea isn't focused yet.
Turn this rough idea into a one-sentence concept using the format: [Product] helps [audience] do [outcome] by [mechanism], unlike [current alternative]. My idea: [describe it]. Give me 5 versions to choose from.
You don't need to build the whole thing to learn if people want it. Find the smallest version — or even a "fake door" — that tests real demand.
Given this concept: [paste your one-line concept], what is the smallest possible version I could build, fake, or mock up in one week to test whether people actually want it? Suggest a "fake door" test if one fits.
You don't need a full financial model yet — just an honest first pass at how this could make money, and which assumptions are shakiest.
Using a basic business model canvas, help me fill in revenue streams, cost structure, and key assumptions for this idea: [paste your concept]. Flag which assumptions are the riskiest and least tested.
Before Week 2's peer consultation, get comfortable defending the idea out loud — even to an AI playing devil's advocate.
Play the role of a skeptical investor hearing this idea for the first time: [paste your concept]. Ask me the 5 hardest questions about it, one at a time, and don't let me move to the next until I give you a real answer.
The steps above apply to any idea, but the details shift depending on what you're building. Pick the closest match.
For apps and digital products, the fastest signal is how quickly a new user reaches their first real "aha moment" — and how ruthlessly you can cut scope before you build anything.
Sketch a 5-screen user flow for this app idea: [describe it], focused on getting a brand-new user to their first "aha moment" as fast as possible. Call out anything that could be cut from this flow.
What's the single feature I could cut from this idea — [describe it] — and still deliver about 80% of the value? Argue for ruthless scope-cutting, not for keeping everything.
Physical products and inventions live or die on real-world feasibility — cost at scale, materials, and what has to be true before a prototype is worth building.
For this physical product idea: [describe it], outline the likely manufacturing steps and materials involved, and give a rough per-unit cost range at low volume (about 100 units) versus higher volume (about 10,000 units).
What general categories of regulation, safety standards, or intellectual property should I research before building a prototype of this: [describe your invention]? Point me toward the right kinds of resources, not legal advice.
Service businesses live or die on positioning and getting the first handful of real customers — pricing tests and outreach matter more than a polished deck.
Help me price this service idea three different ways — hourly, package, and retainer: [describe your service]. Which pricing model fits best for [describe your target audience], and why?
Write 3 outreach messages I could send to find my first 10 customers for this service: [describe it], each testing a different angle or hook.
Content and brand ideas live or die on a niche specific enough that a stranger could describe it back to you, and a distribution channel that matches your real bandwidth.
Help me define a niche for this content or brand idea — [describe it] — specific enough that a stranger could describe it back to me correctly after hearing it once.
Given this brand concept: [paste it], suggest 3 content formats and where to post them that fit both the audience and my actual bandwidth as a solo creator — not a full studio's worth of output.
Once your problem is validated and your concept is written down, you're ready for peer consultation — where real people (not just an AI) stress-test your idea.