Here's a trick that turns the App Store into a free market research team. Instead of guessing at an app idea, you take an app that's already winning, feed it to Claude, and let it tell you exactly who the customer is, what they're frustrated by, and where the gap is for a better, AI-powered version. Then you hand the plan to Google Stitch and watch the whole thing get designed on screen in seconds. Here's the full flow and the two prompts that do the heavy lifting.

What you need

  • The App Store (or Google Play), for inspiration and real reviews.
  • Claude, to read the app and find the opportunity.
  • Google Stitch, Google's free AI design tool that turns a brief into real app screens. It's in Google Labs and free as of early 2026.

Step 1: Mine the App Store

Open the App Store, pick any category, and find one of the top apps in it. You want something with a lot of reviews, because the reviews are the gold. Then grab three things:

  • Screenshots of every app screen and preview image.
  • The full description.
  • As many reviews as you can copy, especially the 1, 2, and 3 star ones. That's where people tell you exactly what's broken.

Step 2: Reverse-engineer it with Claude

Drop all the screenshots into Claude, paste the description and reviews, and run this prompt. It turns Claude into a product strategist that reads the whole thing and hands you the opportunity.

You are a senior product strategist and market researcher. I'm giving you everything about a top app: screenshots of its screens, its App Store description, and a big batch of its real reviews. Reverse-engineer it and tell me how to build a better, AI-powered version.

Work through this in order:
1. The product: in plain English, what does this app actually do and what is its core promise?
2. The ideal customer: who is this really for? Their situation, the job they're hiring the app to do, and how much pain they're in.
3. The real pain points: read the reviews like a detective. Pull out the most common complaints, the feature requests, the "I wish it did X," and the 1 and 2 star frustrations. Find the patterns, not the one-offs.
4. The gaps: where is this app weak, dated, overpriced, or annoying? What do users clearly want that it doesn't give them?
5. The AI angle: for each top pain point, how could an AI-native version actually solve it in a way this app can't? Be specific about the feature, not just "add AI."
6. The positioning: give me a one-sentence positioning statement for a new app that beats this one, the wedge to enter the market, and the first type of user to target.
7. The verdict: is there a real, monetizable opportunity here, or is this a crowded dead end? Be honest.

Base everything on the screenshots and reviews I gave you, not generic assumptions. Quote real review language where it makes the point stronger.

What comes back isn't fluff. It's a positioning breakdown built on real user complaints, the kind of thing a strategy consultant would charge you for. Read it, push back on it, ask follow-ups. By the end you'll know if the idea is worth building.

Step 3: Turn the plan into a design brief

Once you like the direction, stay in the same chat and have Claude turn the strategy into a design brief you can hand straight to Stitch. Run this:

Now turn that strategy into a complete app design brief I can paste straight into an AI UI design tool (Google Stitch). Give me:

- The app name and a one-line description.
- The core user flow, screen by screen, starting with the 4 to 6 screens that matter most (for example: onboarding, home or dashboard, the core action screen, the results screen, settings).
- For each screen: its purpose, the key elements on it, and the one primary action.
- The visual style: the vibe, color palette, typography feel, light or dark, and the kind of brand it should feel like.
- The components and patterns to use (navigation, cards, lists, charts, and so on).
- Exactly where the AI features from the strategy show up in the interface.

Write it as clear, direct design instructions, not paragraphs of filler, so I can paste the whole thing in and get a clean multi-screen design on the first try.

Copy that entire response. That's your design spec.

Step 4: Design it in Google Stitch

Go to Google Stitch, paste in the whole brief Claude wrote, and hit send. Stitch reads it and generates real, connected app screens in front of you, often the full first flow at once. You watch your idea become an actual interface in seconds, and it looks genuinely good. Tweak any screen by just telling it what to change in plain English.

Step 5: Actually build it

Two easy paths from the design:

  • Export it. Stitch gives you the screens as code and exports to design and coding tools. Hit export and drop it into a vibe-coding tool to turn the design into a working app.
  • Hand it back to Claude. Give Claude the designs and the brief and have it build the app from scratch with Claude Code. You design in one tool and build in the other.

One honest note

This is for inspiration and market research, not cloning. You're not copying someone's app, their name, or their exact screens. You're studying what a market already wants, finding the complaints the leader ignores, and building something genuinely better and different around an AI angle they don't have. Take the insight, not the identity. That's the difference between competition and a copy.

The best app ideas aren't invented in a vacuum, they're found in the reviews of apps people already pay for. The complaints are a to-do list for a better product. All this does is read that list faster than any focus group could.

Pick one app in a category you actually care about tonight, run the two prompts, and see what Claude pulls out of the reviews. The gap it finds is the whole opportunity.
– Anir

If this is the kind of AI thinking you want working on your own business, apply to work with me. And to catch the next breakdown, join the newsletter.