Suppose you could…
Build a tiny app from a big little idea
Go from “wouldn’t it be nice if…” to a simple app you can actually use.
A useful little possibility.
You don’t need to begin with a business idea. Start with a tiny tool: a meeting-cost calculator. You’ll describe what you want, ask AI to create a single-page app, and test it with simple numbers. Aim for a working first version in 15 minutes: about three minutes to generate it, five to preview and test, and the rest to make one small improvement. The result is a local prototype, not a production-ready service.
Let’s make a start.
Use an AI chat tool that can generate code. Paste the starter prompt. If your tool offers a preview, use it. Otherwise, copy the generated HTML into a plain-text editor, save it as meeting-cost.html, and open it in your browser.
Use only a standalone HTML file. Don’t install packages, run terminal commands or provide credentials for this exercise.
Create a simple meeting-cost calculator as one standalone HTML file. Use embedded CSS and JavaScript, no external libraries, no network requests and no data storage. Inputs: number of people, average hourly cost per person, and meeting duration in minutes. Output: people × hourly cost × minutes / 60. Label every input, reject negative numbers, support keyboard use and make it work on a phone. Include a reset button and a short explanation. Return the complete file and plain-language instructions for previewing it.
Keep personal and confidential information out of external AI tools. Start with the sample material.
Turn a first try into your thing.
Test 4 people, 30 per hour and 30 minutes. The result should be 60. Try zero minutes, an empty field and a negative number. Ask the AI to fix anything that behaves strangely.
Save the version that works. That is your tangible result. If you have time left, make one small change, such as a currency label or a calmer colour palette. Leave extra features for a later remix.
Pause for the “aha.”
You have described behaviour, checked a result and refined a prototype. Those are real product-building skills.
AI-generated code can have mistakes. A working preview isn’t proof of security or accessibility. Keep this experiment local and avoid private data. Hosting apps with user accounts, payments or stored data needs a separate review.
One possibility leads to another.
Remix the brief into a reading-time estimator or a hobby-budget calculator. Keep it to three inputs and one useful result.
Show someone the local app and ask them to use it without instructions. Note where they hesitate. That is your next improvement.
You tried something. That counts.
Keep a note of what you made or what surprised you. It stays in this browser.