A year ago, 'build your own tool' was advice for programmers. Now a motivated non-developer can go from a nagging problem to a working tool over a weekend, using an AI assistant as the engineer. The skill that matters isn't coding. It's describing what you want clearly.
Saturday: describe the problem
Start by writing down the annoying task in plain language, including a real example. 'Every Monday I download a spreadsheet of orders, and I have to split it into one file per store and email each manager.' That paragraph is your spec. Hand it to an AI assistant and ask it to build the smallest thing that does it.
Sunday: make it runnable by someone else
A tool that only runs on your machine is a personal win. A tool anyone can run is a product. Ask your assistant to add:
- A SETUP file with the exact steps, or a one-line command.
- Clear notes on what it needs (a folder, a runtime, any keys).
- A tiny sample so a buyer can try it without their own data.
The bottleneck used to be building. Now it's noticing which of your annoyances other people share.
The part that's actually hard
The building is the easy 20% now. The valuable skill is taste: noticing that a problem you have is a problem thousands of others have, and describing the fix so plainly that both a buyer and their AI assistant understand it instantly. That's not an engineering skill. It's the same instinct that makes someone good at their job. Which is exactly why the people best placed to build these tools are professionals, not programmers.