Pick a real problem

Most people, handed an AI, ask it to do something impressive. "Write me a novel!" "Plan my whole life!"

Builders do the opposite. They find something small, real, and annoying — and they fix that.

This unit, you're going to build a tiny AI workflow: not a chat, but a little repeatable process that does a real job. And the whole thing lives or dies on step one: picking the right problem.

Don't build something cool. Build something you'll actually use twice.

One idea: small, real, and repeated

A good first project passes three tests. Miss any one and you'll build something you abandon.

Test	Ask yourself	Bad answer
Small	Can I describe the whole job in one sentence?	"It does everything."
Real	Is this an actual annoyance in my life?	"Someone might want this."
Repeated	Does this happen again and again?	"It happened once."

The best projects are chores. The thing you do every Sunday. The message you rewrite every time. The stuff you look up over and over. Boring problems make great first builds because you'll feel the payoff immediately — and you'll know instantly if it works.

Do the thing

Brainstorm five annoyances from your real life. Not big-world problems — your week. For example:

• "I always forget what to pack for practice."
• "Explaining my homework question to a tutor takes forever."
• "I never know what to cook with what's in the fridge."
• "Writing the same kind of message to my group chat every week."
• "I lose track of which library books are due."

Now score each on the three tests (small? real? repeated?). Circle the one that scores highest. That's your project for the rest of this unit.

Quick check. If you're tempted by an exciting-but-huge idea ("an app that does my homework"), shrink it until it's boring and doable ("a helper that turns my messy question into a clear one before I ask my tutor"). A finished tiny thing beats a dreamed-of giant thing every single time.

Why this matters

Choosing well is most of the battle. People don't fail at building because they can't code — they fail because they picked a problem that was too big, not real, or a one-time thing.

• The problem picks the project. Get this right and the rest is downhill.
• Boring and repeated is a feature. It means you'll use it, test it, and actually know if it's good.

Next lesson, we take your one problem and map the stack — break it into steps and hand each step to the right tool, using the five verbs from Unit 1.