Lesson 1Great for AI, and needs your judgment
Not all parts of building something are equal, and AI is far better suited to some parts than others.
- Great for AI: boilerplate that follows a known pattern, converting one format to another, a first draft of a function you already know the shape of, explaining an error message, generating test data.
- Needs your judgment: architecture decisions, anything touching real user data or security, business logic specific to your product, and any code you can't yet explain to someone else.
Real-life example: Kofi, a student in Next Door Tech's Python track, is building an inventory tracker for a friend's shop. Asking AI for boilerplate to read a CSV of products is a great handoff — a well-known pattern. But how the shop's discount rules should work, and whether customer phone numbers are stored in plain text or protected — those are decisions Kofi needs to understand and make himself.
The ship test
If you can't explain what a piece of code does and why, in your own words, you shouldn't ship it — even if AI wrote it and it “works.”