Built With AI

Mama says AI is the devil. Mama was dramatic, but also slightly right. AI isn't a deity or a miracle worker. It's a tool that generates text quickly - be it code or marketing copy - and occasionally hallucinates with confidence. That last bit is the part you want to keep an eye on.

This project treats AI the same way any sane construction worker would treat a power drill: useful, capable of doing good work quickly, and also capable of turning your thumb into a "learning opportunity" if you lose focus.

What AI is for here (practical)

• Helping implement new features: let it build what it can, but humans always review (and almost always improve) the output.
• Test data and fixtures: generate realistic enough but not sensitive sample data and fixtures for unit tests, integration tests, and end-to-end tests.
• Drafting marketing copy: quick first-pass prose and outlines, generating text is kinda what generative AI does best.
• Review pull requests: always appreciate a second set of eyes and AI does a good job.

What AI is not for here (non-negotiable)

• No autonomous production decisions. Feature flags, access control, and releases are controlled by humans. Never let a model flip the switch for you.
• Not a substitute for code review. Generated code and suggestions go through the same review, build, and test gate as everything else.
• Not an owner. If the AI drafts something useful, a human owns it, edits it, and accepts responsibility.
• No runtime dependency on any external model. This app's behavior is deterministic.

The short version

AI helps us move faster and get unstuck. It saves keystrokes and gives plausible ideas. Humans still do the thinking, the hard tradeoffs, and the code review. If you want the machine to do the dangerous stuff, you can, but don't be surprised when it comes back to bite you.

Mama was dramatic, but also onto something. Use the tool. Responsibly.