Phase 0 · Tools · Talking to APIs
Capstone: ad-copy reviewer
Combine files, functions, JSON and an LLM call into one small tool.
Why it matters
This is the shape of most useful AI tools: plain code does everything that can be checked with rules, cheaply and the same way every time, and a model handles the one step that needs judgment or writing. Agents, evals and content pipelines in Phase 1 are bigger versions of what you build here, and the pattern of checking the model's output with your own code is one you'll use constantly.
What you'll be able to do
- Break a real task into load, check, report and generate steps
- Write rule checks as small functions that return a list of problems
- Save results as a JSON report other tools can read
- Use one LLM call for the part rules can't do, and check what comes back