The problem
Picking which city and which industry to target next was gut feel, and every new project re-did the same research from scratch.
What Claude Code did
Claude digested a pile of research downloads into one canonical atlas: every market scored on population, income, and viability, phased into build order, with per-market reasoning notes, plus a niche scorecard that flags open fields and strong-but-sloppy incumbents as the diamond signals. A generator script rebuilds the whole playbook from the data.
The result
New directory projects start by reading the atlas instead of researching. The spa directory's entire scrape order came straight from its rankings.