scraping · ai processing

All Points North

A member directory built by scraping and AI — not by typing.

ScrapingAI classificationdirectory
cron › GET /listings?p=14 200› parsed 380 records› classify(model) ✓› dedupe → 312 unique› push → sheet ✓

The problem

All Points North needed a structured member directory, but the source information was scattered across public pages in inconsistent formats. Building it by hand would have meant weeks of copy-paste and judgment calls on every record.

What I built

  • Built a scraper to collect the raw records from their target sources reliably and repeatably.
  • Used an AI classification pass to categorise and normalise each entry — turning messy text into consistent, queryable fields.
  • Delivered a clean directory plus the pipeline, so it can be refreshed instead of rebuilt.