Most workshops teach a framework by having you follow a tutorial that was written to go smoothly. This one doesn't. Attendees build a Company Value Extractor: a Cloudflare Worker that crawls a company's website and uses Workers AI to extract its mission, vision, values, and culture.

Each phase is built around a real incident pulled directly from the project's own decision log, not a hypothetical: a Workers AI response shape that crashes, an API token leaked in plaintext because it was set as a var instead of a secret, a stale deploy that silently kept serving old code for days, and a character-budget bug that quietly truncated the exact content the AI needed most. Attendees don't just read about these. The real fix and its reasoning are revealed only after they've tried.

Attendees will leave with a working, deployed Worker that adaptively crawls a real site's structure, falls back to real browser rendering only when needed, and reports its own diagnostics.

Attendees will be able to:

* Deploy and iterate on a Cloudflare Worker with Workers AI, Browser Rendering, and secrets management.
* Diagnose "it looks like it's working but the output is wrong" bugs using response-level diagnostics instead of guesswork.
* Recognize the difference between a code bug, a config problem, and a process failure — and know which fix actually belongs where.
* Design a crawler that adapts to a real site's structure instead of guessing at a fixed list of paths.

Format: Hands-on workshop, 4 hours, 6 phases
Audience: Developers comfortable with JavaScript who are new to Cloudflare Workers, or who want a structured tour of building and debugging a real AI-backed Worker
Prerequisites: Node.js, a Cloudflare account, npm install -g wrangler