Context. Women in Tech UK wanted practical AI upskilling for their community, and a talk wasn’t going to cut it. We agreed a four-week Claude Code working group: weekly Saturday sessions, homework between them, and a hard rule that theory earns its place only if someone uses it that day.
What I did. I co-designed the curriculum, taught the sessions and built the delivery infrastructure myself: an attendee hub with recorded lectures, week-by-week PDFs, downloadable skills and copy-paste prompts, all hosted on GitHub Pages with QR codes in the room. Each week had a run sheet planned to the minute, with fallback demos for when live AI decides to be slow in front of an audience.
What happened. By the final session, every participant had built and deployed a live website of their own, in-session. The series earned a retrospective with the organisers and a follow-on partnership.
What I’d reuse. The shipping rule. Confidence with AI tools comes from the moment your own thing goes live, and no amount of watching someone else demo replaces it. Also the hub-first format: when everything from the session lives at one link, the session keeps working after everyone goes home.