PastPUBLISHED AUTOMATICALLY WHEN THE WORKSHOP ENDED
Workshop 6: Agents for Studies
Learn about building AI agents to support with workflows related to studies and learning. Everything from the night, free to re-run.
Slides · 38Present ↗
// RECORDING
NO RECORDING FOR THIS ONE
This was a build-along — the slides and prompts below are the whole thing.
BUILDER PROMPTS5 · COPY ANY
01Learn anything → pick itSegment 4 · Live build
I want to learn [a topic: how vaccines work, Finnish verb types, compound interest…]. Before you build anything, ask me 3 quick questions: what I already know, why I want to learn it, and how much time I have.
02Learn anything → build itSegment 4 · Live build
Build me a one-page interactive learning game for this topic. Start with a 60-second explainer. Then 10 questions that get harder, one at a time: mostly multiple choice, plus one I answer in my own words. When I'm wrong, explain why in one sentence and bring that question back later. Show my score and streak, and end with a list of what I still don't get.
03Learn anything → check itSegment 4 · Live build
Be critical. Check every fact and every answer in the game. Flag anything that's wrong, oversimplified or uncertain, and fix it. Then add one challenge question that's harder than anything so far.
04Critic mode (add to any build)Segment 5 · Build block
Be critical. Evaluate this, don't agree with me. Point out what's wrong, what's weak and what's missing before you say anything nice.
05Fix an agent that misbehavedSegment 5 · Build block
It did: [what happened]
It should have: [one sentence]
Which instruction in CLAUDE.md or .claude/skills caused this? Fix only that file, keep the change minimal, and show me exactly what you changed.