Start learning →Glossary Learning record

Delegating to an AI agent: a practical introduction

This course is for someone who already chats with AI but has never handed a real, important task to an agent. You have a coding agent like Claude Code or Codex available, and you want to stop treating it like a fancy search engine and start treating it like a capable but imperfect colleague. You will not learn to code. You will learn to delegate, inspect, and correct.

By the end you'll be able to (course objectives):

Tell the difference between a chat question and an agent task, and choose the right kind of ask.
Rewrite a vague wish into a clear ask with a task, requirements, boundaries, and acceptance criteria.
Verify an agent's deliverable against your own criteria instead of trusting "looks good."
Name the four common agent mistakes (hallucination, stale facts, scope shrink, non-answer) and spot them in a real output.
Record a caught mistake and turn it into a better result and a better ask next time.

Prerequisites: None beyond being able to chat with an AI agent and open a local course page. No terminal, IDE, or programming knowledge is assumed.

Practice environment: Your own agent chat window and everyday tools (notes app, spreadsheet, browser). This course has no embedded sandbox.

Lessons

#	Topic	You'll learn
01	The agent is a worker, not a talker	agent vs chatbot, delegation, what an agent needs to act
02	Write the ask	task, requirements, boundaries, acceptance criteria
03	Check the result	verification checklist, evidence trace, pass/fail
04	Spot the mistakes	hallucination, stale facts, scope shrink, non-answer
05	Catch and correct	correction loop, mistake log, final exercise

Sources listed in sources.md.