Skip to content

All courses

Course 6 of The AI Decision Course

Agents in production

For leaders deciding where AI agents can do real work: which processes are ready, how much autonomy to give and when, error rates, and keeping agents in check once they run.

19 short lessons. Each one is a five-minute read with a card your team fills in and three AI prompts to test it.

Join for $299 a yearRead the free lesson

Lessons in this course

  1. Ask what a so-called agent actually does4-minute read
  2. From answering the question to finishing the job4-minute read
  3. Which processes are ready for an agent5-minute read
  4. Which back-office work an agent can take today4-minute read
  5. Five decisions that move an AI agent from pilot to production7-minute read
  6. What error rate can each process live with?4-minute read
  7. How much autonomy to plan for, and how soon4-minute read
  8. Specify the agent before it is built, starting with its limits3-minute read
  9. Getting your experts' judgement into the agent3-minute read
  10. Redesign the interface, or describe what you already have4-minute read
  11. Where each agent rule belongs: code, plain words, or a person4-minute read
  12. Widen an agent's permissions one tool at a time5-minute read
  13. The checks that run between the agent and the customer4-minute read
  14. Who checks the work when agents do thousands of tasks5-minute read
  15. Four decisions that improve an agent after launch4-minute read
  16. One coordinating agent, or a network of specialists4-minute read
  17. Three decisions on agents that dig into incidents3-minute read
  18. How long a task can get before an agent's errors pile up4-minute read
  19. What an agent should remember about a customer, and when it lets go4-minute read

All courses