Skip to content

AI Agent Architecture

The production runtime around the model.

agent runtimememoryRAGtoolspoliciesobservability

Lessons

  1. 33.1The agent runtime: production architecture around the modelA real AI agent is mostly not the model. It's the system around it that keeps bookings correct when things go wrong. 30 min Pick the tools
  2. 33.2Policies and guardrails: what the agent is allowed to doHow to let an AI agent issue refunds and send emails without simply trusting it to behave. 30 min Prompt injection lab

On the job: try it on your own project

Take an AI agent you run or plan to build. Ask Claude Code to draw its runtime: the loop, the memory, the retrieval, the tools, and where policies and guardrails apply. Mark where a bad output could cause real damage. Write a one-page agent architecture note with the guardrails you will enforce in code, not just in the prompt.

Can you do this now?

0/4. Anything you can't tick yet: reopen that lesson's practice ladder.