Skip to content

Ch 16: Multi-Agent Systems Architecture - Introduction

Track: Advanced | Try code in Playground | Back to chapter overview

Read online or run locally

You can read this content here on the web. To run the code interactively, either use the Playground or clone the repo and open chapters/chapter-16-multi-agent-systems-architecture/notebooks/01_agents_and_messaging.ipynb in Jupyter.


Chapter 16: Multi-Agent Systems — Notebook 01 (Agents & Messaging)

We start from the atom of the system: a single agent with a role, tools, and memory, running a perceive-decide-act loop. Then we let two agents talk through typed messages and a shared blackboard.

What you'll learn

Topic Section
What an agent is (role, tools, memory, loop) §1
The perceive-decide-act cycle §2
Typed message performatives §3
Shared state with a blackboard §4

Time estimate: 3 hours


Key concepts

  • Agent — a policy plus tools and memory that acts toward a goal.
  • Performative — the intent of a message (inform, request, propose…), not just its text.
  • Blackboard — a shared workspace; powerful but a contention hazard.
  • Autonomy vs control — more agent freedom means more capability and more failure modes.

An agent is a policy + tools + memory running a perceive-decide-act loop. Messages carry intent via performatives, and a blackboard shares state at the price of contention. These primitives compose into everything that follows.

Run the full notebook in the chapter folder for code and outputs.


Generated by Berta AI