Multi-Agent Orchestration

6 Interactive Sessions

Short, interactive sessions: watch it work, try it, then build it yourself. You can go deeper at any time in the written chapters.

1. 1

   Why a team of specialist agents beats one agent

   The competitor memo needs four different skills: finding facts, working out numbers, writing code and writing the memo. One agent doing all four does each one poorly. A team of specialists, each with its own tools and instructions, does better.

2. 2

   The supervisor: plan the work, hand it out, put it together

   Four specialists do not organise themselves. The supervisor reads the goal, breaks it into tasks, works out what has to wait for what, sends each task to the right agent and puts the results together. It never does the work itself.

3. 3

   Shared memory: how one agent’s findings reach another

   The Writer needs what the Researcher found. Without a shared memory, each agent redoes the work or makes the facts up. Three stores, each with its own lifespan and limit, let findings pass between agents with their author attached.

4. 4

   When agents disagree: checks, votes and safe handoffs

   The Researcher says NovaTech has 24.5% of the market. The Analyst’s model says 19.3%. A quality check and a vote settle it before the memo ships a contradiction, and a handoff with a token moves the work on without doing it twice.

5. 5

   Show the run: an app that makes the agents’ work visible

   A run of several agents is a black box unless you can see inside it. The app shows how the answer was produced, which agent did which task and how sure it was, so you can tell what happened and where it went wrong.

6. 6

   Run it in production: numbers, circuit breakers and budgets

   On day two an agent gets stuck retrying, two agents pass a task back and forth, and the bill climbs. Teams of agents fail in new ways. Per-agent numbers, circuit breakers and cost caps are the controls that catch them.

How this certificate is scored

Your certificate shows how many of your written answers met the rubric below, checked by an AI grader. Anyone who checks your certificate sees it.

• You answer 6 short questions, one for each module, in your own words (at least 40 words each).
• An AI grader checks each answer against what its module asks for: it must answer the question, use the course's practice case specifically, and be correct.
• To pass, at least 5 of the 6 answers must meet that bar. The certificate says the answers were graded by AI.

The rubric

1. Module 1, Agent Design: 4 specialist agents with capabilities.
2. Module 2, Supervisor Pattern: Task decomposition + routing.
3. Module 3, Shared Memory: 3-tier memory with TTL eviction.
4. Module 4, Consensus & Handoff: Voting + quality gates + handoffs.
5. Module 5, Orchestration App: Next.js agent delegation UI.
6. Module 6, Production Pipeline: Metrics + circuit breakers + costs.

What you walk away with