Phase 6 · Modern LLM engineering · Systems & safety
Multi-agent systems
Coordinating many agents on one large job.
Open this lesson Create a free account
Why it matters
Phase 1 built single agents: a model calling tools in a loop. Larger jobs, such as researching a market from dozens of sources, auditing a big website, or processing a quarter's campaign data, are increasingly split across many agents working in parallel. Done well, that's faster and more thorough; done badly, it multiplies cost and errors. This lesson is about telling the two apart.
What you'll be able to do
- Describe the main multi-agent patterns: orchestrator-workers, parallel sections, pipelines, evaluator loops
- Estimate the speedup from parallel subagents, and see how dependencies cap it
- Calculate how errors compound across agent hand-offs, and what a verifier buys
- Decide when several agents beat one, given their extra cost and coordination problems
Before this lesson
Next in this phase: Evaluating agents