Tutorials · Chapter D (4/4) · ~11 min
Multi-agent in code
Try it → see it → read → next
**Orchestrator + workers** — split tasks, cap steps, merge results without runaway loops.
Optional on: Professional
Try tools: Agents & automation
Try yourself
Playground
Multi-agent routing
Orchestrator plans and merges — workers do bounded subtasks with step caps.
Gather docs from 3 sources
Recap
What you just did
You routed subtasks to orchestrator vs worker roles.
Read
Pattern
- Orchestrator receives goal + worker roster
- Workers run bounded tools (max 3 steps each)
- Orchestrator merges JSON summaries
- Global step cap across all agents