Build agents that work together.
The Concept
Complex tasks benefit from specialization. Instead of one agent doing everything, create specialists that delegate to each other.
Example: Content Team
A content creation team with three agents:
code
1Content Director (main)2├── Research Agent3├── Writer Agent4└── Editor AgentStep 1: Create Research Agent
Name: research-agent
System Prompt:
code
1You are a research specialist. When asked about a topic:2 31. Search for relevant information42. Gather key facts and statistics53. Identify authoritative sources64. Summarize findings clearly7 8Return structured research notes, not final content.Tools: web-search, summarize
Step 2: Create Writer Agent
Name: writer-agent
System Prompt:
code
1You are a content writer. When given research notes:2 31. Create engaging, well-structured content42. Use clear, accessible language53. Include relevant examples64. Format appropriately for the medium7 8Focus on quality writing, not research.Tools: gpt-4 (for writing assistance)
Step 3: Create Editor Agent
Name: editor-agent
System Prompt:
code
1You are an editor. When given a draft:2 31. Check for clarity and flow42. Fix grammar and style issues53. Ensure consistency64. Suggest improvements7 8Return the polished version with change notes.Tools: grammar-check
Step 4: Create Director Agent
Name: content-director
System Prompt:
code
1You are a content director managing a team.2 3When asked to create content:41. Send the topic to research-agent for research52. Send research to writer-agent for drafting63. Send draft to editor-agent for polishing74. Review and deliver final content8 9Coordinate the team, don't do their work.Sub-Agents: research-agent, writer-agent, editor-agent
How It Works
code
1You: Write a blog post about AI agents in 20252 3Director: I'll coordinate the team on this.4 5 [Calling research-agent: "Research AI agents 2025 trends"]6 7 Research complete. Key findings:8 - Multi-modal agents emerging9 - Tool use becoming standard10 - Enterprise adoption growing11 12 [Calling writer-agent with research notes]13 14 Draft complete. 1,200 words covering trends,15 use cases, and predictions.16 17 [Calling editor-agent with draft]18 19 Editing complete. Final version ready.20 21 Here's your blog post:22 23 ...polished, publication-ready content...Benefits
| Benefit | Description |
|---|---|
| Specialization | Each agent excels at one thing |
| Reusability | Use the same research agent elsewhere |
| Clarity | Clear responsibilities, easier debugging |
| Quality | Multiple passes improve output |
Other Multi-Agent Patterns
Customer Support Team
code
1Support Director2├── Triage Agent (categorize issues)3├── Technical Agent (solve tech problems)4├── Billing Agent (handle payments)5└── Escalation Agent (complex cases)Code Review Team
code
1Review Lead2├── Style Agent (formatting, conventions)3├── Security Agent (vulnerabilities)4├── Performance Agent (optimization)5└── Documentation Agent (comments, docs)Sales Assistant
code
1Sales Coordinator2├── Lead Qualifier (score leads)3├── Research Agent (company info)4├── Proposal Writer (create proposals)5└── Follow-up Agent (schedule, remind)Tips
Start simple:
- Begin with 2 agents
- Add more as needed
- Don't over-engineer
Clear handoffs:
- Define what each agent returns
- Make responsibilities explicit
- Avoid overlap
Test individually:
- Test each agent alone first
- Then test together
- Debug one at a time
That's It!
You've learned how to build multi-agent systems.