[Suggestion] Multi-agent voice orchestration with specialist delegation and context handoff (Python)

[deepgram-robot]

What to build

A voice agent system that demonstrates multi-agent orchestration — a primary agent that delegates to specialist sub-agents (e.g., billing specialist, technical support, scheduling) and transfers conversational context between them. Show how to build agent-to-agent handoff patterns using Deepgram's Voice Agent API.

Why this matters

As voice agent applications grow more complex, developers need patterns for routing conversations between specialized agents without losing context. This is essential for enterprise deployments where a single monolithic prompt cannot handle all domains. Developers building contact centers, healthcare triage, or multi-department support systems need a reference architecture for agent delegation with Deepgram.

Suggested scope

• Language: Python
• Deepgram APIs: Voice Agent API with Nova-3 STT + Aura TTS
• Architecture: Router agent + 2-3 specialist agents (e.g., billing, tech support, scheduling), each with their own system prompt, tools, and domain knowledge
• Key patterns: Context transfer between agents, conversation history forwarding, warm handoff with summary, return to router
• Complexity: Medium-high — multiple agent configurations, routing logic, session management

Acceptance criteria

• Runnable with minimal setup (clone, add API key, run)
• README explains the multi-agent architecture clearly with a diagram
• Uses current SDK version
• Demonstrates at least 2 specialist agent handoffs with context preservation
• Shows how conversation history transfers between agents
• Includes logging that shows the delegation flow

---

Raised by the DX intelligence system.