[Suggestion] LangGraph multi-step voice agent with Deepgram STT and TTS (Python)

[deepgram-robot]

What to build

A Python example demonstrating a LangGraph stateful agent that uses Deepgram STT for voice input and Deepgram TTS for voice output, with multi-step tool calling and conditional branching between agent nodes.

Why this matters

LangGraph is the fastest-growing agentic AI framework, enabling developers to build stateful, multi-step agent workflows with cycles and conditional logic. Developers building voice-enabled AI agents need a reference implementation showing how to wire Deepgram's streaming STT and TTS into LangGraph's graph-based execution model. This is distinct from the existing LangChain example — LangGraph handles stateful agent loops, not linear chains.

Suggested scope

• Language: Python
• Framework: LangGraph (langgraph >= 0.2)
• Deepgram APIs: Streaming STT (Nova-3), TTS (Aura), Audio Intelligence (optional sentiment)
• Pattern: Microphone → Deepgram STT → LangGraph agent graph (with tool nodes) → Deepgram TTS → speaker
• Complexity: Medium — requires LangGraph graph definition, tool node implementation, and audio I/O
• Backend only (CLI-based), no frontend required

Acceptance criteria

• Runnable with minimal setup (clone, add API key, run)
• README explains the pattern clearly
• Uses current SDK version
• Demonstrates at least 2 tool nodes in the LangGraph graph (e.g., weather lookup + calendar check)
• Shows conditional branching based on STT transcript content
• Audio streams through Deepgram STT and TTS (not batch)

---

Raised by the DX intelligence system.