Google's Multimodal Live API delivers real-time voice conversations with 600ms latency through native audio processing and bidirectional streaming.

We built an AI shopping assistant that understands products and guides purchases using voice . By combining Live API with RAG, our system accesses product catalogs and knowledge bases through function calling, creating seamless consultation experiences.

This talk shares our development journey: integrating Live API with RAG infrastructure, designing function calling patterns, managing conversation context, and optimizing performance. You'll learn practical approaches to voice commerce, RAG integration strategies, and lessons from production deployment.