Voice AI + RAG Support Prototype
Marie Auzannet
Voice AI + RAG Support Prototype
Purpose
Demonstrate how a conversational workflow can retrieve grounded information, convert the answer to speech, and keep escalation paths explicit.
System
The prototype combines document ingestion, vector retrieval, response generation, voice output, fallback logic and human handoff.
My role
Conversation mapping, knowledge boundaries, routing, prompt structure, test scenarios, documentation and demo assembly.
Evidence boundary
This is a synthetic portfolio demonstration with no client data. It does not claim a production deployment, call-volume result, booking conversion or specific client outcome.
Synthetic system view — voice interface, grounded retrieval and explicit human handoff; no client data.
Synthetic voice-receptionist demo — no client or production data.
Like this project
Posted Aug 13, 2026
A synthetic, no-client-data demo of a voice workflow that retrieves grounded answers, speaks them and preserves a clear human handoff.
