AI Voice Call Center
AI answers the business’s incoming calls; summaries and the knowledge base are in a mobile dashboard.
Incoming calls are forwarded to a carrier-provided number; the conversation is conducted in Turkish through an STT → LLM → TTS loop, and each call is summarized and shown in a multi-tenant dashboard.
The business owner enters the knowledge base (opening hours, services, price ranges) in the mobile app; the assistant uses this information to answer calls, take appointments and request notes, and produce a structured summary when the call ends. The backend consists of a low-latency conversation engine and an HTTP API; the data layer on Supabase isolates customers with row-level security (RLS). Webhook-based integration with Turkish carriers such as Netgsm / Verimor is planned.
Highlights
- Turkish-first STT → LLM → TTS conversation loop
- Automatic structured summary after each call
- Multi-tenant schema, data isolation with RLS
- Flutter admin dashboard (calls, summaries, knowledge base)
- Forced version updates and mock-model test infrastructure
Technology
- Python / FastAPI
- Claude API
- Supabase (Postgres + RLS)
- Flutter
- Netgsm / Verimor