Skip to content
BEYWARE.
BEYWARE.
← All projects
AI & agentsStatus: Pilot

AI voice agent for inbound calls

Challenge

Inbound calls were to be answered around the clock in natural language: answer frequent questions, capture call-back numbers and messages, and hand requests to the right people, with nobody sitting at the phone.

Technology

  • Node.js
  • Fastify
  • WebSocket
  • Telefonie-API
  • Realtime-Sprachmodell
  • Vektorsuche
  • Next.js
  • Docker

Requirements

  • Real-time voice dialogue with low latency
  • Answers grounded in a proprietary knowledge base
  • Phone number capture, including by keypad
  • Multi-tenant management of agents, numbers and knowledge
  • Notifications in the team messenger

Approach

  1. 01WebSocket server bridging telephony media streams to a real-time speech model in both directions
  2. 02Function calling for knowledge lookup, number capture and ending the call
  3. 03Vector search over the knowledge base
  4. 04Web admin for voice agents, phone numbers, knowledge and conversation history
  5. 05Development mode with simulated calls for testing without real telephony

Contact

Which workflow is holding your team back?

Describe it in three sentences. Within 24 hours you get an honest assessment of whether and how we can help.