Enterprise AIOrganization: Enterprise (confidential)12 weeks · 2024
Building an AI That Actively Participates in Enterprise Meetings
27+
27+
AI-callable tools across data, content, and visualization
Multi-model
Multi-model
Claude and Gemini support with seamless switching
Real-time
Real-time
Sub-second STT/TTS with continuous listening
The challenge
Enterprise teams needed an AI presence in meetings that could listen, speak, see shared content, answer questions with business context, and generate dynamic visualizations — all in real-time with sub-second latency.
Our approach
We architected a model-agnostic AI meeting participant with a multi-MCP plugin system. Real-time audio via WhisperLive STT and ElevenLabs TTS, Gemini vision for screen comprehension, and a 3-tier memory architecture (Redis + Qdrant + PostgreSQL) for persistent context.
Technology stack
PythonFastAPITypeScriptClaude APIGemini APIElevenLabsRedisQdrantPostgreSQLDockerMCP
Facing a similar challenge?
We'll tell you how this approach adapts to your problem — and what it would take to ship it.
Book a strategy call