Vexa - AI Meeting Notetaker and Knowledge Chat for Professionals and Teams

Extract business knowledge from Google Meet | Microsoft Teams | Zoom | Discord | Slack | and more

NOTE: The service is live and running, currently being prepared for open sourcing. Planned public release - Late February 2025.

👉 Try it for free at vexa.ai

Vexa combines the best of AI meeting notetakers (like Otter.ai, Fathom.fm, Fireflies.ai, Tactiq.io) with powerful knowledge management and AI chat capabilities. While other notetakers focus primarily on transcription and follow-ups, Vexa transforms your meetings into an intelligent knowledge base - imagine having Claude Projects' capabilities but trained specifically on your company's meeting knowledge, with enterprise-grade security and access controls.

Our distributed system provides:

• Real-time transcription with speaker detection
• Instant knowledge extraction from conversations
• Contextual AI chat that understands your company's entire meeting history
• Enterprise security with granular access management

🚀 Features

During meetings:

Real-time speaker-aware meeting transcription with AI contextual support and chat

Chrome Extension: Real-time transcription and AI assistance during meetings

After meetings:

• Meeting knowledge extraction from conversations and documents
• RAG-powered chat with full company context awareness
• Fine-grained data access control

Dashboard: Knowledge exploration and team collaboration

🏗 System Architecture

Built on scalable microservices architecture:

User Interfaces

Google Chrome Extension

• Real-time transcripts with enhanced readability
• Interactive contextual support with single-click actions
• Contextual chat with access to company-wide knowledge base

Meeting and Chat Interface

• Company-wide data exploration by:
  • Speaker
  • Project
  • Meeting
• Advanced entity search:
  • People
  • Companies
  • Products
  • More

Backend Architecture

1. Streamqueue Service

• Real-time audio stream collection from client browser

2. Audio Service

• Speech-to-text conversion using Whisper
• Load balancer for distributed model instances (coming soon)
• Speaker-aware transcription processing

3. Engine Service

• Core business logic
• Knowledge extraction and accessibility

🛠 Technology Stack

• Frontend: React, Chrome Extension APIs
• Backend: Python 3.12+
• Data Layer: Redis, PostgreSQL, Qdrant, Elasticsearch
• Infrastructure: Docker, Docker Compose
• AI: local Whisper, Openrouter for LLM calls

🔗 Quick Links

• Product: vexa.ai
• LinkedIn: @vexa.ai
• X: @grankin_d
• Discord: vexa discord server invite link

---

⭐ Star this repository to get notified when it becomes public!