Vexa
Open-source meeting bots, real-time transcripts, and sandboxed AI agents.
Stream live transcripts from Google Meet, Teams, and Zoom into your own Markdown knowledge base.
Vexa sends automated bots into video calls to stream speaker-attributed transcripts and feed sandboxed AI agents. It suits engineering teams and technical professionals who want to own their meeting data without relying on third-party SaaS clouds.
What it is
The platform combines a meeting capture API for Google Meet, Microsoft Teams, Zoom, and Jitsi with an agentic runtime. Captured calls compile directly into a git repository of Markdown files, implementing an LLM Wiki pattern where agents maintain structured entity pages.
Who it's for
It suits developers and technical teams with Docker proficiency who want to self-host meeting intelligence or build custom transcription pipelines. It does not suit non-technical users looking for a simple web app that works out of the box without command-line configuration.
Key features
Multi-platform bot fleet
Spawns automated bots into Google Meet, Microsoft Teams, Zoom, and Jitsi calls.
Live speaker transcription
Streams draft-then-confirmed, speaker-attributed transcripts in real time over WebSockets.
Git-based knowledge compilation
Automatically compiles meeting notes into a portable, grep-friendly Markdown knowledge base.
Sandboxed agent runtime
Executes coding agents in isolated Docker or Kubernetes containers with no unauthorized network egress.
Up and running in 3 steps
Our honest verdict
What's great
- Captures live meetings across major video platforms using your own self-hosted or cloud infrastructure.
- Compiles conversations into standard Markdown files inside a git repository you fully own.
- Runs agents in isolated ephemeral containers that safely read and write your workspace data.
- Provides a complete standalone transcription API if you choose to ignore the agent features.
Where it falls short
- Running the full stack requires Docker engine v26, Make, and technical configuration.
- Transcription requires external API credentials or a self-hosted GPU transcription unit.
- Building from source for contributors demands a machine with 8 vCPUs and 16 GB RAM.
Pricing & license
Free under the Apache-2.0 licence when self-hosted, with an optional hosted SaaS at vexa.ai that provides $2 of free bot credit and charges approximately $0.30/hr thereafter.
Open remote AI-training roles
880 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Which video conferencing platforms does Vexa support?
Vexa officially supports Google Meet, Microsoft Teams, Zoom, and has offline-proven support for Jitsi.
Can I use the transcription API without the AI agents?
Yes, the transcription API is a complete standalone product where you send a bot and read the stream without running the agent lane.
Where is my meeting data stored?
When self-hosted, your transcripts compile into a local git repository of Markdown files on your own infrastructure.
