Vexa v0.6 — The Open-Source Meeting API, now with Microsoft Teams

AI is hungry for data. Vexa delivers.

Vexa v0.6 is how developers and enterprises access real-time meeting data from Microsoft Teams and Google Meet. Build agents, automations, and applications that listen and act on live conversations.

Start hosted, go self-hosted when you're ready. Same APIs, zero rewrites. Your transcripts stay in your house—not someone else's cloud.

What's new in v0.6

Microsoft Teams support: Send a bot to any Teams meeting from a share link (similar to Google Meet). Cross-platform coverage with Google Meet in one API.
Real-time WebSocket streaming: The fastest way to receive transcripts for immediate agent reactions and live integrations.
MCP server for agents: Connect Claude, Cursor, Windserf, and other MCP-enabled agents to Vexa directly—no custom glue code.
Production hardening: Reliability upgrades from a month of operating our hosted service in the wild.

Why it matters

Interchangeability: Start hosted for speed, migrate to self-hosted for control. Same APIs. No lock-in.
Data sovereignty: Keep meeting data inside your infrastructure when you need to. Compliance without compromises.
Developer speed: REST + WebSocket APIs, simple auth, and examples that get you from idea to production fast.
Enterprise reliability: Battle-tested in multi-user, multi-tenant environments.

Key features at a glance

Platforms: Microsoft Teams + Google Meet transcription API
Streaming: WebSocket real-time delivery for sub-second reactions
Agents: MCP server support (Claude, GPT, and others)
APIs: REST + WebSocket for flexible integration
Deployment: Hosted and self-hosted with identical APIs
Reliability: Production-hardened connection + error handling
Open Source: Apache-2.0 licensed

Getting started

Hosted (fastest path)

Create an API key: vexa.ai/dashboard/api-keys
Start a bot and stream transcripts in minutes

Self-hosted (full control)

git clone https://github.com/Vexa-ai/vexa.git
cd vexa
make all            # CPU default (Whisper tiny) — great for dev
# For GPU:
# make all TARGET=gpu    # (Whisper medium) — recommended for production quality

API examples

Start a Teams transcription bot

curl -X POST https://api.cloud.vexa.ai/bots \
  -H "Content-Type: application/json" \
  -H "X-API-Key: <API_KEY>" \
  -d '{
    "platform": "teams",
    "native_meeting_id": "<NUMERIC_MEETING_ID>",
    "passcode": "<MEETING_PASSCODE>"
  }'

Start a Google Meet transcription bot

curl -X POST https://api.cloud.vexa.ai/bots \
  -H "Content-Type: application/json" \
  -H "X-API-Key: <API_KEY>" \
  -d '{
    "platform": "google_meet",
    "native_meeting_id": "<MEET_CODE_XXX-XXXX-XXX>"
  }'

Stream transcripts in real time (WebSocket)

Once a bot joins your meeting, connect to the WebSocket endpoint to receive streaming transcript segments for immediate processing. See documentation for message shapes and event types.

Docs: WebSocket streaming guide

Agents love MCP

MCP-enabled agents can talk to Vexa without custom API calls. Point your agent at Vexa's MCP server to fetch transcripts, start bots, and manage meetings.

Docs: MCP setup

Built for developers, ready for enterprises

For developers

Real-time streaming for instant agent reactions
Simple, predictable API surface (REST + WebSocket)
Examples for popular stacks and automation tools

For enterprises

Self-hosted deployment keeps data in your environment
Interchangeable hosted ↔ self-hosted with zero rewrites
Horizontal scalability and operational visibility

FAQ (short and useful)

Does Vexa support Microsoft Teams and Google Meet? Yes. v0.6 supports both, using one API and a consistent streaming model.

How do I get real-time transcripts? Use the WebSocket streaming API. It delivers transcript segments as they are generated for sub-second reactions.

Can I switch between hosted and self-hosted? Absolutely. Start hosted, migrate to self-hosted when ready. Same APIs, no vendor lock-in.

Is Vexa open source? Yes. Apache-2.0 licensed. You can self-host the entire stack.

Is it ready for production? Yes. v0.6 includes reliability improvements from running the hosted service in real-world multi-tenant environments.

Do you support MCP for agents? Yes. Agents using the Model Context Protocol can connect to Vexa directly.