Vexa

Open-source meeting bot API that feeds live transcripts to your agent

Advanced API
Visit VexaBack to AI tools

Own Vexa? Verify ownership to get a dofollow link

Screenshot of Vexa, Open-source meeting bot API that feeds live transcripts to your agent

What is Vexa?

Vexa is an AI tool. Open-source API and MCP server that sends bots into Google Meet, Teams, Zoom and Jitsi calls, streaming live speaker-tagged transcripts over WebSocket. Self-host it for free under Apache 2.0 or pay per seat or per bot-hour in the cloud.

Vexa is a meeting transcription API built around bots that join Google Meet, Microsoft Teams, Zoom or Jitsi calls on your behalf. The bot captures audio and streams a live transcript over WebSocket while the meeting is still running, tagging each segment with a speaker name. Each meeting carries its own language setting, so a single account can transcribe calls in English, German, French, Hindi or any other language without reconfiguring the stack. The product is really two offerings sharing one engine. The first is an agent connector: paste a prompt into Claude Code, Codex or any MCP-compatible client and meetings land in the assistant you already use, billed at $12 per seat per month for one person with one concurrent bot. The second is a developer API: send a bot with a REST call, stream the transcript over WebSocket, receive signed webhooks for meeting start, end and failure events, and keep the recorded audio for up to 12 months. Builders pay $0.30 per bot-hour, plus $0.20 per hour for real-time transcription, which lands at $0.50 per hour all in. Underneath, everything is open source under Apache 2.0, including a Next.js dashboard you can fork. The self-hosted path puts transcription and the AI model on your own infrastructure via Kubernetes, OpenShift or Docker Compose, with Postgres holding all meeting data. Enterprise engagements add a two-week fixed-fee activation sprint, security and data residency review, and a process where your engineers open PRs against the engine. A finance-focused variant addresses fund distribution, private credit and wealth teams that cannot send meeting audio to a third-party cloud. The tool suits engineering teams embedding meeting capture into their own products, agent users who want transcripts in Claude Code or Codex, and regulated companies that need meeting knowledge to stay on their own servers. It competes with hosted meeting bot APIs such as Recall.ai on the main differentiator of self-hosting and auditable source code, rather than on a broader feature set.

How do you use Vexa?

  1. 1Get an API key
    Create an account to receive $2 in free bot credit without entering a card, then generate an API key from the dashboard for use in REST and WebSocket calls.
  2. 2Connect your agent over MCP
    If you only want transcripts inside an existing assistant, copy the one-line prompt from the connect page and paste it into Claude Code, Codex or another MCP client. The agent then joins meetings and reads the transcript live.
  3. 3Send a bot to a meeting
    Call the bots endpoint with a POST request and a meeting URL for Google Meet, Teams or Zoom. The bot joins, captures audio and begins streaming speaker-labelled segments over WebSocket.
  4. 4Wire up webhooks and recording
    Register a webhook endpoint to receive signed events when a meeting starts, ends or fails, and fetch the stored audio later through the API if you need playback or archival.
  5. 5Review transcripts in the dashboard
    Open the included web dashboard to see bots, transcripts and meeting status, or fork the Next.js front end if you plan to self-host and customize.
  6. 6Evaluate self-hosting for compliance
    Teams that cannot use a vendor cloud can compare the open-source and enterprise paths, including the finance variant, before committing to a deployment on their own servers.

Pros and cons

Pros

  • Source code is Apache 2.0 and auditable, so regulated teams can verify what handles their meeting audioAI
  • Self-hosting keeps recordings, transcripts and the transcription model inside your own networkAI
  • Pay-as-you-go pricing is granular, at $0.30 per bot-hour base and $0.20 per hour extra for live transcriptionAI
  • MCP support drops live meeting data straight into Claude Code, Codex or any compatible clientAI
  • New accounts start with $2 of bot credit and no card, enough for roughly six hours of bot timeAI

Cons

  • Self-hosting shifts real operational weight onto your team: Kubernetes or Compose, Postgres and a transcription unit to run and patchAI
  • MCP and shared workspaces are early, and shared workspaces are flagged as self-hosted early onlyAI
  • The Individual seat allows just one concurrent bot, which is tight for anyone running parallel callsAI
  • Per-hour transcription billing plus bot time can climb quickly for products with long or always-on meetingsAI
  • No native mobile or desktop client is described; control happens through the API, dashboard or an agentAI

How much does Vexa cost?

Free trial

Every new account gets $2 in free bot credit with no credit card required, covering roughly six hours at $0.30 per bot-hour with all features available. Top-ups start at $10.

Learn more

Pricing

Four paths: self-hosted open source at $0 under Apache 2.0; Individual at $12 per seat per month with one concurrent bot; Build on Vexa at $0.30 per bot-hour plus $0.20 per hour for real-time transcription, or $0.50 per hour all in; and Enterprise with custom pricing, volume rates and dedicated support.

Learn more

Support

Community support on the open-source tier, dedicated support and volume pricing on Enterprise, plus a fixed-fee two-week activation sprint with security and data residency review for self-hosted deployments.

Integrations

Works with Google Meet, Microsoft Teams, Zoom and Jitsi, plus Claude Code, Codex, OpenClaw and any MCP-compatible client. API access runs over REST and WebSocket with webhook events.

Learn more

Features

Bots for Google Meet, Microsoft Teams, Zoom and Jitsi; real-time speaker-labelled transcription over WebSocket with per-meeting language settings; audio recording and 12-month storage; signed webhooks for start, end and failure events; Postgres-backed history and search; an MCP server for Claude Code, Codex and other clients; REST API, WebSockets and a forkable Next.js dashboard.

Learn more

Frequently asked questions about Vexa

  • How much does Vexa cost?
    Four paths: self-hosted open source at $0 under Apache 2.0; Individual at $12 per seat per month with one concurrent bot; Build on Vexa at $0.30 per bot-hour plus $0.20 per hour for real-time transcription, or $0.50 per hour all in; and Enterprise with custom pricing, volume rates and dedicated support.
  • Does Vexa offer a free trial?
    Every new account gets $2 in free bot credit with no credit card required, covering roughly six hours at $0.30 per bot-hour with all features available. Top-ups start at $10.
  • Does Vexa have an API?
    Yes, Vexa offers an API.
  • How do you use Vexa?
    The walkthrough on this page covers 6 steps: 1. Get an API key 2. Connect your agent over MCP 3. Send a bot to a meeting 4. Wire up webhooks and recording 5. Review transcripts in the dashboard 6. Evaluate self-hosting for compliance.
  • What platforms does Vexa support?
    Vexa is available on Web App.
  • What does Vexa integrate with?
    Works with Google Meet, Microsoft Teams, Zoom and Jitsi, plus Claude Code, Codex, OpenClaw and any MCP-compatible client. API access runs over REST and WebSocket with webhook events.

Status

StatusActive
Views0
Outbound clicks0
Added10/21/2026

Platforms

Web App

Pricing

FreeFreemiumPaid