Introduction

Welcome to MeetStream!

One API to send bots into Zoom, Google Meet, and Microsoft Teams — record, transcribe, and interact in real time

View as Markdown

MeetStream.ai provides a unified API to deploy interactive bots and AI agents into meeting platforms such as Zoom, Google Meet and Microsoft Teams.

Use one integration to 

  • capture media
  • generate transcriptions
  • interact in real-time

Start Building

Quickstart

Put your first bot in a meeting with one API call.

API Reference

Every endpoint for bots, transcription, calendars, and MIA agents.

Build with AI

Connect the MCP server so Claude or your coding agent integrates MeetStream for you.

Meeting Bot and Agent Features

Core Capabilities

  • Universal Integration: Connect to any major meeting platform via a single API using only the meeting URL.
  • Low-Latency Media: Capture raw audio and video with ~ 200ms latency.
  • Active AI Participants: Deploy bots that speak via TTS, share screens, or inject video streams.
  • Managed Infrastructure: Scale from one to thousands of concurrent sessions without managing bot clusters.
  • Rich Metadata: Track participant join/leave times, speaking duration, and in-meeting chat.
Media Capture & Streaming
  • Record: Export mixed audio, isolated participant tracks, or composite MP4s.
  • Stream: Access real-time WebSocket audio (base64) and RTMP video feeds.
Transcription & Diarization
  • Live Events: Receive webhooks for real-time speaker attribution.
  • Post-Meeting: Access full, speaker-labeled transcripts via the MeetStream engine or third-party providers.
Bot Interaction
  • Messaging: Read and write in-meeting chat via API.
  • A/V Output: Stream live web pages, display static images, or speak using TTS.
Reliability & Management
  • Automation: Sync calendars for auto-dispatch and trigger auto-leave rules.
  • Control: Pause, resume, or mute bots mid-meeting via API.
  • Resilience: Built-in network auto-retries and platform consent handling.

Use Cases

  • Bot joins live calls, streams real-time attributed transcripts via WebSocket
  • NLP layer detects intent signals and objections from the live transcript stream
  • Streams live attributed transcript via WebSocket, enabling real-time coaching cues to be surfaced to the rep mid-call
  • Post-call agent auto-generates follow-up summaries and action items for rep review