Give your AI agent ears for every meeting
ScribeNode transcribes meeting audio and exposes it as tools your AI agents can call. Upload via the web dashboard, or pipe audio directly through the MCP gateway.
99.2%
Cantonese accuracy
<2 min
Agent response time
MCP
Native protocol
100 MB
Per file
For AI agents
Meeting transcription your agents can call
Your Claude, Cursor, or custom agent can now upload audio, retrieve transcripts, and act on meeting insights — all through the MCP protocol.
MCP Native
Two tools: start_transcription and get_transcription_result. Drop into any MCP host.
API Key Auth
Generate scoped API tokens from the dashboard. Authenticate with x-api-key — no OAuth dance.
Context Injection
Pass meeting context via initial_prompt. Your agent can provide names, terminology, or topics for sharper results.
Async Polling
Transcriptions run on GPU. Your agent polls for results — non-blocking, event-driven. Perfect for autonomous workflows.
Email Alerts
Agent or human — get notified when transcription completes. Hook into webhooks or watch your inbox.
Claude & Cursor Ready
Pre-configured for Claude Desktop and Cursor. One-line config. Works with any MCP host.
Your agent's workflow
From audio file to actionable transcript — automatically.
Upload
Agent calls start_transcription with an audio path.
Process
Audio hits GPU. faster-whisper transcribes with context awareness.
Retrieve
Agent polls get_transcription_result. Non-blocking.
Act
Agent uses transcript — summarize, extract actions, update tickets.
Give your AI agent meeting superpowers
One API key. Two MCP tools. Your agent transcribes meetings autonomously.