Description
Most AI meeting tools hand you notes after the call, but the hard part is during it: a client asks for a number you don't have, you've been talking for ten minutes straight without noticing, and the question you meant to ask only comes to mind once the call is over.
Call.md puts the AI inside the meeting. It records your mic and system audio together and transcribes you and them as two separate live streams, then suggests in a side panel what to say next and what to ask. It tracks talk ratio, your speaking pace and how many questions you've asked in real time, and gently nudges you when you've slipped into a monologue.
The interesting part is MCP. When the conversation calls for information, it triggers the MCP tools you've configured and shows the results right in the meeting panel, so you don't have to go digging mid-call. When the meeting ends it writes a summary, key points and action items, and can push everything to n8n, Zapier or your CRM via webhook. Transcription and AI features run on VideoDB's cloud and need a VideoDB API key (there's a free tier); settings, history and transcripts are kept in a local SQLite database. There's a one-line installer for macOS; Windows needs a build from source.
Dual-channel live transcription: mic and system audio are transcribed separately so you always know who said what, in a language you pick or auto-detected.
Live assist: contextual suggestions for things to say and questions to ask, plus a read on the other side's sentiment.
Conversation metrics and nudges: live talk ratio, words per minute, question count and a conversation health score, with rate-limited nudges when the call needs steering.
MCP auto-triggering: detects information needs in the conversation, calls your MCP tools and displays the output inline as markdown, links or structured data.
Meeting prep: describe the meeting and it drafts probing questions and a discussion checklist, with optional Google Calendar sync.
Post-meeting summaries: a narrative overview, key points by topic attributed to participants, and concrete action items, exportable as Markdown with the full transcript.
Bookmarks and history: mark key moments during the call and review them later with the transcript; recordings cap at 2 hours with a 5-minute warning.
Workflow webhooks: send meeting data to n8n, Zapier or a CRM automatically when the call ends.
Call.md puts the AI inside the meeting. It records your mic and system audio together and transcribes you and them as two separate live streams, then suggests in a side panel what to say next and what to ask. It tracks talk ratio, your speaking pace and how many questions you've asked in real time, and gently nudges you when you've slipped into a monologue.
The interesting part is MCP. When the conversation calls for information, it triggers the MCP tools you've configured and shows the results right in the meeting panel, so you don't have to go digging mid-call. When the meeting ends it writes a summary, key points and action items, and can push everything to n8n, Zapier or your CRM via webhook. Transcription and AI features run on VideoDB's cloud and need a VideoDB API key (there's a free tier); settings, history and transcripts are kept in a local SQLite database. There's a one-line installer for macOS; Windows needs a build from source.
Features
Dual-channel live transcription: mic and system audio are transcribed separately so you always know who said what, in a language you pick or auto-detected.
Live assist: contextual suggestions for things to say and questions to ask, plus a read on the other side's sentiment.
Conversation metrics and nudges: live talk ratio, words per minute, question count and a conversation health score, with rate-limited nudges when the call needs steering.
MCP auto-triggering: detects information needs in the conversation, calls your MCP tools and displays the output inline as markdown, links or structured data.
Meeting prep: describe the meeting and it drafts probing questions and a discussion checklist, with optional Google Calendar sync.
Post-meeting summaries: a narrative overview, key points by topic attributed to participants, and concrete action items, exportable as Markdown with the full transcript.
Bookmarks and history: mark key moments during the call and review them later with the transcript; recordings cap at 2 hours with a 5-minute warning.
Workflow webhooks: send meeting data to n8n, Zapier or a CRM automatically when the call ends.
