# Conversation intelligence API for sales call capture

> Conversation intelligence starts with reliable call capture. Get recordings, per-speaker transcripts, and real-time audio from every sales call via one API.

Source: https://meetstream.ai/use-cases/conversation-intelligence

## Conversation intelligence built on reliable call capture

Sales teams trust conversation intelligence only when every call is captured and every speaker is attributed correctly. MeetStream is the capture layer: one API that records, streams, and transcribes sales calls on Zoom, Google Meet, and Microsoft Teams.

### Reliable capture is the whole product

A conversation intelligence platform with gaps in its call history is a platform reps stop trusting. Coaching scores, deal signals, and forecast insights all inherit the quality of the underlying capture, and capture is exactly where in-house bot infrastructure fails first: a platform update ships, joins start failing on one provider, and a week of calls disappears.

MeetStream maintains the bot layer across Zoom, Google Meet, and Microsoft Teams so your capture rate is not tied to your team's ability to chase platform changes. Connect your users' Google or Outlook calendars and bots schedule themselves onto booked calls up to three months ahead, so coverage does not depend on reps remembering anything.

### Who said what, exactly

Attribution is where sales insights get credibility. MeetStream provides separate audio streams and files per participant, plus transcripts with speaker identification, so talk ratios, monologue lengths, and quoted objections are computed from ground truth rather than guessed from a mixed track.

That precision compounds downstream. A coaching note that quotes the exact sentence a prospect said, with a timestamp into the recording, lands very differently from a vague summary.

- Per-participant audio for exact talk-time and interruption metrics

- Diarized transcripts for accurate quote attribution in call reviews

- Join and leave times to see which stakeholders actually attended

### Coach live, not just after the call

Real-time audio streams over secure WebSocket at around 200ms latency, with a separate stream per participant. That is fast enough to power live features: surfacing a battlecard when a competitor topic comes up, flagging pricing questions to a manager, or nudging a rep who has been talking for four straight minutes.

If you would rather not run your own speech-to-text, real-time transcripts with speaker labels arrive at your webhook during the call, ready for your detection logic.

### Ship it with one API call

Send a POST request with the meeting URL and a bot joins the call, no host permission or paid meeting plan required. Webhook events track the full lifecycle, so your platform knows when recording starts, when the meeting ends, and when media and transcripts are ready to process.

```
curl -X POST https://api.meetstream.ai/api/v1/bots/create_bot \ -H "Authorization: Token $MEETSTREAM_API_KEY " \ -H "Content-Type: application/json" \ -d '{ "meeting_link": "https://zoom.us/j/74648392" , "bot_name": "MeetStream Bot" }'
```

### From recordings to deal intelligence

After the call you get the MP4 recording, per-participant files, the async transcript, and meeting metadata. Feed the diarized transcript to your models for topic detection, risk signals, and next steps, and use attendance metadata to track multithreading across the buying committee.

Enterprise buyers will also ask about your vendors. MeetStream encrypts data in transit and at rest, is ISO 27001 certified and GDPR compliant, is under audit for SOC 2 Type 2, and supports bring-your-own S3 storage when customers want recordings kept in their own buckets.

### How it works

- Connect calendars or send a POST with the meeting URL to put a bot on every sales call.

- Stream live audio and diarized transcript segments while the conversation happens.

- Collect the recording, per-participant audio, and metadata when the call ends.

- Run your scoring, coaching, and deal-signal models on clean, attributed data.

### Frequently asked questions

Very, because they are computed from separate per-participant audio streams rather than inferred from a single mixed track. You know exactly who was speaking and when, down to join and leave times for each attendee.

Yes. Audio streams over WebSocket at around 200ms latency and real-time transcripts arrive via webhook during the call, which leaves enough headroom for detection logic and an in-call alert while the conversation is still happening.

No. Once a calendar is connected, bots are scheduled automatically for upcoming meetings, and the bot joins as a visible participant. No host permission or paid meeting plan is required on either side.

Pricing is usage based, see the pricing page. You can start with free credit and dedicated Slack support while you build your integration.

### Keep exploring

### Build it on MeetStream

Spin up your first meeting bot in minutes. One API to join, record, stream, and transcribe across Zoom, Google Meet, and Microsoft Teams.

---

Full site index for agents: https://meetstream.ai/llms.txt
API contract (OpenAPI 3.1): https://meetstream.ai/openapi.json
MCP discovery: https://meetstream.ai/.well-known/mcp