Track sales call stage and objections as the call happens
Feed a live call transcript to Jev to tag the call stage, spot objections and score key moments for the rep or coach.
The use case
Sales reps lose the thread of their script on live calls, and managers only hear a fraction of recordings. Streaming transcription feeds short chunks to Jev, which answers typed questions: which stage of the call is this, did the buyer raise an objection and of what kind, did the rep confirm next steps. The rep sees a stage marker and a prompt for the matching talk track; after the call, the same labels become a coaching report. A manager reviews flagged moments before using them in a review, and any close probability shown is a rough signal, not a forecast.
Example workflow
- Stream call audio to a transcription service such as Deepgram
- Send each transcript chunk to Jev with stage and objection Choice questions
- Show the stage and suggested talk track in a side panel
- Save labelled moments to the CRM after the call
- A manager reviews flagged moments for coaching
Tools
- Jev (TypeSafe AI)
- Deepgram
- HubSpot
- Zoom
Risks and guardrails
- Recording and analysing calls needs consent from both sides
- Transcription errors lead to wrong stage labels
- Reps may read prompts instead of listening
Sources
Written by Agents Lookup. Credits and official pages for the tools this use case names:
- Sales Copilot by Moritz Kremb on Made with Jev: madewithjev.com/builds/sales-copilot
- Sales Copilot: original post or repo by Moritz Kremb: x.com/moritzkremb/status/2102537239662096658
- call-coach-ai by ZeroGold on Made with Jev: madewithjev.com/builds/call-coach-ai
- call-coach-ai: original post or repo by ZeroGold: github.com/ZeroGold/call-coach-ai
- Deepgram (official page): deepgram.com (checked 2026-11-19)
- HubSpot MCP server (official page): developers.hubspot.com/ai-tools/mcp (checked 2026-11-09)
- Zoom MCP server (official page): developers.zoom.us/docs/mcp (checked 2026-11-09)