Sales Call Analyzer.
Drop in a call recording, get a coaching report and a CRM-ready summary.
End of run
Two things land at once: a coaching note for the rep (what worked, what didn’t, specific moments to listen back to) and a tidy CRM summary of the call (key topics, objections raised, next steps, decision-maker signals). The recording stops being an opaque 50-minute MP3 and starts being structured data the team can act on.
Challenges it solves
The post-call writeup tax. Reps either rush a half-summary into the CRM or skip it entirely. The agent does both in one pass, a coaching artifact and a CRM artifact, so the rep doesn’t have to choose between coaching themselves and updating the system.
Coaching needs specificity. Generic “ask more discovery questions” feedback is useless. The analysis step runs on the full transcript with GPT-5 so feedback can cite specific moments: where a buying signal was missed, where a question went answered too thinly, where the rep over-talked the close.
Two formats from one input. Coaching prose and CRM-shaped data
are different writing tasks. The workflow handles them separately
(Generate Feedback, then Create CRM Summary) so neither has to
compromise.
Under the hood
Linear pipeline: Start → Upload Audio → Transcribe Audio → Analyze Call → Generate Feedback → Create CRM Summary → Format Output → End. Single model (GPT-5) across the analysis steps;
deterministic upload, transcribe, and format steps around them.