AI Systems
AI Sales Call Review Automation
Built an AI-powered sales call auditing system that automatically processes call transcripts and generates structured performance reviews.
AI Automation Engineer / Make.com & Zapier Developer
Business summary
- Automated sales call transcription review and scoring.
- Reduced manual QA work for sales managers.
- Improved feedback speed and coaching consistency.
What I built
When a Zoom call ends, Fathom generates the transcript and Zapier captures it, then sends the payload (transcript, rep name, call date) to a Make.com webhook. Make parses the JSON and runs a multi-step OpenAI (GPT / o-series) prompt chain that evaluates the call against a custom audit rubric, scoring areas like discovery, objection handling, pitch quality, and pricing discussion. The final structured review is posted to a Slack channel (#sales-call-reviews) and tagged to the relevant sales rep, with retry logic and failure alerts routed to #automation-errors.
Automated sales call evaluation and improved coaching feedback speed and consistency.
Architecture
A simple operating path with explicit ownership.
Challenges
- Ensuring consistent scoring across different call types
- Handling incomplete or noisy transcript data
- Maintaining reliable webhook delivery between Zapier and Make.com
- Reducing false negatives in AI evaluation outputs
Results
- Faster sales coaching feedback loops
- Reduced manual call review workload
- More consistent and structured performance evaluations
- Improved visibility into sales team performance
Tech Stack
Gallery
