CASE STUDY: Live AI Guidance for Discovery Calls

We built a copilot that joins our own sales calls, tracks what has been covered, coaches in the moment, and has the follow-up written when the call ends.

→ The Challenge

On a discovery call, the person selling has to listen and track a checklist at the same time. Notes pull attention off the prospect, and memory decides what got covered.

Pain Points

  • Notes Compete With Listening: Every minute spent writing is a minute not spent on the person talking, so either the notes or the conversation suffers.
  • No Honest Record of Coverage: A topic mentioned in passing feels covered afterward, and nothing checks whether the question was answered.
  • Follow-Up Rebuilt From Memory: The summary, the next steps, and the follow-up email all get written after the call, from whatever was remembered.

→ Our Solution

We built SalesCoach, a real-time copilot that joins the meeting, listens, and coaches during the call, so the person selling does nothing extra but talk.

We Built

  • Live Meeting Capture: A session is created for a prospect and a call type: discovery, demo, closing, or intro. An AI bot joins the Zoom meeting and transcribes the conversation as it happens.
  • In-Call Checklist and Coaching Cards: Each call type has its own checklist of objectives. Items turn yellow, then green, as they are covered. Short cards suggest what to ask next and flag risks such as a stall, scope creep, or a new stakeholder entering the deal.
  • Second-Pass Verification: A smarter background pass re-checks the checklist against the transcript, so an objective is credited only when it was accomplished, not when it was only mentioned.
  • Automatic Post-Call Report: When the call ends, the system produces a summary, objections, buying signals, gaps against the call plan, next steps, and a follow-up email drafted in the seller’s own voice.

Why We Built It

“I found myself on dozens of nearly identical discovery calls, wondering each time if I’d asked everything I needed to.”

– Michael Trezza, CEO Lithyem

Real-Time Sales Copilot screenshot

Real-Time Sales Copilot

6

Parts in Every Post-Call Report

→ A summary, objections, buying signals, gaps against the call plan, next steps, and a follow-up email are ready when the call ends.

4

Call Types, Each With Its Own Checklist

→ Discovery, demo, closing, and intro calls are each tracked against the objectives that kind of call needs to hit.

75¢

AI Cost Per 45-Minute Call

→ An estimate from a short scripted test call, scaled up to a full-length call.

95%

Of Call Objectives Covered

→ The average share of checklist items covered across real calls.