An AMS training cohort, week one
New reps need reps. The fastest way to get them is to fail safely against an AI that pushes back.
GoodLeap
GoodLeap's AMS program runs new sales hires through structured training before they take live homeowner calls. The challenge with classroom training is the asymmetry — coaches outnumber learners 1-to-many, and a rep can only run a handful of practice calls before they're on the floor for real.
The team wanted reps practicing every objection, every customer profile, in conditions that felt close to a real call. They wanted the practice to be voice — not a script on a slide.
Voice, not text
We built a real-time voice AI roleplay system. Reps make a call exactly as they would on the floor — the AI talks back with realistic homeowner profiles, raises objections, hesitates, misunderstands, and tests close attempts in real time.
Each session is a full conversation, not a quiz. The reps that handle it well early are the same ones who close on day one of live deployment.
A controlled space to fail in
Managers can drop into any session, score it, or set up scenario packs targeting specific weaknesses — a particular price objection, a hesitating spouse, a rate-comparison question. New reps run twenty objection drills before lunch, something a human roleplay partner could never sustain.
The point isn't replacing the human element of training. It's giving reps something they can't run out of: practice.
“The reps who use this their first week ramp on a different curve.”
Ramp time compressed. Confidence on first live calls went up because by the time reps got there, the AI had already heard their pitch dozens of times.
- Client
- GoodLeap (AMS program)
- Surface
- Real-time voice training tool
- Stack
- Realtime voice AI, scenario engine