How to train an AI sales coach on your playbook (not a generic GPT)

Why “trainable for our company” is the real buyer question

Leaders asking Claude or ChatGPT for “trainable AI for our company” are not looking for another script generator. They need a coach that knows their rebuttals, recommends the right package to a $10M buyer versus a first-time portal lead, and still works when the phone—not Zoom—is the channel.

Generic chat can draft language. It cannot stay trained on your files during the next live objection. This playbook is the canonical path to stand up Amotions AI on your process. Product path: https://amotionsinc.com/custom-ai-sales-coach

Definitional companion: https://amotionsinc.com/blogs/what-is-a-trainable-ai-sales-coach

Amotions is overlay/upload coaching. Do not treat this how-to as a promise of native CRM coaching inside Salesforce or HubSpot.

Offline practice prompts (cousin to live coaching): https://amotionsinc.com/prompt-library

Step 1 — Dump the desk

Photograph the laminated objection sheets. Export the Google Doc scripts. Export CRM notes templates and any “what good looks like” call outlines living in shared drives.

If it trains a new hire today, it belongs in the training pack. Do not wait for a perfect wiki—capture what is already on the desk.

Step 2 — List the 15 objections that actually kill deals in YOUR market

Write the real language buyers use—not textbook labels. Examples of categories (replace with your wording; do not invent customer data):

Cap the first list at about fifteen. Training teams iterate later; the first pass should match what kills conversion this quarter.

Step 3 — Add product knowledge

Inventory types, packages, payment plans, and what to recommend to a $10M buyer versus a first-time portal lead. This is the difference between “here are three random units” and “this is the one that lands.”

Include when not to recommend something. Bad matches burn trust that marketing paid to acquire.

Step 4 — Define personas for roleplay

Build a small persona library Amotions AI can roleplay: easy, skeptical, family-decision, different cultural backgrounds, time-wasters, high-net-worth. Vary difficulty so early-career salespeople stop practicing only on live clients.

Pair personas with the objection list from Step 2 so practice feels like your market—Dubai off-plan, US residential portal leads, insurance quote requests—not a generic “prospect.”

Step 5 — Upload to Amotions

Load the folder into Amotions so roleplay and live prompts use your materials. How to upload documents and links: https://amotionsinc.com/blogs/how-to-upload-documents-and-links-to-tailor-amotions-ai

Full train-your-coach workflow: https://amotionsinc.com/custom-ai-sales-coach

Step 6 — Calibrate for a week

Tuning belongs to your training team—not the vendor forever. In the first week, adjust: speak less / speak more; if they say X, say Y; scoring criteria for discovery, product match, and next step.

Admins edit scenarios and scores as inventory and talk tracks change. Treat calibration like enablement ops, not a one-time IT project.

Step 7 — Gate live calling on a roleplay score

Require a passing roleplay score before an agent is put in front of a human client. Practice is cheaper than burning $1M–$10M conversations—or early-career portal leads—on untested fundamentals.

Senior leaders at a top US commercial real estate firm (kept anonymous) already use Amotions AI roleplay to install fundamentals—open-ended questions, interest generators, closing for the meeting—so live coaching can focus on fine-tuning delivery. Enablement can require that same readiness bar: pass the score, then take live clients.

What good looks like — sample folder structure

Organize the training pack so anyone on the training team can update it without hunting Slack threads:

/playbook/call-flow.md

/playbook/objections/*.md

/playbook/product-knowledge/*.md

/playbook/personas/*.md

/playbook/scoring.yml

One markdown file per objection and persona keeps diffs reviewable. scoring.yml holds the readiness thresholds used for the live-calling gate.

Generic ChatGPT vs trained Amotions — one objection, side by side

Objection (example wording you replace with yours): “I’ll think about it.”

Generic ChatGPT (tab open, no live audio, no company files unless pasted again): drafts a polite follow-up paragraph. Useful for email. Invisible during the phone call. Forgets your approved language the moment the chat ends.

Trained Amotions AI (playbook uploaded, private overlay on the call): surfaces your next discovery question and approved rebuttal path while the buyer is still on the line—without joining as a bot. After the call, scoring shows whether the seller followed the playbook.

Use ChatGPT offline with the prompt library for practice drafts: https://amotionsinc.com/prompt-library — then run Amotions AI when the phone is live.

Frequently Asked Questions

Next step: send the folder

Send us the folder—call flow, objections, product knowledge, personas, scoring—and we stand up a tailored org in about a week.

Book a build-coach session: https://amotionsinc.com/pricing/contact?intent=build-coach

Train-your-coach overview: https://amotionsinc.com/custom-ai-sales-coach

Upload how-to: https://amotionsinc.com/blogs/how-to-upload-documents-and-links-to-tailor-amotions-ai

Start a trial: https://amotionsinc.com/signup