Model training and data

How is the AI coach trained, and is our call data used to train it?

Amotions AI learns your process from what you upload: playbooks, talk tracks, objections, and product knowledge. It applies that material at coaching time instead of retraining a model on your calls. Customer conversations are never used to train third-party foundation models, and live audio runs in isolated, encrypted Google Cloud workloads.

Written and reviewed by Pianpian Xu Guthrie, Founder and CEO · Published · Last updated

How it works

How does Amotions AI learn your playbook and use it on a live call?

Three stages. You upload your sales material and Amotions calibrates to your language, typically in about a week. During a call it reads the conversation against that material and shows one next action to the rep. Afterward it scores the call against your rubric.

  1. 1Learn from your material

    Upload playbooks, approved objections and rebuttals, product knowledge, and talk tracks. Amotions calibrates to that language so prompts and scores follow your process. Foundation-model weights are not retrained on your conversations.

  2. 2Coach during the live call

    The desktop app or browser overlay listens on the rep’s computer. Audio travels over encrypted connections to isolated Google Cloud workloads, and one next best action returns only to that rep’s screen. No bot joins the meeting.

  3. 3Score and store in your workspace

    After the call, scorecards are visible to the rep and their team. Playbooks, transcripts, scorecards, and uploads are stored encrypted and kept separate by customer workspace.

Vendor evaluation call · Teams
Only you see this
  • RepHappy to walk through how onboarding works.
  • ProspectFirst, does our data end up training your AI?
Security concern

Suggestion

Answer from the approved data-handling talk track, then offer a security review.

Phases

2/4

  • Opener
  • Discovery
  • Security questions
  • Next step
Illustrative example

Example

What does “trained on your playbook” look like on a real call?

The prompt comes from the material your team uploaded, not a generic script. Here a buyer raises a data question mid-call, and the overlay points the rep to the company’s own approved answer and next step.

Software sales call · ZoomIllustrative example
  • Rep:Does the rollout plan look right for your team?
  • Prospect:It does, but legal will ask whether our customer records get used to train your models.
Legal review aheadOnly the rep sees this

Suggestion

Give the approved data-use answer in one sentence. Then ask who on their side runs vendor reviews.

From your playbook

Objection: data use · Talk track: security review offer

Phases

Demo ✓ · Rollout ✓ · Security questions · Next step

What the rep said next

“Good question to settle early. I’ll send our data-use terms today. Who runs vendor reviews on your side, so I can set up a call with our security lead?”

Compare

How does this compare with other AI coaching tools?

Post-call tools usually learn from recorded calls and report after the fact. Roleplay apps learn from scenarios you write. Amotions AI applies your playbook live and adds practice and scoring. Data policies differ by vendor, so ask each one the same questions.

  • How it learns your process

    Amotions AI
    From your uploaded playbooks, objections, product knowledge, and talk tracks, applied at coaching time.
    Post-call analytics tools
    Often from your recorded call library and the trackers you configure.
    Roleplay-only tools
    From the scenarios, personas, and rubrics you set up.
  • Use of your call data for model training

    Amotions AI
    Never used to train third-party foundation models.
    Post-call analytics tools
    Varies by vendor; check its published policy and contract.
    Roleplay-only tools
    Varies by vendor; practice sessions may not include real customer data.
  • When guidance reaches the rep

    Amotions AI
    During the live call, as a private overlay only the rep sees.
    Post-call analytics tools
    After the call, in a review or dashboard.
    Roleplay-only tools
    Before any real buyer is on the line.
  • What the buyer sees

    Amotions AI
    Nothing. No bot joins the meeting.
    Post-call analytics tools
    Often a recorder or meeting bot in the participant list.
    Roleplay-only tools
    Not applicable: no buyer is present.
  • Where it is the stronger choice

    Amotions AI
    Reps who need your approved answer while the buyer is still talking.
    Post-call analytics tools
    Searchable call libraries, deal analytics, and forecasting.
    Roleplay-only tools
    Large scenario libraries and certification programs.

Amotions is a vendor in this category. The columns describe categories, not specific vendors; confirm any individual product’s features with that vendor.

Proof

Who has trained Amotions AI on their own playbook?

A top US commercial real estate firm uses AI roleplay to drill open-ended questions, interest generators, and closing fundamentals. Venu AI’s founder gives new reps the same structure and questions through live coaching instead of shadowing every call. Both run on tools their teams already use.

Works where your calls happen

All integrations

Pricing and free trial

Basic is free. Starter is $49/seat/month billed annually ($59 billed monthly). Every plan is listed on the pricing page.

FAQ

Questions people ask

Short, direct answers on how it works, what it needs from you, privacy, and pricing.

Is our call data used to train AI models?

Customer call transcripts, audio uploads, and chat sessions are private to your organization and are never used to train third-party foundation AI models. Playbook personalization applies your process at coaching time; it does not retrain foundation-model weights on your conversations.

How is Amotions AI trained on our sales playbook?

Customers upload their playbooks, approved objections and rebuttals, product knowledge, and talk tracks. Amotions AI calibrates to that language—typically about one week—so roleplay scores and live prompts reflect the firm’s process rather than a generic script library. Live prompts can show the exact wording from the uploaded talk tracks, follow rules such as not moving to the pitch until set discovery items are answered, and do not ask for information the buyer already gave. Managers can review whether approved prompts were followed on real calls after the fact. The material stays in your organization’s workspace and is applied each time the coach reads a conversation.

What happens to call audio during a live coaching session?

The desktop app or browser overlay listens on the user’s computer. Audio and context travel over encrypted connections to isolated Google Cloud workloads, including Cloud Run, where speech and language models interpret the conversation. The next best action returns only to the user’s overlay.

Which AI models does Amotions use?

Amotions is built on speech-recognition and large language models, including third-party foundation models. What Amotions builds is the layer around them: reading the six signals together, grounding every read in the customer’s own playbook and history, deciding on one action in real time, and learning from the outcome. Customer conversations are never used to train third-party foundation models.

Can another company’s coach see our playbook or calls?

No. Organization playbooks, transcripts, and uploads are separated by customer workspace, and each organization’s knowledge stays isolated to it. Data is encrypted in transmission and in storage.

Does Amotions have SOC 2 or ISO 27001?

Amotions does not publish a SOC 2 report or ISO 27001 certificate on its site. If your review requires one, ask for current status during your security review.

How do we run a security and IP review of Amotions?

Book a demo at https://amotionsinc.com/pricing/contact and say you need a security review. Bring your questionnaire and any data-use, retention, or contract requirements. The team covers current technical controls, data handling, and deployment for your organization.

Written and reviewed by

Pianpian Xu Guthrie

Founder and CEO, Amotions AI

M.Phil. / M.A. in Sociology, Yale University (thesis on social mobility and psychological well-being). Pianpian Xu Guthrie founded Amotions to coach the behavioral side of live conversations: tone, emotion shifts, and listening.

About the founder

Bring your security questions to the demo

Walk through training, the live-call data path, and your review requirements with the team.

Never joins your calls as a bot · Start free on Basic