Agent Trainer
Agent Trainer is the WinnerWare training module for building repeatable coaching scenarios and letting agents practice them through chat or audio simulations.
Overview
From a business perspective, Agent Trainer helps teams standardize coaching scenarios, measure performance, and review transcripts in a consistent format. Administrators can manage stories, simulator options, and reports in one place.
The module is split into three feature areas:
| Feature | Purpose |
|---|---|
CloudSolutions.AgentTrainer | Core story authoring, simulator UI, reports, transcript handling, and settings. |
CloudSolutions.AgentTrainer.Storage | Recording storage, download endpoint, trimming settings, and the background cleanup task. |
CloudSolutions.AgentTrainer.Storage.Azure | Stores recordings in cloud storage when that option is enabled. |
Key Features
- Builds training scenarios as Interaction Story content items with reusable steps and classifications.
- Supports chat and audio simulations from both admin and frontend routes.
- Offers optional AI assist during simulations.
- Imports interaction stories from uploaded
.opusor.wavfiles. - Gives managers report screens for transcript and performance review.
- Supports recording retention and secure downloads when the storage feature is enabled.
How It Works
Build training content
Create story categories first and then build training stories for each scenario they want agents to practice. Each story is made up of guided steps, prompts, expected response patterns, and optional scoring rules.
Run practice sessions
Launch the simulator, choose the story they want to practice, pick Chat or Audio, and decide whether to turn on AI Assist. The system then walks the learner through the scenario as an interactive practice session.
Review results
After sessions are completed, managers can review transcripts, compare performance, and use the reporting screens to spot coaching needs, interruptions, and assisted versus unassisted outcomes.
Configuration
Enable the right feature set
Enable the core Agent Trainer feature for story authoring, simulation, reports, and transcript management.
Enable the storage features only when your team needs to keep and download audio recordings:
- Agent Trainer Storage for recording retention and download support
- Agent Trainer Storage - Azure when recordings should be stored in cloud storage
Site settings
The module adds an Agent Training settings group under Settings. The current settings page includes:
| Setting | Purpose |
|---|---|
| Title Template | Controls how simulator sessions are titled in the admin area. |
| Simulator Chat Deployment | Chooses the default AI option used for simulator responses. |
| Simulator Utility Deployment | Chooses the default AI option used for scoring and validation. |
| Retention Days | How long recordings are kept before the cleanup task permanently deletes them. |
These settings are intended for the administrators who manage training setup.
Storage options
If recordings are enabled, administrators should decide whether they want local storage or cloud storage before rollout. This choice mainly affects retention, governance, and how recordings are managed over time.
Import prerequisites
The import screen accepts .opus and .wav files up to 20 MB.
If the required audio services are not available, the import screen shows a warning instead of starting the import.
Access and operations
In practice, administrators manage configuration, transcript access, and report visibility, while agents mainly need access to the simulator experience itself.
Usage
Create and organize stories
Create Interaction Story Category items first, then create Interaction Story items for each coaching scenario. Each story can contain a bag of Interaction Story Step items, plus optional voice and scoring setup.
One simple example you could build is:
- Category: General Category
- Story: VPN support website access review
In this example, the story description explains that this is a short support call about a VPN issue and a blocked website request. The opening statement starts the practice with a clear expectation: the user will review the issue in 5 quick steps.
That example story could then walk the learner through a simple support flow:
- Describe the VPN issue
- Have you restarted the VPN client
- Which website do you need
- What access message do you see
- What approval do you need
This makes the story easy to build because each step only needs to capture one clear part of the conversation.
The content list below shows how categories and stories are managed in the WinnerWare admin.
Use the step editor to define:
- The prompt the simulated customer will use, such as Describe the VPN issue
- A pro tip that remains visible to the learner, such as reminding them to ask whether the VPN fails to connect or disconnects after sign-in
- Whether the response should be AI-generated or selected from predefined responses
- How the learner response should be validated so the step knows what counts as a successful answer
Using the VPN example, a good pattern is:
- Put the customer question in the main prompt field
- Use Pro Tip for the coaching reminder the learner should keep in view
- Use the response setup to decide whether the next reply should feel open-ended or tightly controlled
- Use validation to define what a correct answer should contain, such as naming the website, describing the error message, or confirming that the VPN client was restarted
The step editor below shows where those authoring options appear when a story step is being configured.
If the step needs formal scoring, authors can turn on Add scoring criteria and define the scoring rules and maximum points for that step.
Import a story from media
Users can use Content -> Import Interaction Story to upload audio and create a draft story. The title names the story, and the uploaded file provides the conversation that WinnerWare turns into a draft story structure.
Run simulations
Users can open the simulator, choose a story, select Chat or Audio, and decide whether to enable AI Assist.
The story choice decides which scenario is loaded, Chat or Audio decides how the practice session runs, and AI Assist controls whether extra guidance is available during the session.
Review transcripts and recordings
Users can open Simulator -> Manage Transcriptions to review stored training results. This is where teams can compare how the session went, review responses step by step, and download recordings when recording storage is enabled.
Operational Notes
- Use categories to keep story libraries organised by business topic, queue, or training program.
- Keep scoring criteria focused on behaviors that supervisors actually review in coaching sessions.