Training Panel 101
The Training Panel gives you full control over how your AI agents are trained, tested, and improved. With real-time feedback, version snapshots, and detailed activity tracking, you can manage the entire training workflow in one place.
This update is available across all AI agents in your Quack workspace.
Train with Confidence and Clarity
The Training Panel is designed to help you move fast and improve with precision. You can view feedback by topic, see what changed and why, and know exactly what is live and what is still in progress. Everything is organized in one structured view so your team can work more efficiently and deliver smarter AI across every channel.
Build a Network of Smarter Agents
Quack lets you deploy specialized AI agents across your product, documentation, and internal tools all of which can be seen in your Training Center Dashboard.
Each agent is trained for a specific job but shares one training environment. Together, they create a seamless and intelligent support experience across your entire stack.
💡 Pro Tip: Use feedback from one agent to inform improvements across all others
The Training Panel gives you a full view of how each agent is performing, what needs attention, and what has changed. It includes three key areas: the left-side topic list, the center session view, and the right-side insights panel.
Here’s what you can see and do:
Real-Time Feedback Panel (Right Side)
View counts for Good, Bad, and No Feedback
Monitor how each topic is performing in live conversations
Understand which areas need more training or review
See active instructions, brief content, and knowledge base toggles
Version Summary
View the current version number at the top of the panel
See a list of topics with New, Has Updates, or Live status
Identify content that is ready to be published
Quickly scan which topics were changed in the current version
Activity Page
Access a full timeline of edits, tests, and training actions
See who made each change and when
Track updates to instructions, briefs, and training responses
Filter by action type, topic, or feedback event
Topic Status Labels
New: The topic has been created but not yet trained
Has Updates: Edits were made that are not yet released
Live: The topic is part of the current live version
Pending Review: A training log was created but not yet reviewed
💡 Pro Tip: Use filters and version tags to quickly find which topics need your attention before releasing updates
This overview gives your team the context needed to train effectively, respond to real feedback, and ship improvements with confidence.
Step-by-Step: Train and Improve Your Agent
Create Your Agent
Go to Training Center Dashboard
Click + New AI Agent and choose a channel
Your agent appears in the list, ready to set up
Add Topics and Instructions
Select your agent from the list
Add or update topics, briefs, and instructions
Test by asking a real question in the Training Panel
Provide feedback with thumbs up or down to record results
Track Changes and Activity
Click the Activity tab to view all edits, tests, and updates
Filter by topic, feedback type, or training action
See who made each change and when
💡 Pro Tip: Use the Has Updates label to find topics that need to be reviewed before publishing
Monitor Feedback in Real Time
Select any topic from the left panel
View current feedback counts and instruction details in the right panel
Track how each version performs based on actual user input
Search and Filter Customer Tickets
Now you can search, filter, and test with real customer tickets right from the left panel in the training area. This makes it easier to train Quack AI using real-world examples.A Simpler Way to Improve Results
We’ve unified the feedback and improvement process. The new, AI-powered experience helps you review and optimize answers all in one place.
💡 Pro Tip: Training with actual tickets gives Quack AI a better understanding of what your customers are really asking.
Compare and Release Versions
Open the Version History tab
Compare two versions side by side
Review what changed and how feedback shifted
Go to Release Manager and click Release to push updates live
A new version is created and feedback counters reset
Prerequisites in the Training Panel
Set up prerequisites in the Training Panel to make sure Quack AI asks critical questions in the right order—before moving forward with user interactions.
What are prerequisites?
Prerequisites are requirements you define for Quack’s conversations. They help Quack check for key information before giving more details or instructions.
For example:
Set a prerequisite so Quack always asks, “What’s the purpose of this meeting?” before providing agenda options.
How to set prerequisites
Open the Training Panel from your dashboard.
Select the conversation or topic you want to edit.
Click Add Prerequisite.
Enter your requirement—for example, “Confirm meeting purpose.”
Click Improve: Quack AI will optimize the prompt.
Click Submit to activate the prerequisite.
Your changes apply instantly. Quack will now check for the prerequisite every time before continuing with the rest of the training flow.
💡 Pro Tip: Use prerequisites to keep conversations focused and avoid missing critical details.
Key Details and Definitions
What Does “Pending Review” Mean?
This appears when a training log exists but has not been reviewed or tested yet. You’ll see a badge on the topic list and details inside the training view.
How Is Feedback Tracked?
Each thumbs up or down is logged by version. You can measure how changes affect performance over time.
Where Can I See What Changed?
Open the Version Details tab to view updated topics, instructions, and feedback trends. Use the New and Updated labels to review quickly.
The Training Panel helps you create, test, and improve AI agents with structure and speed. Whether you're managing a single agent or a network across products and teams, you now have a complete training system that supports fast iteration and better performance.
“AI Response Steps” -Chain of Thought
Quack gives managers deeper visibility into how each AI-generated response is created. The AI Response Steps and Chain of Thought features make it easier to understand, monitor, and improve how Quack processes requests.
AI Response Steps
The AI Response Steps button appears above every generated response in Quack. When you click it, you’ll see a detailed breakdown of how Quack built its answer.
This includes:
Available Skills: Lists the active skills Quack can use for the task.
Prerequisite Checks and Results: Shows which prerequisites were tested and whether they passed or failed.
Language Detection: Identifies the language of the input to ensure accurate responses.
Knowledge Sources: Displays the internal or external sources Quack searched to generate its answer.
Strict Mode Check: Indicates if strict mode is on and whether it affected the output.
💡 Pro Tip: Use these views when debugging or optimizing Quack’s setup. They help you understand what happened behind the scenes and pinpoint where to fine-tune your briefs, tone of voice, or topic descriptions for better results.