Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.quiva.ai/llms.txt

Use this file to discover all available pages before exploring further.

Help your assistants learn and improve by rating messages or entire conversations. Assistant Learning captures your feedback on assistant responses, consolidates insights using AI, and uses those learnings to build a knowledge source to enhance future assistant performance.
Assistant Learning works best when users consistently rate conversations. The more feedback you provide, the better your assistants become at handling similar interactions in future conversations.

​
Key Features at a Glance

  • Rate Conversations — Quickly thumbs-up or thumbs-down assistant responses and conversations to provide feedback
  • View Learning Insights — See consolidated feedback and insights in your assistant’s Learning tab
  • Automatic Consolidation — AI-powered system analyses feedback patterns and generates actionable insights
  • Contextual Learning — Assistants automatically use consolidated learnings in future conversations when enabled
Learning is per assistant, and any ratings you give will improve only your assistant responses in your account and do not inform training data.

​
Getting Started with Assistant Learning

​
Rate Your First Conversation

Rating conversations is the simplest way to help your assistants learn. You’ll see rating options right in the chat interface after each assistant response.
  1. During or after a conversation with an assistant, look for the rating buttons below the assistant’s message
  2. Click the thumbs up (👍) to indicate a good response, or thumbs down (👎) for a response that could improve
  3. To rate the entire conversation, use the ‘Rate this conversation’ option at the bottom of the chat
  4. Select your rating—the system captures your feedback immediately
  5. View your learning by navigating to the “Learning” tab by clicking on the assistant settings. You can review and edit the learnings manually, as well as “consolidate” any pending feedback
Assistant message with thumbs up and thumbs down rating buttons visible below the response
The feedback you provide is stored securely as knowledge in your own account and used to help your assistants improve over time.

​
Understanding Assistant Learning Insights

​
Viewing Consolidated Learning

Each assistant has a Learning tab where you can see patterns and insights generated from all user feedback.
  1. Navigate to your assistant’s settings or details page
  2. Select the Learning tab to view feedback summary and trigger consolidation of new feedback into learnings
  3. Review consolidated insights—AI-generated summaries of what’s working well and areas for improvement
  4. These insights are used automatically as a knowledge source to refine assistant instructions, behaviour, or scope
Assistant Learning tab showing consolidated feedback insights and user ratings summary

​
How Consolidation Works

When feedback is collected, the system uses an AI model to analyse patterns and generate high-level insights. This consolidation process:
  • Identifies common themes in user feedback
  • Summarises what users consistently rate positively or negatively
  • Generates actionable observations about assistant performance
  • Provides context for assistant behaviour refinement
You can review consolidated feedback anytime and use these insights to improve your assistant’s prompts, instructions, or capabilities.

​
How Assistants Use Learning

​
Automatic Learning Context

When Assistant Learning is triggered for an assistant, the consolidation insights are automatically included as additional context for the assistant during new conversations. This means:
  • Assistants can reference previous learnings to improve responses
  • Insights about common user preferences inform assistant behaviour
  • Assistants become more aligned with user expectations over time
  • The learning loop continuously strengthens assistant performance
To trigger the learning file generation, the “Consolidate” button needs to be clicked on the learning tab which will trigger an updated using all the votes that have been cast.

​
Common Questions About Assistant Learning

Assistant Learning works by rating conversations and triggering the consolidation process. Consolidated insights become available shortly after and are immediately updated for you to review and edit.
Yes. In the Learning tab, you can view all collected feedback and edit the consolidation notes. This is useful if you want to refine the insights or remove feedback that’s no longer relevant.
Assistant learning needs to be triggered by voting, and then running the consolidation of votes. Not using this or removing the contents of the file will stop the assistant using the context.
Monitor your assistant’s performance in real conversations and compare responses before and after feedback consolidation. Look for assistants responding more accurately to common scenarios, handling edge cases better, and aligning more closely with your preferences.
All feedback is stored securely within your workspace. Only users with access to the assistant can view learning insights. The consolidation process is performed server-side and follows your organisation’s security and privacy policies.
Yes. Different users can rate the same conversation independently. All ratings are captured and contribute to the consolidated learning insights.
The learning forms part of the assistant configuration. If it is deleted, the learning file will also be deleted.

​
Best Practices for Assistant Learning

​
Provide Consistent Feedback

Rate conversations regularly to build a comprehensive feedback library. The more data you provide:
  • The better the AI consolidation becomes
  • The more accurate the learning insights
  • The faster your assistants improve

​
Be Specific About Why You’re Rating

When possible, provide context for your rating. If you’re rating a response negatively, consider what could improve. This helps the consolidation process understand root causes. The best way to provide this feedback is to correct the assistant in the chat, and then vote positively for the corrected response.

​
Use Learning Insights to Refine Assistants

Don’t just collect feedback—act on it. Use the consolidated insights to adjust:
  • Assistant instructions and system prompts
  • Assistant capabilities and scope
  • Response guidelines and tone You can always copy feedback from the learning file and incorporate it into your assistant instructions directly.

​
Review Learning After Consolidation

Check your assistant’s Learning output to ensure it is accruate once it is consolidated. This helps your assistants perform to your expectations.

​
Next Steps

Consolidate Something

Use past conversations to add votes and create your first learning file

Monitor Learning Performance

Keep an eye on the performance of your learnings and how they are impacting the assistant behaviour
Assistant Learning transforms user feedback into continuous assistant improvement. Start rating conversations today to help your assistants learn from real interactions and deliver better results tomorrow.