Skip to main content

Overview

Learned Logs are created when you record meaningful interactions from conversations within your space. These logs serve as training data and reference material for improving AI assistant performance and creating evaluation datasets.

Viewing Learned Logs

In the space sidebar panel:
  • Learned Logs section with search functionality
  • Shows “No learned logs yet” when empty
  • Search bar to find specific logs
  • Description: “Learned logs are created when you record meaningful interactions in conversations”

Creating Learned Logs

From Conversations

Learned logs are created by:
  1. Having a conversation with an assistant
  2. Identifying valuable interactions
  3. Recording/saving the interaction as a learned log
  4. Log becomes searchable and reusable

What to Log

Record interactions that:
  • Demonstrate correct responses
  • Show desired behavior
  • Provide good examples
  • Represent edge cases
  • Illustrate best practices

Using Learned Logs

For Training

Learned logs can be used to:
  • Improve assistant performance
  • Fine-tune responses
  • Establish response patterns
  • Create consistent behavior

For Evaluation

Logs serve as:
  • Test cases for evaluations
  • Benchmark examples
  • Quality standards
  • Performance metrics

For Reference

Team members use logs for:
  • Understanding past interactions
  • Learning from examples
  • Reusing successful patterns
  • Maintaining consistency

Search Functionality

Find learned logs by:
  • Keyword search - Search log content
  • Interaction type - Filter by category
  • Date range - Find logs from specific periods
  • Assistant - Logs from specific assistants

Best Practices

Record Quality: Log only high-quality, representative interactions.
Organized Search: Use consistent terminology for easier searching.
Regular Review: Periodically review logs to identify patterns and improvements.
Privacy Awareness: Don’t log sensitive or confidential information.

Integration

With Conversations

Learned logs connect to conversations:
  • Created from conversation interactions
  • Link back to original conversation
  • Preserve context and history
  • Enable follow-up reference

With Evaluations

Logs feed into evaluation systems:
  • Used as test datasets
  • Benchmark performance
  • Track improvements
  • Validate changes

With Assistants

Logs help improve assistants:
  • Training examples
  • Behavior patterns
  • Response quality
  • Consistency standards

Empty State

When no logs exist yet:
  • Explanation message displayed
  • Instructions to start recording
  • Information about purpose
  • Encouragement to begin logging

Common Use Cases

Customer Support

  • Log successful resolutions
  • Record helpful responses
  • Build knowledge base
  • Train support assistants

Technical Assistance

  • Document code solutions
  • Save debugging processes
  • Record technical explanations
  • Build technical knowledge

Content Creation

  • Save effective prompts
  • Record successful outputs
  • Build style guides
  • Maintain brand consistency

Data Analysis

  • Log analytical insights
  • Record interpretation methods
  • Save successful queries
  • Build analysis patterns

Managing Logs

Viewing Details

Access log information:
  • Original conversation context
  • Timestamp and creator
  • Associated assistant
  • Interaction outcome

Organization

Keep logs organized by:
  • Consistent naming
  • Clear descriptions
  • Relevant tagging
  • Regular cleanup

Privacy and Security

Considerations for logged data:
  • Remove sensitive information
  • Respect privacy guidelines
  • Follow data policies
  • Maintain confidentiality

Troubleshooting

Can’t Find Logs

If logs aren’t appearing:
  1. Check search terms
  2. Verify logs were created
  3. Check space permissions
  4. Refresh the view

Can’t Create Logs

If unable to log interactions:
  1. Verify conversation permissions
  2. Check feature availability
  3. Confirm space access
  4. Review subscription tier
I