Learned Logs - Pulze.ai
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:
- Having a conversation with an assistant
- Identifying valuable interactions
- Recording/saving the interaction as a learned log
- 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:
- Check search terms
- Verify logs were created
- Check space permissions
- Refresh the view
Can’t Create Logs
If unable to log interactions:
- Verify conversation permissions
- Check feature availability
- Confirm space access
- Review subscription tier