Cost: Reduce LLM Token Usage in Log/Event Analysis

[elliotxx]

What would you like to be added?

Implement frontend pre-processing for logs and events to extract key information before sending to LLM API, reducing token consumption and improving cost efficiency.

Why is this needed?

Currently, the log and event aggregators send full content to LLM APIs (despite length limits), causing unnecessary token consumption.

This improvement will:

1. Frontend Pre-processing:

• Extract key information from logs/events
• Filter out redundant data
• Identify critical patterns

2. Implementation Details:

• Add frontend utility functions for data extraction
• Implement pattern recognition for common log formats
• Create configurable filtering rules

3. Expected Benefits:

• Reduced token consumption
• Lower API costs
• Faster response times

This optimization will make our AI features more cost-effective while maintaining analysis quality.