Overview
Request Logs capture a wealth of data about every AI request. Following these best practices ensures you're using that data effectively — catching issues before they impact users, optimizing costs, and maintaining a reliable AI-powered application.
Monitoring practices
Regular Log Review
Establish a routine for reviewing Request Logs to catch issues proactively. Set up daily reviews, error monitoring, and trend analysis to stay ahead of problems.
Using Filters Effectively
Master the filter system to cut through noise and focus on relevant requests. Learn when to use each filter mode, how to combine filters, and strategies for efficient log navigation.
Monitoring Failed Models
Track provider failure frequency to identify instability before it impacts users. Set up failover rate monitoring, analyze failure patterns, and make data-driven provider decisions.
Tracking Webhook Reliability
Monitor webhook delivery success rates to ensure your backend stays in sync. Catch endpoint issues early, optimize delivery performance, and use retry strategies effectively.
Organization practices
Separating Environments
Use seed_batch prefixes and filter modes to keep production, testing, and playground data cleanly separated. This prevents test noise from obscuring real issues and helps you focus on the right data.
Optimization practices
Token Usage Optimization
Review token usage patterns to identify cost-saving opportunities. Learn techniques for reducing prompt tokens, managing conversation history, and choosing cost-effective models without sacrificing quality.
Duration & Performance
Use duration data to monitor request latency, identify slow providers, and optimize for speed. Set performance baselines, track trends, and balance speed with cost and quality.
Debugging practices
Request & Response Inspection
Master request and response body inspection for effective debugging. Learn what to look for in payloads, how to compare requests, and techniques for reproducing and resolving issues.
Using Timeline Context
Leverage the timeline view for complete request lifecycle understanding. Go beyond the final result to see failover chains, webhook flows, and callback completion for the full picture.
Quick reference
| Practice | Key benefit |
|---|---|
| Regular Log Review | Catch issues proactively |
| Using Filters | Cut through noise |
| Monitoring Failed Models | Provider stability |
| Tracking Webhooks | Backend reliability |
| Separating Environments | Clean data |
| Token Optimization | Cost savings |
| Duration & Performance | Speed optimization |
| Request Inspection | Effective debugging |
| Timeline Context | Complete understanding |
Next steps
- Use Cases & Scenarios — Practical scenario guides
- Timeline Components — Deep dive into timeline details
- Back to Observability — Return to the overview