Follow the full workflow
Bring model and tool activity together to understand the steps behind an agent response.
Agent observability
See what happens between a request and a result.
Follow model and tool activity across agent workflows. Understand latency, reliability, and usage in one place, then see where to improve.
Workflow trace
Illustrative activity view
Total latency
1.42s
Across 3 workflow steps
Cache activity
1hit
In this example trace
Step latencySample view
Agent request
Input received
Model call
gpt-4.1 · completion
Search tools
2 calls · 1 cached
Make every step clearer
Understand the work behind every answer.
A joined-up view helps teams identify what is slow, what is repeated, and where to focus next.
Find performance bottlenecks
See where latency, errors, and cache behavior affect the experience.
Understand usage and cost
Connect usage to the actions that drive it and spot repeated work worth optimizing.
Observability and caching work together: find repeated work, then decide where reuse makes sense.
Explore caching