Foglamp
Trace AI-agent calls and track costs, latency, and response quality.
Retrace
Record agent execution, replay failures, and compare fixes before release.
Meet Retrace
Retrace records model calls, tool activity, costs, and errors from AI agents through Python and TypeScript SDKs. Developers can inspect an execution timeline, replay a recorded run, or branch from the step where behavior went wrong. Changing a prompt, input, or model then produces a new path that can be compared with the original.
The platform turns recorded failures into regression checks and supports evaluation gates in continuous integration. Runtime controls can limit budgets, repeated loops, or execution steps, while multi-agent views expose relationships between sessions. Its fix-verification workflow reports whether a change improved, worsened, or left the evaluated behavior unchanged.
Discussion
Tried it? Share a useful thought or ask a question.
Good conversations start with a hello.
Sign in to commentProduct details
- Website
- retraceai.tech
- Status
- Launched
- Category
- Developer tools
- Listed
Explore related topics
Similar products
Raindrop
Trace AI agent behavior, investigate recurring failures, and test changes.
Tessary
Find and investigate failures in production AI agents
Axiom
Collect, query, and monitor machine data in a managed event store.
Plural
Use infrastructure-aware AI agents for DevOps investigations and changes
HyperProbe
Investigate production incidents with read-only runtime probes