What shall we analyze?

Recent Charts
Autonomy
Evals Last 30 days
Sandbox

Autonomy runs experiments here as it investigates. You can also start one manually.

Datasets

Static collections of (input, output, eval) rows. Used for golden sets, calibration, annotation, and sandbox runs. Autonomy builds these as it works; you can also create one manually.

Loading…
Vocabulary

Help the analyst understand your agent's structure and goals

Strategy

Describe your agent's goals and what you're optimizing for

Terminology

Define business terms, KPIs, and domain-specific concepts

Data Structure

Span hierarchy and metadata fields (auto-generated from your traces)

Agent Change Log

Record dated changes to your agent (model swaps, prompt updates, deploys) so analysis can tie trends to causes