Core Concepts

Warpmetrics models your AI agent workflows as a hierarchy of primitives. Understanding these concepts will help you get the most out of the platform.

The hierarchy

Every agent execution is modeled as a tree. Runs contain groups and calls. Groups contain calls and other groups. Outcomes can be attached at any level. Acts connect one run to the next.

Run
├── Group (planning phase)
│   ├── Call (gpt-4o)
│   └── Call (gpt-4o)
├── Group (execution phase)
│   ├── Group (sub-task)
│   │   └── Call (gpt-4o-mini)
│   └── Call (gpt-4o)
├── Outcome ("failed")
└── Act ("retry")
    └── Run (follow-up attempt)

Concepts

Runs

The top-level unit of work. One run per agent execution — a support ticket, a code review, a data pipeline job.

Groups

Logical phases within a run. Organize calls into steps like planning, execution, and review. Supports nesting.

Calls

Individual LLM API invocations. Automatically captured with full context: messages, response, tokens, cost, latency.

Outcomes

Named results on any entity. The bridge between 'the agent ran' and 'it worked.' Drives success rate tracking.

Acts & Feedback Loops

Actions taken after outcomes. Connect retries, refinements, and iterations into traceable chains across runs.

Quick reference

run(label, opts?)Create a top-level runwm_run_

group(target, label, opts?)Create a group inside a run or groupwm_grp_

call(target, response, opts?)Link an LLM call to a run or group—

outcome(target, name, opts?)Record a result on any entitywm_oc_

act(outcome, name, opts?)Record an action after an outcomewm_act_

ref(target)Get the tracking ID for any entity—

flush()Manually send pending events—