ScoutML

ScoutML

ScoutML is the sessions platform for agent work. Log tool use, code execution, notes, and outcomes in one place so teams can inspect how runs actually unfolded.

Request Access Browse Sessions Read the Docs
Coverage Action-by-action history

Capture searches, code runs, tool calls, notes, failures, and recoveries in one session timeline.

Telemetry Decorator-first tracking

Add lightweight decorators to your tools and send each action directly to the ScoutML session view.

Use Case Operational visibility

See what agents actually did, where they stalled, and how behavior changes across repeated runs.

// Core Product

What the core platform does

The primary product is built around the artifacts you need to evaluate agent work directly: sessions, metrics, timelines, and operational evidence.

Trace every action in one session

See each code execution, tool call, search, note, and failure in a single chronological view instead of scattered logs.

code search tool

Compare how runs actually behave

Measure differences in pace, intervention rate, completion, and repeated failure patterns across agents and tasks.

timelines metrics comparisons

Keep evidence you can audit later

Export session artifacts for retrospectives, reviews, incident analysis, and team reporting without rebuilding the context from scratch.

exports audit reporting
// Platform

What teams actually measure

The main application is the telemetry layer for agent work: sessions, metrics, timelines, exports, and operational evidence.

Session tracking Store each run with action history, timestamps, payloads, and completion state.
Metrics and analytics Summarize action counts, compare runs, and inspect how behavior changes over time.
Exports Preserve traces for offline analysis, audit trails, retrospectives, and operational review.
Developer APIs Authenticated endpoints for starting sessions, logging actions, and closing runs cleanly.
For technical teams

Understand process, not just output

Measure which failure patterns repeat, which tools get overused, and how quickly agents converge on workable implementation paths.

  • Compare session timelines side by side with structured metrics.
  • Track efficiency, quality, resource usage, and intervention points.
  • Preserve session timelines for later analysis and review.
For governance and safety

Build empirical footing for capability claims

Capability discussions need artifacts, traces, and longitudinal measurements. The platform is designed to make those records inspectable rather than anecdotal.

  • Observe repeated patterns in execution, tool use, escalation, and recovery.
  • Collect comparable metrics across agents, tasks, and time windows.
  • Ground policy and deployment decisions in operational evidence.
Get Started

Start with sessions and metrics

The core workflow is to log an agent run, inspect the resulting timeline, and compare its metrics against prior work. ScoutML is now scoped to that telemetry and evaluation loop.

Browse Sessions View Metrics Read the Docs