Skip to content

Observability & Performance Monitoring

The Activity Dashboard provides visibility into how your agents are being used within AI Refinery® Studio. Access it anytime from the left global navigation pane, under Workspaces and Library.

The dashboard allows you to track usage across the SDK and Studio, helping you better understand performance and efficiency.

The dashboard has three tabs — Token Usage, Performance, and Agents — giving you a layered view from high-level consumption down to individual agent behavior.

The Token Usage tab shows your total agent team count, workspace count, and members, along with a token usage breakdown by model and agent class.

Activity Dashboard Token Usage tab showing total agent teams, workspaces, members, and token usage by model and agent class

The Performance tab surfaces latency trends, request rates over time, and dependency call volumes — giving you instant visibility into where slowdowns may be occurring.

Activity Dashboard Performance tab showing latency trends, request rates, and dependency call volumes

The Agents tab breaks down each agent's success rate, orchestration time, runtime, collaboration patterns, and token cost — helping you pinpoint which agents are driving performance or cost.

Activity Dashboard Agents tab showing per-agent success rate, orchestration time, runtime, and token costs

Filtering

As a member, metrics are aggregated across all agent teams you own and can be filtered by:

  • Workspace
  • Team
  • Time range

Token Usage

Optimize cost and performance with clear, aggregated insights that show exactly which agents and models are driving token consumption. Toggle usage over time and in aggregate by model or by agent class.

Performance Metrics

Monitor performance with instant visibility into:

  • Latency — Response time per request
  • Request rates — Volume of requests over time
  • Dependency calls — Calls made to downstream services

These metrics allow quick diagnosis of bottlenecks to maintain consistent system performance.

Agent Activity

Track aggregate agent effectiveness including:

  • Success rate — Percentage of requests completed successfully
  • Orchestration time — Time spent coordinating between agents
  • Agent runtime — Time each agent spent executing

Drill into the drivers to understand:

  • Which agents did the work
  • How long each agent took
  • How agents collaborated with each other
  • What the total token cost was

This enables smarter tuning and scaling of your agent teams.

Programmatic access

To query observability data — logs, metrics, and traces — from your own tools or pipelines, see the Observability API.