Agent analytics
Operational rollups for your agents: runs, errors, cost, latency, run status, failing tools, and per-agent volume.
Alongside the four report types, Trodo rolls your agent runs and spans into ready-made operational views. These answer "how are my agents doing right now" without building a report.
Key metrics
The agent overview surfaces:
- Total runs, error rate, total cost, and number of agents.
- Latency over the recent run sample: p50 and p95 duration, tokens in, and tokens out.
- Run status breakdown: success, failed, running.
- Failed tools: the tool calls that error most.
- Agents by volume: a table of each agent with its runs, errors, cost, and last-seen time.
- Recent runs: the latest runs, each linking to its full trace.
Histograms and conversations
- A run histogram buckets runs over time (total, errors, ok) so you can spot spikes.
- A conversations rollup groups runs by conversation, with run count, last activity, and the agents involved.
Evaluation scores over time
If you run Evaluations, each evaluator has a time-series of passed, failed, and errored results, plus pass rate, latency percentiles, and cost. See Evaluations results.
Workspace home
Your home workspace is a dashboard of preset widgets (team pulse, agent operations, web activity, reliability and errors, tool performance, and more). Add your own board cards alongside them.
Next
- Observability for the trace data behind these rollups
- Evaluations to score quality