Telemetry
Telemetry
Telemetry gives inside-out visibility into applications and services so teams can debug failures that synthetic checks only detect from the outside.
Supported telemetry data
- Logs: structured events and error records.
- Metrics: numeric measurements such as latency, error counts, throughput, and cost.
- Traces: request or workflow traces with spans, duration, status, and service context.
- Agent telemetry: AI-agent run records, tool spans, tokens, cost, duration, and errors.
Setup
- Create an API key from the API keys page.
- Use the key as a bearer token when sending telemetry.
- Send OTLP JSON to the telemetry ingestion endpoint.
- Use the telemetry setup page for examples and endpoint details.
- Keep keys scoped to the organization that owns the telemetry.
How telemetry is used
- Telemetry is shown in dashboards for logs, metrics, traces, and trace detail.
- Trace detail pages help locate slow or failing spans.
- SLO burn-rate alerts can use telemetry to open service incidents.
- Agent telemetry powers agent health, cost, latency, and error-rate views.
Troubleshooting
- If telemetry does not appear, confirm the Authorization header uses a valid non-revoked API key.
- Confirm timestamps are valid and payloads use the expected JSON shape.
- Confirm the API key belongs to the same organization where you expect data to appear.
- Use API key last-used timestamps to verify ingestion authentication is succeeding.
Related documentation
API keys
Create and revoke telemetry ingestion keys for SDK and OTLP-style data.
AI agent monitoring
Monitor AI-agent quality, latency, tool spans, token usage, cost, and failure rate.
Services and SLOs
Model owned services, link monitors, define dependencies, and track service-level objectives.
Audit log
Review access, configuration, API key, incident, review, and team-management events.