Getting started

atrim.ai turns your existing OpenTelemetry into topology, ranked critical paths, and automatic analysis — you point your data at it and the work starts on its own. This is the whole path from the landing page to your first spans arriving.

The path

1. Land and pick a route. Open atrim.ai. Two ways in, each one click:

   • Connect my app — you already emit OpenTelemetry and want your own services in.
   • Explore the demo — nothing to connect yet, so start on a live OpenTelemetry demo with incidents firing throughout the day and see what atrim.ai does with it.

2. Sign in. A magic link to your email, or GitHub — one step, and your workspace is ready.

3. Copy your config. You get an API key and the OTLP endpoint, ready to paste — the key is what routes your telemetry to your workspace. The App route needs no code change; the Collector route adds an exporter beside the ones you already run.

4. Watch it arrive. The app watches for your first spans and tells you what it sees — the service name, the span count, and which signal arrived. If nothing shows up, the same screen links to troubleshooting.

5. Explore. Once data is arriving, start at Critical Paths — the ranked view of where time and errors actually accumulate — or open Service Topology for the map.

Copy the config

The App route needs no SDK and no code change — four standard OpenTelemetry environment variables:

Environment variables

OTEL_EXPORTER_OTLP_ENDPOINT=https://otlp.atrim.ai
OTEL_EXPORTER_OTLP_PROTOCOL=http/protobuf
OTEL_EXPORTER_OTLP_HEADERS=x-api-key=atrim_key_YOUR_KEY_HERE
OTEL_SERVICE_NAME=your-service-name

The Collector route adds an exporter rather than replacing one, so whatever you already send telemetry to keeps receiving it. Both snippets, and the full per-language reference, are on Connect your app.

Using a coding agent?

The onboarding flow hands you a short prompt — your credentials plus a pointer. Your agent already knows how to wire OpenTelemetry, and the depth it might want lives at https://atrim.ai/llms.txt, which it can fetch. The prompt is editable in the UI before you copy it.

Where to look next

atrim.ai groups its surfaces into four entry points. Once your first traces land:

• Intelligence — start here. Critical Paths ranks the request paths through your system by where the time and the errors actually are. Metric Signals surfaces the metrics that moved.
• Exploration — Service Topology for the map, and the Traces, Logs & Metrics explorers for individual requests, lines and ad-hoc metric queries.
• Incidents — automatic detection with a full lifecycle: an incident that opens on a real condition and closes itself when the condition clears.
• Admin & Governance — your API key, alerting webhooks, appearance, and account.

Connect your appEndpoint, auth header, per-language setup, collector config, and troubleshooting.

Try the OTel DemoA realistic microservice graph without instrumenting anything of your own.