Skip to content
GitHub
Start a 14-day trial

What atrim.ai does

You point your existing OpenTelemetry at atrim.ai and the analysis starts on its own — no dashboards to build first, no thresholds to configure, nothing to install in your cluster. This page is the map of what you get, organised by the four entry points in the product’s navigation.

Intelligence — what to look at, ranked for you

Section titled “Intelligence — what to look at, ranked for you”
Critical Paths The request paths through your system, ranked by where time and errors actually accumulate, with on-demand analysis of any one of them and the evidence one click away.
Analyze One click on a trace, a path or an incident, and I read the request for you: where its time and errors went, and which service a failure started in, with the evidence.
Copy for agent The view you’re on — evidence, findings and a concrete ask — as text you paste into your own coding agent.
Metric Signals The metrics that moved, ranked statistically against each service’s own history — a workload overview, a ranked concern list, and the arithmetic behind every rank on screen.

Exploration — the raw data, one query shell

Section titled “Exploration — the raw data, one query shell”
Service Topology The full-system service graph, with rate, error and duration on every node and edge.
Traces Individual traces and waterfalls with the critical path overlaid, searchable and filterable.
Logs Log search and pattern grouping, with an agent that reads a window and writes up what it finds.
Metric Explorer Ad-hoc metric queries straight over your raw OTLP metrics, with exemplars back to the traces behind a spike.
Ask in plain English Describe what you want in a sentence; I turn it into filters you can see and edit, and run the query.
Compare traces A slow or failing request beside a healthy one, with every operation that was added, went missing or slowed down marked.
Find in trace Narrow a long trace to the spans you care about and step through every match.

Incidents — detect, notify, resolve, on its own

Section titled “Incidents — detect, notify, resolve, on its own”

A condition in your telemetry opens an incident, a message goes to your sign-up email and/or a webhook you created, and the incident closes itself when the condition clears. It is a complete loop, on by default, with detection that runs without any thresholds for you to set up first.

Webhooks are created, edited and test-fired in-product, filtered by severity, and a Slack or Discord webhook URL is recognised and formatted automatically.

Admin & Governance — your workspace, your control

Section titled “Admin & Governance — your workspace, your control”
Ingest token and endpoint Your OTLP endpoint and an ingest token, ready to copy from the app.
Sign-up A magic link to your email, then your workspace.
Your team Invite teammates by email and give each an owner, admin or member role. New people are set up on the spot; existing atrim.ai users join with one click.
Alerting and Notifications Tune what fires and how loud, and choose where it goes — webhooks (Slack and Discord formatted automatically) and email.
Usage Your metered usage this hour and month-to-date, always on and read-only.
Training Data By default your workspace contributes masked telemetry examples that improve static detection rules, never to train models. Masking doesn’t guarantee anonymity; an owner or admin can opt out at any time.
Reset Tenant Data A one-action wipe of telemetry and analysis that keeps your settings, ingest tokens and webhooks — handy right after onboarding with a test app.

These docs are published as llms.txt and llms-full.txt so a coding agent can read them directly, and the onboarding flow hands you a short prompt to paste into whichever agent you already use.