Detect issues before your customers do
Honeycomb Anomaly Detection learns what normal looks like for every service, then tells you the moment it changes—no thresholds to set, no metrics to guess at.
Why you'll love Honeycomb
Zero config, real coverage
Anomaly Detection automatically finds which of your services have enough steady traffic to model reliably, onboards them, and starts learning. There's nothing to threshold, tune, or name in advance.
Detection that starts the investigation
The moment a service behaves differently than its own history says it should, Honeycomb kicks off an auto-investigation in Canvas, running queries and testing hypotheses before you've even logged on.
Built on the data you already send
Runs on the same high-cardinality, OpenTelemetry event data flowing into Honeycomb today. No new agent, no separate pipeline, noor re-instrumentation.
Feature highlights
Baselines that build themselves
Anomaly Detection establishes a statistical baseline of normal behavior for every eligible service, then flags real deviations from it. It covers two signals,: error rate and presence, —so you know the moment a service starts failing more than usual, or stops reporting altogether.
Get pulled in before customers notice
When something changes, you're notified in Slack or the channel you already watch. Honeycomb also kicks off an investigation in Canvas automatically, so the "why" is already being answered by the time you open it.
Coverage you can actually trust
Anomaly Detection only models services with enough steady traffic to baseline reliably. You can see which services are covered, which aren't yet, and why—not a wall of alerts you learn to ignore.
No new pipeline or re-instrumentation
Anomaly Detection runs on the OpenTelemetry data you're already sending to Honeycomb. It's also available through our API and MCP server, so you can pull anomaly status into the workflows and agents you're already running.
I didn't know what good observability looked like until we adopted Honeycomb. It makes our jobs easier—fast insights, no fighting the tool—and it gives us the power to detect issues before our customers do.
Brenden Petracek
Observability Team Lead, iQmetrix
Related features
Bringing your data into Honeycomb is just the start. Explore how Honeycomb's Anomaly Detection connects to the rest of the platform.
Want to know more?
Talk to our team to arrange a custom demo or for help finding the right plan.