System status
/health endpoint from your browser when the page loads. It confirms the API answers; it does not exercise the neural engine. We do not run a public uptime monitor or offer a public uptime SLA; availability is tracked internally with AWS CloudWatch alarms.Components
Monitor it yourself
The API serves an unauthenticated health endpoint you can poll from any monitoring tool (Datadog, BetterStack, Pingdom, Uptime Kuma):
{"ok":true,"service":"usezn-api","version":"0.1.0"}Incidents
From 12:04 UTC on 9 October to about 11:30 UTC on 10 October, the cloud detection API (POST /v30/analyze) ran without its neural classifier (at the time, Galvanize-60M). A deploy shipped without the classifier's runtime dependencies, so the API kept answering with its deterministic rules only. zn-gate on your machine was not affected: it does not depend on our cloud.
We restored the dependencies on 10 October. Deploys now check that the classifier loads in the new version before it receives traffic, and an alarm fires if it fails to load.
If you think an outage affected you, email hello@usezn.com and we will tell you what happened.