Stop writing alert rules. Agents read the telemetry and decide.
Join the waitlistCritical latency degradation in checkout-edge worker
Critical latency degradation detected in checkout-edge worker: 18x+ P99 latency spikes sustained for 12 minutes
edge-gateway Cloudflare Worker ··· checkout-edge Cloudflare Worker ··· 
hd-prod Hyperdrive ··· 
payments-db PlanetScale ··· cart-svc Cloudflare Worker ··· A spike isn't an incident. Polylane knows the difference.
-
Your charts become checks
Anything your team already charts gets watched: save a query or build a dashboard and Polylane picks it up.
-
Baselines that know your traffic
Normal at 2pm on a Tuesday is not normal at 2am on a Sunday. Polylane compares against your rhythm, not a flat line.
-
Blips get ignored
A move has to be big, sustained, and in the direction that's worse. Isolated spikes are rejected.
-
An agent makes the call
It looks at the whole picture, including what changed recently. The default answer is that nothing is wrong.
-
Evidence attached
When something is wrong, you get the exact queries behind the finding, not a graph and good luck.
Some problems never spike. Polylane reads for those too.
A steady stream of crashes looks flat on every chart, so agents also read the logs themselves. And some states are wrong at any baseline: a CrashLoopBackOff doesn't need judgement, a plain threshold catches it instantly.