Find production issues before your users do.

Polylane learns what normal looks like for each of your services and spots the problems nobody set an alert for, so you hear about them first.

Get started for free

No credit card required.

Hear about a slowdown before anyone files a ticket.

Polylane learns how each service normally behaves and opens an issue when something drifts. No thresholds to set, and a slow Monday morning won't page you.

checkout · p99 latency
checkout-edge · last hour
3.8s usual 210ms
0 1s 2s 3s 4s 5s
19:3019:4520:0020:1520:30
deploy 9f3c2a1 · 20:14
Checkout p99 up 18x since 20:14
checkout-edge · Cloudflare Worker
Critical

Polylane writes monitors for your routes and jobs from your code.

Polylane reads your code and sets up monitors for the routes and jobs that matter, using the logs and metrics you already send.

src/payments/stripe.ts
41 const charge = await stripe.charges.create(params);
42 if (charge.status === "failed") {
43 metrics.increment("payment.failed", {
44 reason: charge.failure_code,
45 });
46 }
New signal: payment.failed per minute
Watches failures on the payments route, from line 43

Spot a security breach early.

Root logins, MFA turned off, a repository made public: Polylane flags risky changes in AWS and GitHub as they happen.

Root account signed in
prod account · from a new country
Critical 4m
Branch protection removed on main
coreplanelabs/payments
High 1h
MFA removed for deploy-bot
IAM user
High 6h

Fifty alerts about one outage become one issue.

Alerts from Datadog, Sentry, CloudWatch and Polylane's own checks are grouped by the problem behind them, so one outage pages you once.

Checkout p99 up 18x since 20:14
3 alerts grouped into this issue
Critical
checkout p99 over 1s
Datadog monitor · 20:15
TimeoutError in POST /checkout
Sentry · 283 events
5XXError alarm on checkout-alb
CloudWatch · 20:16

Forwarded alerts get checked, and false alarms get dismissed.

Forward alerts from your monitoring tools and Polylane checks each one, keeps digging into the real ones and dismisses the rest with a reason.

Issue · CPUUtilization high on batch-worker
CloudWatch 02:10

ALARM: CPUUtilization > 90% on batch-worker for 10 minutes.

Polylane 02:11
getMetricslistJobRuns

This is the nightly backfill, which runs at 02:00 and finishes by 02:40 every night. Nothing is waiting on it. Dismissed.

Dismissed
Scheduled backfill, expected
No incident

Outcomes

Real numbers from Polylane customers.

Common questions.

Do I need to set up alerts first?

No. Polylane watches your stack from the moment you connect it. Forward the alerts you already have and they join in.

Will it flood me with notifications?

No. Repeats fold into the open issue, and noise is dismissed before it reaches you.

Which alerts can I forward?

Alerts from Datadog, Sentry, CloudWatch and the other monitoring tools on the integrations page.

How quickly does Polylane notice a problem?

Forwarded alerts are picked up as they arrive. Polylane's own checks run most often on the resources everything else depends on.

Platform
Surfaces

Nobody should be on-call.