Monitoring built in your account. Yours to keep.
Most providers watch your cloud from their platform, and the tooling leaves when you do. We built the opposite. The monitoring we run for you is deployed into your own cloud account, as code, in a repository you own. This page explains exactly how it works.
Not our platform. Your stack.
The industry norm is monitoring that lives on the provider's side: their dashboards, their data, their logins. Leave, and you start over from nothing. We install your monitoring in your environment instead, from official packages at standard paths, readable and maintainable by any Linux administrator. If we ever part ways, you keep a working monitoring stack. That's not a concession. It's the design.
Your account. Your data. Your keys.
- Deployed in your cloud account, as code, in a repository you own
- Dashboards at your URL, behind your login. We link out to them, never sitting between you and your own data
- Only alert notifications reach us. Your metrics and your logs stay in your environment
- Our monitoring access can't read your application data, can't alter audit records, and can't widen its own permissions
- A strict touch policy on your servers: no firewall changes, no global configuration edits, nothing removed
- We never interfere with monitoring you already run
If you ever leave, offboarding removes our alert destination from your configuration, and that is the whole procedure. Your dashboards, your history, and your local alerting keep working without us. We've tested that, not just claimed it.
Rather not run another server? We'll host your dashboards in a private workspace on our side, walled off from every other customer. Same dashboards either way.
Eighty-seven ways to see trouble coming.
The stack watches your servers, databases, load balancers, network, storage, and backups with 87 distinct check types across 16 resource categories. Most checks run every three minutes; security and compliance posture is checked daily. Cloud resources are monitored without installing anything. For what can only be seen from inside a server, memory, disk, services, databases, a small monitoring stack from standard packages goes into your environment, and it's yours. And where something genuinely can't be seen without deeper access, we show you the gap instead of quietly leaving it off the list.
The 10-minute promise doesn't rest on one notification.
- If a critical alert goes unacknowledged, the system re-pages at five minutes and pulls in a second engineer at ten. The commitment survives a missed phone, a dead battery, or a bad night.
- Your monitoring sends an alive signal every 25 seconds, and we check for it every minute. A missed check is re-checked one minute later; a second miss alerts our staff. Whether your stack died or our path to it broke, we know within about two minutes and treat it as an emergency, not something to discover during your outage. Monitoring that dies quietly is worse than no monitoring, because it looks like calm.
- An alert that arrives outside your coverage window isn't dropped. It's recorded in full, and an engineer picks it up the moment your window opens.
The numbers behind it.
| What happens | How fast |
|---|---|
| A failed server becomes an open incident, full context attached | About 3.5 minutes |
| Your monitoring goes quiet and our staff is alerted | About 2 minutes |
| Your stack sends its alive signal | Every 25 seconds |
| A recovered resource clears its own incident | Within 5 minutes |
| Standard checks run | As often as every 3 minutes |
Measured in our own environment, August 2026. We'll publish customer numbers when we have them, not before.
What we never do.
- Cross the lines you draw. Your guardrails mark what we must not do on our own. Before anything past them, failing over a database, restoring from a backup, we call you first. They appear on the engineer's screen next to the alert, and a person reads them and decides.
- Run automated remediation against your production. Engineers act; software informs.
- Flood your inbox. Alerts go to our engineers, not to you. When something needs your attention, the first thing you read was written by the engineer handling it.
- Lock you in. Everything on this page is the opposite of lock-in, on purpose.
One stack, three ways to buy it.
The same stack sits under every service we offer. Cloud Monitoring Service: we watch and respond while you run everything else. Managed Cloud Services: we also run the operations you hand off. Managed DevOps Services: we build pipelines and infrastructure as code, and run what ships.
See it in your account, not a sales deck.
Tell us what you run and the hours you want covered, and we'll send you a quote. No obligation.