Monitoring and on-call alerting

When something stops, we text, call and email you. Once it is over, we get you to write down what happened.

Your servers, backups and nightly jobs check in with us on a schedule, or we check them ourselves. If one stops, we contact your people one at a time until someone answers. Nobody has to be awake to notice.

NIS2 deadlines built in · no incident is ever silenced · the person woken needs no password

Betalfil03:14
The hourly payment file has not arrived.
Kari was woken 03:12. No answer yet. You are next.
I have it
Incident report
Incident #7: Betalfil
Aware 03:12 · answered 03:16 2 contacted, each attempt logged Postmortem written 08:40 NIS2 early warning: met

How it works

Three steps, and the third is the one nobody else makes you do.

  1. You say what must keep happening

    Pick a nightly job, a backup, a server, or a task someone must do every three months. It tells us when it is done, or we check it ourselves.

  2. If it goes quiet, people get called

    Nobody has to spot it. We contact your people one at a time, with a wait between each, until one answers. One tap stops it.

  3. Afterwards, you write down what happened

    If the check asks for a postmortem, the incident stays open until someone writes it. We email the person who answered when it is owed, and again a day before it is due.

The report an auditor reads, written while it happens

Everywhere else this gets put together later, out of a chat log, by someone who was asleep when it started. Here the incident stays open until the postmortem is written, so it gets written while people still remember.

Aware
How we became aware, and at which minute.
Contacted
Everyone we called, how we called them, and what happened each time.
Postmortem
Why it happened, in the responder’s own words.
The authority
What we owed them, and whether we met it.
See a finished incident document →

The legal clocks: 24 hours, 72 hours, one month

NIS2 and Norway’s digitalsikkerhetslov set three deadlines, all of them counted from the moment you knew. Here that moment is a timestamp, not something anyone has to remember.

Hour zero
The minute the check should have arrived and did not.
Early warning
24 hours.
Incident notification
72 hours.
Final report
One month.

The line you paste

One line at the end of a job you already run. Nothing to install.

bash
# Nightly backup finished
curl -fsS -X POST \
  https://app.amondinops.com/api/vs/ping/9f2c41e8a7b3

It answers with the time the next one is due, so the job can log that too.

It runs in Europe
The app and its database run in Ireland. Email is sent from Frankfurt. Texts go through a Norwegian provider. Calls are placed by Twilio, a US company that keeps the call data in Ireland.
An agent can drive it
An AI assistant connects at app.amondinops.com/mcp with a key from your workspace. It can make a check, fetch the check's key and send the first ping. It cannot answer an alarm or write a postmortem. A person does those.

A check is set up one of seven ways. A machine tells us it ran, or we call it. A job starts a timer and stops it when it finishes. A person confirms a task was done. A date runs out. A domain's daily mail reports stop arriving. Or the monitor you already have, Grafana or Alertmanager, raises an alert. The last three catch the quarterly access review nobody did, the agreement that ran out in silence, and the mail that started bouncing.

When most of a check's alarms turn out to be nothing, the check says so and offers to stop ringing phones. It still opens an incident when it breaks for real.

Start with one check tonight

Add one line to the end of a job you already run. If the line stops arriving, we tell your people at once, not at nine the next morning.

Create an account

Looking for a whistleblowing channel?

Then you are in the wrong place. The Norwegian word varsling means two things, and this is the other one.

That is something else