Incident response that never sleeps
A team that notices the error at 3am, works out what changed, and has the story and a proposed fix ready when you wake up.
You'll end up with an on-call team that watches your error tracker, triages anything new against your recent changes, and tells you in Slack, asking you before it changes a line of code.
About 15 minutes.
- 01
Hire the DevOps & Incident Desk
On the dashboard, press New team and pick DevOps & Incident Desk. Give it a budget per month; it can't go over it.
- 02
Connect your code and your monitoring
In Connections, sign in to GitHub and your error tracker (Sentry, Grafana, Axiom or Honeycomb). GitHub asks before any change by default: pushes, merges, comments and new issues all wait for you.
- 03
Give it the runbook
Upload your runbooks to Docs, or let the team write one: its first job audits your deployment setup and drafts a runbook for your top three failure modes, for you to publish. Pin the runbook so every incident starts with it.
- 04
Put it on call
Under On-call triggers, add a check of your error tracker every few minutes. It acts only on issues that are genuinely new, not the same error twice. If your alerting can call a URL with a header (Grafana and PagerDuty can), add a webhook trigger instead and point the alert at it.
- 05
Hear about it where you are
Link Slack or Telegram in Settings. When an incident comes in, the team reads the errors, checks what changed in recent commits, writes up what it found, and proposes a fix, which waits for your approval like any other change to your code.
Start the trigger in shadow mode if you want to see a week of its write-ups before you rely on them. Every team has a kill switch, and the activity log shows each step it took.