Set up basic on-call for my app. Even at team-of-one, I need a plan for when I'm the only one who can respond. A tiny team is not an excuse to skip this; it's a reason to do it well. Set up: 1. Alert routing: a real pager (PagerDuty, BetterStack On-Call, Grafana OnCall, Opsgenie) with SMS/phone notifications for SEV1 alerts. Email alone is not on-call. Slack alone is not on-call. 2. Escalation: if the primary doesn't acknowledge within 5-10 minutes, escalate to a secondary contact (a partner, co-founder, backup, or 'call this friend who can wake me up'). Team of one? Still add a backup number for someone who can physically wake you. 3. Access: everyone who could plausibly be on-call has current credentials for the platform, the database (read-only is fine for triage), the deploy pipeline, the error tracker, and the log viewer. Verify by trying, don't assume. 4. Runbook access: rollback runbook, deploy runbook, and incident response plan are linked from the alert message when possible, or from a single 'when you get paged, start here' doc. Middle-of-the-night is not the time to hunt for a link. 5. Handoff (for real teams): a written weekly handoff summary if you have more than one on-call person. What's currently deployed, any known issues, any expected weirdness (deploys, migrations, third-party maintenance). 6. Post-alert: quiet hours after a page, a policy for follow-up (file a follow-up issue for anything worth fixing so you don't lose it in the fog of the incident), and permission to sleep after mitigation. Don't skip a real pager because you 'always have your phone'. The Slack notification you sleep through is the one that mattered. The deliverable is the pager configured, escalation tested with a fake alert, access verified for anyone who could respond, and runbook links pinned somewhere you can find them at 2am.