CockroachDB vs incident.io

Investigations went generally available in early August: Nexus posts a root-cause hypothesis and its evidence into the incident channel within minutes of declaration. Around it the on-call layer keeps filling in — coverage policies that flag gaps in a rota, escalation reassignment, shift swapping, private incidents scoped to whole teams. Insights picked up 'match any' filtering, and MCP plus the macOS app reached GA in July.

The agent has moved from something you ask to something that acts the moment an incident opens, while everything else hardens the scheduling and escalation machinery beneath it. Coverage policies and vacation conflict checks point at an on-call product being pushed toward guarantees rather than rotas you hope are correct.

Expect Nexus to reach past the hypothesis into the work that follows it — suggested actions, post-incident drafting — and more policy checks that catch scheduling gaps before an incident finds them.