LoginSign Up
← Back to all sparks
R

Rootly

INFRA · APIS
Velocity6.3

Rootly is turning its AI from an incident chat box into an evidence gatherer across the stack

incident-responseagent-nativeservice-catalogon-callobservability-integrations
◆Current state
Rootly's last month splits cleanly between AI and the plumbing that makes AI useful. Rootly AI moved from a chat panel sitting on an incident to a system that connects read-only into observability, code, infrastructure, feature flag, ticketing, and documentation tools and gathers evidence itself. Around it, the catalog gained a CLI sync from GitHub and Backstage, onboarding collapsed to a single command, and responders got an on-call widget, a customisable Alerts table, and mobile alert muting.
◆Where it's heading
The catalog work and the AI work are the same bet. Rootly is assembling a current, machine-readable picture of services, owners, and tooling so its agent can reason about an incident without a human pasting context into a chat panel. Read-only connections with nothing stored is the trust position it is taking to get access to those systems in the first place, and the retrospective AI blocks that show every prompt and source are the same posture applied to output.
◆Prediction
The step these entries set up but do not take is the agent acting rather than only gathering — proposing or executing remediation on the evidence it assembles. More catalog sources landing on the new CLI is the safer near-term bet.

◆Recent moves

  1. 8d ago

    Sync your catalog, and mute pages during Alert storms.

    A CLI that syncs the catalog from GitHub, Backstage, and others, plus mobile alert muting and an Upwind integration. This is the supply side of the AI work — the agent is only as good as how current the service and ownership data underneath it is.

  2. 15d ago

    Rootly AI now gathers incident evidence across your entire stack.

    ⚡ SPARK

    The pivot point of the last month. Rootly AI stops waiting for a human to supply context and goes to get it across observability, code, infrastructure, feature flags, ticketing, and docs, which is what the catalog investment was building toward.

  3. 15d ago

    Rootly AI connects to your observability stack and beyond.

    A duplicate row for the same 28 July release as the entry above, where the crawler captured both the headline and a variant of it. Same change: read-only connections spanning the observability, code, and ticketing stack.

  4. 20d ago

    From sign up to incident-ready in minutes.

    One command sets up team, on-call, alerting, and status page, then pages the operator's phone to prove it works. Onboarding speed matters more once the product's value depends on the catalog and integrations actually being populated.

  5. 27d ago

    The on-call widget, 24-hour time, and a new Alerts table.

    A live on-call widget in the sidebar, 24-hour time, and a customisable Alerts table. Unglamorous, but this is the surface responders stare at while an incident is running, and it had been lagging the AI work.

  6. 1mo ago

    Retrospective templates with customizable AI-blocks

    A duplicate record of the same July 9 retrospective-templates release, differing only in trailing punctuation. Counted once editorially; the feed is emitting these in pairs.