Skip to main content

Responding to Incidents in Slack

@Rootly in Slack brings the parts of Rootly you reach for during an incident directly into where your teams are already operating: Slack! @Rootly can be used in any incident channel, Slack AI assistant pane, or in a direct DM to the Rootly assistant. Ask @Rootly to:
  • “This is now a SEV1. Update the incident severity.”
  • “Draft a comms to our high value customers, letting them know the impact of the incident on their product experience.”
  • “Page the security team, and convert this to a private incident.”

How @Rootly in Slack works

The hardest part of running an incident isn’t just the work to mitigate and resolve the incident: it’s also keeping your teams and stakeholders up-to-date so they can quickly support you. However, responders join late, miss key learnings, and take time to catch up: comms leads need to draft customer updates while still triaging, commanders need to pull in the right stakeholders and leadership must be aware of major updates. To collaborate effectively, everyone needs to have all the context they need, while minimizing distractions on other teammates. @Rootly in Slack collapses context into a single conversational surface directly in the channel where the incident is being run. @Rootly reads from your Rootly data, channel messages, bridge call transcripts (if available), to take actions on your behalf. To start, type @Rootly in Slack, and then ask your question. The assistant leverages context in real-time to decide how to take action. It replies in-thread to keep you aware of any actions taken. Learn more about how to use @Rootly in Slack
@Rootly can only take actions that you as a user have permission to perform. Check your Rootly role to confirm what your permissions are.

Key Capabilities at a Glance

CapabilityDescriptionExample question
Catch me upStructured brief of an incident’s state, key events, current responders, and open action items@Rootly what’s going on right now?
Who’s on callLook up current and historical on-call across teams, services, schedules, and escalation policies@Rootly who’s on call for the Security team?
Page respondersPage a team, user, escalation policy, service, or Slack user group via Rootly On-Call@Rootly page the Search Experience team.
Update an incidentChange severity, status, title, services, custom fields, or roles — with the assistant confirming what it did@Rootly add the Prod DB Team to the incident and bump up the severity to a SEV2.
Manage action itemsCreate, update, list, or delete action items from natural-language requests@Rootly create an action item for all of the next steps we discussed on the bridge.
Find similar incidentsEmbedding-based similarity search across past incidents, scoped to your tenant@Rootly I remember an incident similar happened last week that impacted this service. Summarize it for me.
Draft customer updatesGenerate Slack-ready or status-page-ready update copy from the current incident state@Rootly write a draft to send to the leadership team and our partners.

How Permissions Work

The assistant uses each user’s existing Rootly permissions. If a user can’t see a private incident in the web app, the assistant won’t surface it for them in Slack. If a role doesn’t allow paging, the assistant declines to page when that user asks.
Permissions are managed in Configuration → Roles & Permissions in the Rootly web app. The assistant respects whatever role assignments are already in place — there’s no separate AI permission system.

Where to Go Next

These pages cover specific aspects of using and configuring Rootly AI in Slack:

FAQs

A Rootly admin enables it in Configuration → Rootly AI by toggling Enable @Rootly in Slack on. A Slack workspace admin then needs to reconnect the Slack integration for the update to take effect.
The assistant works best inside an incident channel: where the assistant has full access to incident data. You can also use the assistant via Slack’s AI assistant pane pointed at an incident channel in the main view, or by messaging it directly in Slack. For the latter two surfaces, the assistant can only read information, not take actions.
Yes. The assistant uses the requesting user’s existing Rootly permissions for every action. It cannot do anything in Slack that the user couldn’t already do in the Rootly web app.
The Rootly assistant has to be invited to a channel before it sees anything there and can take actions. User permission still applies: a user who can’t see a private incident in the web app can’t ask the assistant about it from any surface.
Paging happens through Rootly On-Call only. If you have a third-party paging integration configured elsewhere in Rootly, the assistant will not fall back to it.
Claude Sonnet 4.6 from Anthropic by default, with automatic failover to GPT-5 from OpenAI. Both providers are accessed through a managed gateway.