Setu Recipes: reusable workflows across every app.Browse templates
Engineering

Incident command

The timeline built live from alerts, deploys and chat, the right people paged, and a postmortem drafted when it is over.

Chat

We have an incident. Run point.
The errors began two minutes after the 02:08 payments deploy. A rollback is the likely fix. The channel, the pages and a postmortem draft are ready.

Connected apps.

Connect each one once. You sign in on the app's own screen, and you can disconnect any time.

  • PagerDutyPagerDuty
  • DatadogDatadog
  • SentrySentry
  • GitHubGitHub
  • SlackSlack
  • NotionNotion

An example run.

What the agent does, step by step, after you ask. This is an illustration; your results depend on your own data.

Incident command

  • PagerDutypagerduty.get_incidentOpened at 02:14: checkout error rate above 5 percentDone
  • Datadogdatadog.query_metricsErrors began at 02:10, latency doubled at 02:12Done
  • GitHubgithub.list_deploysA deploy of the payments service went out at 02:08Done
  • Sentrysentry.list_issuesNew error: timeout calling the card processorDone
  • Slackslack.create_channelIncident channel opened with the contextWaits for you
  • PagerDutypagerduty.page_userPayments on-call and the incident lead pagedWaits for you
  • Notionnotion.create_pagePostmortem drafted with the timeline, ready to editDone

You stay in control.

Looking things up is automatic. Anything that changes something waits for you, and every step is logged.

Runs on its own

  • Opened at 02:14: checkout error rate above 5 percent
  • Errors began at 02:10, latency doubled at 02:12
  • A deploy of the payments service went out at 02:08
  • New error: timeout calling the card processor
  • Postmortem drafted with the timeline, ready to edit

Waits for your approval

  • Incident channel opened with the context
  • Payments on-call and the incident lead paged

How it works.

  1. Connect your apps

    PagerDuty, Datadog, Sentry, GitHub, Slack, Notion. Setu keeps the sign-in secure and never shows your passwords to the agent.

  2. Ask, or set a trigger

    Run it when an incident opens. You describe the job in your own words.

  3. You stay in control

    Reading is quick. Anything that writes or deletes can wait for your approval, and every step is logged.

Give your agent real hands.

Connect an app, hand your agent the tools, and watch it finish the job.