Engineering · Use case
Investigate an Incident End-to-End
An alert fires. Your agent correlates monitoring data with recent deploys, starts a Slack thread, and posts an initial hypothesis.
The grind
When an alert fires, the first minutes are chaos - and they're spent gathering context instead of fixing the problem.
- Someone scrambles to correlate the alert with recent deploys.
- A Slack thread gets started from scratch, pulling people in one at a time.
- Initial findings live in someone's head until they type them up.
The clock is running, and the response is slow to organize.
The flow
With fastn, your agent runs the first response when an alert fires. Your agent forms the hypothesis; fastn pulls the monitoring and deploy data and opens the Slack thread.
- Read the alert and recent deploys. fastn pulls the incident from PagerDuty and recent deploy history from GitHub.
- Your agent correlates. The agent connects the alert to likely causes and forms an initial hypothesis. This is your agent's reasoning, not fastn.
- Start the Slack thread. fastn opens an incident thread in the right channel and posts the hypothesis and context.
The agent decides what likely happened; fastn does the data-gathering and comms across PagerDuty, GitHub, and Slack.
The trigger
This blueprint is event-driven: it fires the moment an alert goes off. fastn's app-event triggers watch PagerDuty and wake your agent instantly - no one has to start the response by hand.
The payoff
The first minutes of an incident are organized, not chaotic.
- Context is ready. The alert, deploys, and a hypothesis are gathered automatically.
- Comms start instantly. The Slack thread opens itself.
- Faster resolution. Responders start fixing, not gathering.
Make it yours
Swap PagerDuty, GitHub, or Slack for your monitoring, code host, and chat - fastn connects to the tools your customers already use, and your agent runs the same flow. Talk to us to see it on your stack.