An alert after the breach is a report
Most SLA tooling triggers when the clock runs out. By then the commitment is missed, and the notification records failure instead of preventing it.
The useful alert arrives earlier. It is far enough ahead for action but close enough to avoid noise. The right timing depends on the ticket type, not a single global setting.
Acknowledgment separates an alert from a message. Without a state to check, a scheduled flow resends the same warning every run. That trains the channel to ignore it when it matters most.
Business hours quietly break these systems. A timer counting through a weekend causes a Monday morning full of breaches no one could prevent.
How the ActionFlow runs on the canvas
Timers are the whole mechanism here, so the flow is built around them rather than around the message.
A Start Scheduled node runs frequently. An HTTP Request or Jira node reads open tickets, including their creation time and priority.
A Date node calculates elapsed time against the SLA for each ticket's band and handles business-hours logic. A Math node applies the warning fraction, the point before the threshold where an alert still helps.
A Filter node keeps only tickets inside the warning window. An If node checks the acknowledgment field so an already-warned ticket does not warn again.
A Slack node notifies the owner. A Wait node holds. The If node re-checks, and a second Slack node escalates to the team lead if nothing changed. The third level notifies the whole channel.
Every notification writes an acknowledgment state back through an HTTP Request or Jira node. That write-back is what makes the escalation ladder work rather than repeat.
A Google Sheets node logs warnings against actual breaches, which tells you whether the warning fraction is set early enough to be useful.
Nodes this ActionFlow uses
- Start Scheduled: runs frequently enough to catch a window, not a breach.
- Jira (Integration): reads open tickets with created time and priority. HTTP Request covers other helpdesks.
- Date (Util): calculates elapsed time and carries the business-hours logic.
- Math (Util): applies the warning fraction ahead of the threshold.
- Filter (Data): keeps only tickets inside the warning window.
- If (Control): checks acknowledgment state so a warned ticket does not warn twice.
- Wait (Control): holds the escalation ladder between levels.
- Slack (Communication): notifies owner, then lead, then the channel.
Frequently asked questions
Start building AI workflows
Create a free account, open a template or a blank canvas, and run your first ActionFlow.
Newsletter
Get product updates
New nodes, agents, and product notes. We send mail only when we have something worth opening.