Escalate Real Threats to PagerDuty. Ignore the Rest.

Rhombus + Claude Series — Tying Into the Enterprise

Most enterprises already have a paging system — PagerDuty, Opsgenie, or something similar — that handles escalations when something goes wrong in production. What they don’t have is a way to connect physical security events to that same system, because the signal-to-noise ratio is too low. Paging an on-call engineer every time a tree branch moves in front of a camera doesn’t work.

The result is two completely separate incident response workflows. Digital incidents get paged and triaged. Physical incidents get noticed eventually. And when an event crosses both worlds — someone accessing a data center at an unusual hour, for instance — nobody connects the dots.

The fix is to put AI between your alert source and your paging system, so only the signal gets through.

Rhombus exposes alerts as structured data with frames attached. PagerDuty exposes an incident API. Claude decides what’s actually worth waking someone up for.

For the developers:

With the Rhombus plugin, intelligent escalation is one prompt:

“Every hour, review alerts across all sites. Only page PagerDuty for a person in a restricted area outside business hours or multiple cameras offline at the same site.”

Claude triages, PagerDuty pages, the on-call team stays sane.

The payoff: Your physical security events finally share the same incident response discipline as the rest of your infrastructure. Real threats escalate immediately. Everything else stays out of your team’s way. And the camera system stops being a parallel universe with its own operational rules — it becomes part of the same organism as the rest of your ops tooling.