Vercel Agent can investigate monitor alerts before sending a page

Vercel now lets teams send only alerts its agent classifies as critical to an opted-in webhook destination.

**Dek:** Vercel’s October 28 update lets teams opt individual webhooks into a Critical Only path. The change puts agent classification between a monitor firing and an incident destination receiving the alert.

Vercel added an alert validation option on October 28. When a webhook destination is set to Critical Only, Vercel Agent can investigate monitor alerts before they reach that destination, and only those it classifies as Critical are sent there. Operators can enable the option in the monitor’s notification settings or when configuring the destination through the Vercel CLI. Vercel documents availability on Pro and Enterprise with Observability Plus. [Vercel changelog](https://vercel.com/changelog/vercel-agent-can-now-validate-alerts-before-paging-you)

A paging decision is different from a monitor firing

A monitor can detect a threshold crossing that deserves investigation even when it does not require waking an on-call engineer. The new option changes where that second decision is made for an opted-in webhook. A team might use it on a noisy service where many short spikes are harmless. The same team needs to understand what happens to a real incident that the investigation classifies below Critical. The release describes how Vercel routes notifications to the destination. It does not establish a measured reduction in false pages or missed incidents for a particular service. [Vercel changelog](https://vercel.com/changelog/vercel-agent-can-now-validate-alerts-before-paging-you)

A useful evaluation starts with historical incidents or controlled alerts whose severity the team can label independently. Include a clear outage, a noisy recovery and a slow degradation that matters to customers. Compare the classification with what the on-call process would have expected and record how long it takes to reach the destination. During a trial, a separate path for reviewing monitor activity can help an operator identify alerts that the critical-only webhook did not receive.

Plan limits matter for incident workflows

Vercel says Observability Plus includes ten Agent investigations per billing cycle and additional investigations are billed separately. It also lists a data scanning charge for custom monitors and says built-in alerts remain included with Observability Plus. Before applying the feature to a large set of destinations, teams can estimate their alert volume and verify how the operational cost appears in their account. [Vercel changelog](https://vercel.com/changelog/vercel-agent-can-now-validate-alerts-before-paging-you)

The update gives teams another place to decide which alerts page a person. Its value depends on the service’s alert mix, the quality of the investigation and whether operators can still review the activity behind a paging decision.