Vercel Agent can now validate alerts before paging

Vercel Agent has introduced a feature that evaluates alerts from Vercel monitors before they trigger paging systems. When enabled for a webhook destination, the system sends only alerts classified as Critical, preventing non-urgent notifications from reaching incident management tools like incident.io. Users can activate this setting in a monitor’s notification preferences or configure it via the Vercel CLI using the vercel monitors notifications command with the --critical-only flag.

The validation process involves Agent analyzing logs and metrics to assess the actual impact of an alert. For instance, it can distinguish between health-check errors that do not affect users and sustained 5xx errors impacting critical flows like checkout requests. This reduces false positives and ensures teams respond only to incidents with real user impact.

Users can enhance Agent’s investigation by adding contextual details to each monitor, such as descriptions or tags, to guide its assessment. This customization helps refine the validation process and improves the accuracy of alert prioritization for specific services or environments.

The feature is available on Pro and Enterprise plans with Observability Plus, which includes 10 Agent investigations per billing cycle. Additional investigations beyond the included quota are billed separately. Custom monitors incur a cost of $0.0035 per GB of data scanned, while built-in alerts remain included with Observability Plus.