Vercel Observability lets users create custom monitors for alerts on metrics

These new monitors permit tracking diverse signals, including AI Gateway costs, Core Web Vitals, performance metrics, failure rates for Workflows and external APIs, or Vercel Sandbox CPU time.

Each monitor focuses on one metric, perhaps response time, or a ratio derived from two metrics, like errors divided by total requests. Users select either statistical anomaly detection or a fixed threshold before setting the detection window, filters, and grouping.

Creating a monitor can be done by examining a chart's query in Observability, starting from a preset, or navigating to Settings > Alerts for your team.

Alternatively, you can define a custom monitor within a JSON file and deploy it using `vercel monitors add` in the Vercel CLI. To see the schema and examples, execute `vercel monitors schema --type custom`.

Alerts can be delivered via email, Slack, Vercel Inbox, push notifications, or team-level webhooks, and context can be added for Vercel Agent investigation.

These custom monitoring features are accessible on Pro and Enterprise plans that include Observability Plus, costing $0.0035 per GB of data scanned. Built-in alerts remain part of Observability Plus.