Vercel Agent now screens alerts before they page you
Vercel Agent reads logs and metrics for each triggered alert and only forwards the ones it judges critical to your paging system, filtering out the noise.
Vercel announced on 26 October that Vercel Agent can now investigate alerts from Vercel monitors before they reach your paging system. Turn on validation for a webhook destination, such as incident.io, and only the alerts Agent classifies as Critical get sent through. The agent examines logs and metrics to assess impact, telling apart something like a harmless health-check error from a sustained run of errors hitting customer checkouts.
You enable it from a monitor’s notification settings by linking a webhook destination and selecting “Critical Only” next to it, or from the CLI with vercel monitors notifications and the --critical-only flag. It needs Pro or Enterprise with the Observability Plus add-on, which includes 10 agent investigations per billing cycle before further ones are billed separately, plus $0.0035 per GB of data scanned for custom monitors.
Why it matters: most on-call pain is a triage problem, not a detection one, every monitor fires and a human has to work out which page actually matters. Letting an agent make that first pass before anything reaches incident.io is worth trying if your on-call rotation is drowning in noise.
The caveat: this sits behind a paid add-on, and the free quota of ten investigations a cycle will not cover a noisy setup, so watch the bill once you switch it on.