How to monitor ChatGPT scheduled tasks

The problem with scheduled tasks

ChatGPT scheduled tasks run on OpenAI's side. When one doesn't run — a platform incident, a paused task, a usage limit — there's usually no error to catch. You find out when you notice the brief didn't arrive, sometimes days later.

People running many tasks describe the same chore: opening each one every morning to check that it actually ran.

What works today: relay the notification email to your ping URL

Did It Run? check-ins today are HTTP requests to a watch's unique ping URL. A ChatGPT task can't make that request itself, but it can email you when it finishes, and an automation tool that watches your inbox can turn that email into a ping:

1. Create a watch with the same schedule as the task, e.g. every weekday by 9:15am in your timezone, with 15–30 minutes of grace.
2. Turn on email notifications for the task, so each completed run emails your account address.
3. In an automation tool that can watch your inbox (n8n, Make, or Zapier on a paid plan), trigger on new emails from ChatGPT that mention the task, then send an HTTP POST to the watch's ping URL with the email text as the body.

If the task doesn't run, no notification arrives, no ping arrives, and you get an alert. It needs that second tool, so it isn't zero-setup, but it works with what's live today.

Coming soon: email check-ins

The simpler setup, forwarding the notification straight to a watch's own email address with an inbox rule and no other tool, needs email check-ins. They aren't live yet; they're coming soon. Until then, there's no way to watch a ChatGPT task with Did It Run? without a relay like the one above.

Catch "ran, but said nothing"

Ask the task to end its answer with a fixed marker such as BRIEF-OK, send the notification text as the ping body, and set the marker as the watch's required keyword. If the notification includes the task's reply and that reply is an apology, a refusal, or an empty result, you get an empty alert.

One roll-call for every agent

If you run tasks in ChatGPT, routines in Claude, and workflows in n8n or Zapier, the daily roll-call digest puts all of them in one email: ✅ what ran, ❌ what's missing, ⚠️ what came back empty.