ChatGPT Work’s event-triggered tasks let a reporter ignore email for an afternoon
A TechRadar journalist let ChatGPT monitor Gmail for an afternoon; the AI interrupted only three times for actionable messages, suggesting drafts and flagging itinerary changes.
The feature, available on eligible plans in ChatGPT on the web and mobile, allows users to connect Gmail, Slack, and other channels. Users set a trigger, such as a new Gmail message, and give the AI instructions on how to respond. In the test, the journalist connected Gmail and left the trigger broad, while making the prompt selective: ChatGPT was to alert only when a message created an actual obligation or came from a family member, and to draft a reply for approval.
The reporter then closed the inbox and went about the afternoon. Over several hours, ChatGPT interrupted three times. The first alert flagged a message from the reporter's bank, explaining who the sender was and what needed review. The second came from a family member; ChatGPT provided a brief reply draft that the reporter approved with minor additions. The third alert concerned a change to previously purchased event tickets, which was especially useful because the itinerary shift affected other plans.
Afterwards, the reporter checked Gmail and found only two emails that might have been worth seeing earlier, neither of which required immediate action. No problems arose from ignoring promotional emails in real time.
The report offers advice for anyone trying the feature: start with a short test period and review the first few runs in ChatGPT's Scheduled section. If routine receipts keep slipping through, add clearer exclusions. If real requests are missed, explain what counts as an expected response and allow ChatGPT to flag uncertain cases. The report also recommends keeping the first version informational, having ChatGPT summarize the obligation rather than reply automatically, and using push or text notifications to avoid email-based alerts.