On August 1, 2026, the OpenAI Help Center updated "Scheduled Tasks in ChatGPT," focusing not on adding a timed reminder button but on bringing the creation, editing, monitoring, and notifications of scheduled tasks into a clearer entry point. For ordinary users, ChatGPT is starting to resemble an assistant that keeps a close eye on changes; But it still isn't an automated platform that can run infinitely in the background.
What was changed this time?
Scheduled tasks now have independent management pages, allowing web, mobile, and desktop platforms to be viewed and adjusted more intuitively. Users can set one-time reminders or arrange repetitive tasks; Monitoring tasks check for changes periodically, only notifying when meaningful new situations arise, rather than disturbing users every time.
This change solves the previous problem of tasks being scattered across dialogues and unclear entry points for modifications. When creating, it's best to clearly state the inspection object, trigger conditions, notification content, and end time to avoid relying solely on vague context for judgment.
Three boundaries are more important than new entrances
- The maximum execution frequency for a single task is once per hour, which is not suitable for minute-level polling.
- The number of tasks enabled simultaneously is limited by the plan: 3 for Go, 5 for Plus, and 15 for Pro, Business, and Enterprise.
- When creating a task in a project with files, the task cannot read the project file; Long-term inactivity or deletion of related conversations may also cause task pauses.
This means it is better suited for low-frequency, fault-tolerant information tracking, rather than relying on local files, requiring continuous operation, or critical processes that must succeed on time.
Which tasks are best entrusted to it?
Daily briefings, fixed reminders, product status changes, and public information updates are all suitable scenarios. Monitoring prompts can adopt a structure of "check what — notify only of what change — how to handle when no changes occur," and first observe the results several times with short-cycle tasks.
To understand the overall direction of ChatGPT moving from conversation to continuous execution, you can first look at the changes in ChatGPT Work's capabilities since its launch (/article/1908-chatgpt-work-shang-xian-cong-liao-tian-zhu-shou-zhuan-xiang-ke-chi-xu-zhi-xing-d); This update addresses more specific task management layers.
The real impact on users
Planned tasks lower the threshold for continuous monitoring but do not eliminate reliability boundaries. For work involving code, project documents, approval steps, or traceable execution records, more comprehensive automation tools should still be used. Regular users can start with a non-critical reminder, keep manual reviews, and regularly clean up unneeded active tasks.
