ChatGPT is taking another step toward acting less like a chat window and more like an AI personal assistant. OpenAI has added new controls for scheduled tasks, giving users a central place to manage work that ChatGPT can run later or repeat over time.
A new home for active tasks
The main change is a new "Scheduled" page in the sidebar. Instead of managing tasks only where they were created, users now have one place where active tasks can be reviewed and controlled.
From that page, users can view, pause, edit, or delete tasks. That matters because scheduled tasks are only useful if they remain easy to manage. A reminder, research check, or recurring prompt can quickly become noise if there is no clear way to change it or stop it.
The new page makes scheduled tasks feel more like a standing part of ChatGPT rather than a side feature. It also gives users a clearer picture of what ChatGPT has been asked to do on their behalf.
Research tasks can watch for changes
OpenAI is also improving how research tasks behave. These tasks can search the web and connected apps, then send alerts only when something actually changes.
That is an important distinction. A research task that reports every time it runs can become repetitive. A task that waits for a meaningful change is closer to a monitoring tool: it checks the relevant places and only interrupts the user when there is something new to report.
The source article does not list examples of specific research topics or connected apps. The key point is narrower and more practical: research tasks are designed to look across the web and connected apps, with notifications tied to actual changes.
Scheduling is becoming more flexible
Users can schedule tasks for specific times. They can also choose broader parts of the day: morning, afternoon, or evening.
That gives scheduled tasks two kinds of timing. A specific time is useful when the task needs to happen at a precise moment. A broader window is useful when the exact minute matters less than the general routine.
OpenAI says all tasks are now faster and more reliable. The source does not provide benchmarks or technical details, so the practical takeaway is simply that OpenAI is positioning scheduled tasks as more dependable than before.
Access, limits, and automatic pauses
The feature is available to Plus, Pro, Business, and Enterprise users. The number of active tasks varies by plan.
There are also limits on how often tasks can run. Tasks can run at most once per hour. That keeps scheduled activity from becoming continuous automation inside ChatGPT.
Tasks can also pause automatically if the user goes inactive. The source article does not define inactivity, but the behavior suggests that scheduled tasks are connected to ongoing user engagement rather than running without end.
Pulse is being folded in
OpenAI is retiring the previous "Pulse" feature and folding it into scheduled tasks. That makes scheduled tasks the place where this kind of proactive ChatGPT behavior is being consolidated.
The move is small in wording but meaningful in product direction. Instead of keeping a separate feature for proactive updates, OpenAI is putting that capability under the broader scheduled tasks system.
For users, the result is a simpler mental model: create a task, decide when it should run, and manage it from the "Scheduled" page. For ChatGPT, it is another step toward becoming a personal assistant that can track requests, monitor changes, and surface updates without waiting for every prompt to begin from scratch.