OpenAI is launching a new beta feature called Tasks for ChatGPT that allows users to schedule upcoming actions and reminders.
The feature, rolling out to Plus, Team, and Pro subscribers starting today, is an attempt to make chatbots more like traditional digital assistants. Think Google Assistant or Siri. However, it has more advanced language features than ChatGPT.
Tasks work by allowing you to tell ChatGPT what you want it to do and when you want it to do it. Do you want to know the weather forecast at 7am every day? A reminder about your passport expiration date? Or just a knock-knock joke to tell your kids before bed? With ChatGPT, you can use scheduled one-time Now you can handle all of this through tasks or repetitive tasks.
To use this feature, subscribers must select “4o with scheduled tasks” in ChatGPT’s model picker. From there, all you have to do is tell ChatGPT what you want it to do and when. The system can also proactively suggest tasks based on conversations, but users must explicitly approve the suggestions before they can be made. (To be honest, I feel like these suggestions could inadvertently create some annoying slop).
All tasks can be managed directly in the chat thread or through the new tasks section in your profile menu (only available via the web), making it easy to change or cancel tasks you’ve set. Users on web, desktop, and mobile will receive a notification when these tasks are completed. Additionally, the number of active tasks that can run at the same time is limited to 10.
OpenAI hasn’t said when (or if) this feature will be made available to free users, and it’s possible that tasks will remain a premium feature to justify ChatGPT’s subscription costs. It suggests that there is. The company has subscription tiers of $20 and $200 per month.
Scheduling functionality is a common feature for digital assistants, but this marks a change in functionality for ChatGPT. Until now, AI has operated only in real-time, responding to immediate requests rather than handling ongoing tasks or future plans. The addition of tasks suggests that OpenAI is expanding ChatGPT’s role beyond conversations and into areas traditionally occupied by virtual assistants.
OpenAI’s task ambitions seem to go beyond simple scheduling. Bloomberg reported that Operator, an autonomous AI agent that can independently control computers, is scheduled to be released this month. Meanwhile, Reverse Engineer’s Tibor Blaho says OpenAI integrates with Tasks to help ChatGPT search for specific information, analyze issues, summarize data, navigate to websites, and access documents. It appears to be developing a codename “Caterpillar” and users have discovered that they receive notifications. Upon task completion.
The rise of “agent-based” AI in 2025 is more than just a technological advance; it’s also an economic issue
As I previously wrote in October, the rise of “agent” AI in 2025 is not just a technological advance, but also an economic one. These agent-like capabilities represent a strategic way to monetize expensive AI infrastructure. OpenAI’s decision to put this feature behind ChatGPT’s paywall was predictable, but the real questions remain. The question is whether reliable results can be obtained. The last time I got a demo of the OpenAI agent, it produced inaccurate information. The coming months will reveal whether the team has resolved these fundamental reliability challenges.
I think of this new feature as a slightly more sophisticated script, but at the end of the day, the tasks follow a simple, mechanical set of instructions, just like a typical bot. The goal of many frontier AI labs like OpenAI is to evolve these capabilities into ones that can interact with the environment, learn from feedback, and make decisions without continuous human input.
However, questions remain about how reliable these scheduled tasks are and what happens if ChatGPT fails to provide time-sensitive information. OpenAI’s decision to launch Tasks in beta suggests that the company is still working out the details and wants to gather real-world feedback before a broader rollout.
For now, any paid user of ChatGPT can start experimenting with tasks by looking for the “4o with scheduled tasks” option in the model picker. However, this is still in beta, so don’t use it as a reminder for a very important meeting just yet.
