OpenAI has rolled out a new beta feature to its ChatGPT platform, allowing paying users to schedule reminders and recurring requests with the AI assistant. Dubbed "tasks," this feature marks a significant step towards the development of AI agents that can act independently. The rollout is currently underway for ChatGPT Plus, Team, and Pro users worldwide.

The tasks feature enables users to set simple reminders, such as notifying them when their passport expires, and receive push notifications on their preferred platform. Moreover, users can set recurring requests, like receiving a weekend plan based on their location and weather forecast every Friday. This capability is a notable departure from traditional digital assistants like Siri and Alexa, which were not capable of such scheduled information requests.

According to OpenAI CEO Sam Altman, 2025 will be a pivotal year for AI agents, which are expected to "join the workforce" this year. While the tasks feature is a limited version of an agentic system, it demonstrates the potential for AI models to act more independently. Users can access tasks by selecting "4o with scheduled tasks" from a dropdown menu in ChatGPT and sending a message to the AI assistant to create a reminder or action.

In some cases, OpenAI says ChatGPT may suggest certain tasks based on chats. Users can manage tasks through the dedicated tasks manager tab on the web app or by chatting with the AI assistant on any platform. Notably, the tasks feature allows ChatGPT to browse the web on a set schedule, but it will not run continuous searches in the background or make purchases. For instance, users can instruct ChatGPT to check for concert tickets to see their favorite artist in their area once a month, but it cannot alert them the moment tickets go live or make a purchase on their behalf.

OpenAI is using this beta period to gather insights into how users interact with tasks before making the feature more widely available on its mobile app and free tier of ChatGPT. During this beta launch, users cannot set tasks through Advanced Voice Mode. The company is gearing up to release more advanced agentic systems, including an agent reportedly called Operator that can write code and book travel, which may be released in the coming weeks.

As OpenAI and other AI model developers push the boundaries of what computers can do, they must also address the challenges of creating AI assistants that can perform simple tasks like setting timers and reminders. The tasks feature is a significant step towards more advanced agentic systems, but it also raises concerns about the potential problems that may arise with more independent AI systems.

OpenAI's safeguards will likely be tested in the coming months as it rolls out more advanced AI agents. The company's ability to balance the capabilities of its AI models with the need for control and oversight will be crucial in shaping the future of AI agents and their potential impact on various industries.