OpenAI has launched a new “agent mode” for ChatGPT, enabling the AI assistant to perform real-time tasks on users’ behalf.
OpenAI has launched a new “agent mode” for ChatGPT, enabling the AI assistant to perform real-time tasks on users’ behalf. Announced Thursday, the feature uses a virtual computer to let ChatGPT “think” and “act,” executing commands like checking calendars or planning meals. It is now available for Pro, Plus, and Team subscribers.
The mode integrates OpenAI’s Operator and Deep Research tools, enhancing ChatGPT’s capabilities. In a demo, the agent helped plan a wedding, from outfits to hotel bookings, all from one prompt.
The move places OpenAI in direct competition with Google’s Gemini and Apple’s upcoming Siri upgrade. However, OpenAI has imposed safeguards, banning high-risk tasks like bank transfers and requiring user confirmation for sensitive actions such as sending emails.
“OpenAI advises users to be thoughtful when granting the agent access to personal data,” especially for tasks that may not require it.
This development marks a major leap toward AI-powered autonomous digital assistants.