OpenAI has launched a new capability for ChatGPT called ChatGPT agent, which allows users to manage complex, multi-step tasks. This feature is available to Pro, Plus, and Team users and enables the AI to complete assignments from start to finish using a virtual computer. With agent mode activated, users can ask ChatGPT to perform tasks such as reviewing calendars, planning recipes, and analyzing competitors. Unlike typical chatbot interactions, the agent can navigate websites, run code, and compile research into editable formats, with users retaining control throughout the process.
The ChatGPT agent is powered by a combination of Operator for web navigation, deep research for synthesizing information, and ChatGPT’s core model for natural language understanding. To access agent mode, users can select it from the tools dropdown in any ChatGPT conversation. OpenAI has emphasized safety and user control, with the agent designed to avoid high-risk actions without user approval and trained to recognize and reject malicious instructions. The company plans to regularly add new features and improvements, aiming to make ChatGPT more versatile and useful for a broader set of users. This release marks a shift from conversational assistance to hands-on task execution, with oversight and control remaining in the users’ hands.
