OpenAI has introduced a groundbreaking new feature for ChatGPT that transforms the AI chatbot into a semi-autonomous digital assistant.
In a live stream, OpenAI CEO Sam Altman and other company executives demonstrated this new Agent Tool, showcasing how it functions similarly to OpenAI’s Operator Tool while integrating the advanced Deep Research capability.
This new GPT Agent can perform real-world tasks directly on your computer using written prompts and access to linked accounts.
For example, it can manage your calendar, run PowerPoint presentations, and browse the internet just like a human would.
Currently, this ChatGPT Agent feature is being rolled out exclusively to paying users on the ChatGPT Pro, Plus, or Team plans, who can activate it via the Agent Mode in the ChatGPT tool menu.
This is not just another basic upgrade—it’s a fundamental shift in how users interact with ChatGPT. Traditionally, users ask ChatGPT a question, receive an answer, and then manually perform related tasks using other apps.
Now, users can instruct ChatGPT to complete multi-step real-world actions on their behalf. For instance, if you’re planning a trip to another city, you can ask the agent to find affordable hotels, compare pros and cons, and recommend the best options. The chatbot will carry out this task automatically.
While this feature blends capabilities from both the Operator and Deep Search tools, it also introduces unique functionalities by completing tasks with ease and intelligence.
You could even ask it to book a table at a restaurant, and it would not only suggest options but also check your calendar to find a suitable time for the reservation.
Importantly, this AI agent can also interact with your web browser, access your computer files, and operate installed software, making it a powerful step forward in personal AI assistance.
