OpenAI, the company behind ChatGPT, just announced Operator. It is a generative AI service that acts like an agent and performs tasks on your behalf. Using its own browser, Operator looks at a webpage and interacts with it by typing, clicking and scrolling on its own – no need for any input.
The rollout will be gradual, and the first to get it are ChatGPT Pro subscribers in the United States.
Operator can handle various repetitive browser tasks, and OpenAI claims it can fill out forms, order groceries, and even create memes. It can use the same interfaces and tools that humans interact with, and that would also help businesses, opening new engagement opportunities for them.
A research preview of Operator, an agent that can use its own browser to perform tasks for you. pic.twitter.com/wkBBDIlVqj
— OpenAI (@OpenAI) January 23, 2025
Operator is powered by a new model called CUA – Computer-Using Agent. It combines GPT-4o vision capabilities with advanced reasoning through reinforced learning. CUA is trained to interact with GUIs – graphical user interfaces with buttons, menus, and text fields people see on a screen.
When the service is stuck or needs assistance, it simply hands control back to you. You also need to manually input sensitive data, such as passwords or other verification forms.
Operator can work with services such as Doordash, Etsy, Booking.com, Uber, and Instacart, and it can do research through media partners like Associated Press and Reuters.
Source
Discussion about this post