The fresh AI Openai agent can perform various activities for you on the Internet

According to the Entry on the blog, OPENAI provides the “research version” of an artificial intelligence agent called Operator, who can “go to the Internet to perform tasks for you”. “Using his own browser, he can browse the website and interact with it by entering, clicking and scrolling,” says Opeli. It will be launched as the first in the USA for subscribers of ChatgPT Pro OpenAI for $ 200 a month.

The operator uses the model “agent using a computer”, which combines GPT-4O vision capabilities with “advanced reasoning through strengthening by strengthening” in order to be able to interact with the graphic interfaces of the user, says OpenAI. “The operator can” see “(through screenshots) and” interact “(using all the activities that the mouse and keyboard allows) with the browser, enabling it to take action on the Internet without the need for non -standard API integration,” says Opeli.

The operator can apply reasoning to “self -recreate”, and if he gets stuck, he will give the user control. He will also ask the user to take control when the website asks for confidential information, such as login details, and “should” ask the user to approve actions such as sending an e-mail. Opeli also claims that the operator has been designed to “reject harmful demands and block prohibited content.”

Opeli claims that he cooperates with such companies as Dordash, Instacart, Opestable, Priceline, Stubhub, Thumbtack, Uber, that the operator “responds to real needs, observing the established standards.” The company warns, however, that not everything can yet work as expected; The tool currently has problems with “complex interfaces, such as creating slide shows or calendar management.”

In the future, OPENAI plans to provide the operator with Plus, Team and Enterprise users and “integrate these functions from chatgpt”.