Presenting OpenAI’s newest web browsing tool.

On Thursday, OpenAI revealed the ChatGPT agent, a utility aimed at navigating the web and accomplishing tasks on your behalf. As highlighted in an X post prior to the livestream, the ChatGPT agent integrates the autonomous capabilities of its Operator agent with the analytical functions of its Deep Research tool.

OpenAI’s Operator, launched in January for ChatGPT Pro users, was able to navigate, click, and type online but had restrictions and was not broadly available. Deep Research could look up and compile data but lacked the ability to take further actions. The new web browsing agent effectively integrates these functionalities.

“Operator couldn’t engage deeply in analysis or generate comprehensive reports, and Deep Research couldn’t interact with websites to refine outcomes or access content that needed user authentication,” stated the OpenAI announcement. “We realized that many inquiries users made with Operator were in fact better suited for Deep Research, thus we combined the strengths of both.”

OpenAI’s latest tool is a part of the rapidly expanding agentic AI landscape

OpenAI’s ChatGPT agent joins a variety of newly released agentic tools that can execute tasks on the user’s behalf. Although not a complete web browser, it operates similarly to Perplexity Comet’s browser assistant. Anthropic also provides a utility referred to as “computer use,” which can take control of your cursor and write code. As models progress, they gain greater ability to carry out independent tasks. Web browsing is viewed as a crucial domain for AI labs to compete, with OpenAI, Anthropic, and Perplexity all offering functionalities, along with Google’s Project Mariner research prototype.

In