OpenAI is preparing to expand its agentic ecosystem with a Chromium-based browser that integrates ChatGPT Agent for automation. According to newly surfaced details, the browser—internally linked with the codenames “Perplexity Comet” and OpenAI’s Chrome alternative—will allow deeper control of web activity compared to the current cloud-based Agent mode.
How Agent works
ChatGPT Agent already exists as a mode that runs inside a virtual machine hosted in the cloud, powered by Linux and deployed via Microsoft Azure. In this setup, the Agent simulates a browser experience by taking screenshots of a virtual browser window and then performing tasks like filling forms, clicking links, or navigating between pages. However, it has clear limitations: it cannot directly control a user’s local browser, access open tabs, or interact with features such as extensions.
New browser integration
Evidence spotted by developer Tibor shows OpenAI wiring Agent mode to support two execution paths. The first is the existing remote cloud browser, while the second points to a first-party local browser—believed to be OpenAI’s upcoming Chromium-based client. A hidden “Use cloud browser” toggle and a user-agent string referencing “ChatGPT…Macintosh;…Chrome” suggest that the integration is only active on OpenAI’s native app or browser, with the fallback being the virtual browser in the cloud.
Why it matters
By embedding Agent directly into a local browser, OpenAI could enable automation that goes far beyond today’s screenshot-based approach. This would let ChatGPT directly manage tabs, carry out automated workflows, and potentially rival traditional browsers like Google Chrome. It also aligns with reports that OpenAI aims to bring browsing and ChatGPT interaction into a single interface, streamlining productivity tools.
OpenAI’s documentation has long hinted at a “visual browser” for agents, and Reuters previously reported that a Chromium-based OpenAI browser is in development. The move could reshape how users interact with the web, shifting from manual browsing to AI-driven automation. With ChatGPT Agent gaining the ability to run locally, OpenAI’s upcoming browser may directly compete with established players while setting the stage for agent-first browsing.
Discover the latest Business News, Sensex, and Nifty updates. Obtain Personal Finance insights, tax queries, and expert opinions on Moneycontrol or download the Moneycontrol App to stay updated!
