OpenAI on Tuesday announced ChatGPT Atlas, a new artificial intelligence-powered web browser.

The company said Atlas looks and feels similar to existing browsers, but is built around OpenAI’s generative chatbot ChatGPT. That means users will be able to access ChatGPT to create summaries, ask questions and complete tasks as they navigate across any page on the web.

“We think that AI represents a rare once-a-decade opportunity to rethink what a browser can be about,” OpenAI CEO Sam Altman said during a livestream on Tuesday.

The launch of Atlas puts OpenAI in direct competition with rivals like Google and Perplexity AI, which are jostling to attract users and win market share with their own browser offerings.

Perplexity rolled out its Comet browser for free earlier this month, and Google embedded a Gemini model within its Chrome browser in September.

Shares of Alphabet, Google’s parent company, closed down 2% following OpenAI’s announcement. 

Atlas users will see an “Ask ChatGPT” button on the top right corner of every web page, and clicking it will open up the chatbot in a companion sidebar.

Since ChatGPT will follow users across the web with Atlas, there is no need to copy and paste between tabs, Adam Fry, the product lead for Atlas at OpenAI, said during the livestream.

The browser is also equipped with memory, which will help it understand users and become more personalized over time, he said. If users forget where they saw a piece of information, for instance, they could use the homepage to type in a prompt like “search web history for a doc about Atlas core design.”

Atlas also has an agent mode, which means that ChatGPT can take actions for users and complete tasks like booking reservations, flights, shopping or editing a document.

“It’s got all your stuff, it’s clicking around for you, you can watch it,” Altman said. “You don’t have to, but it’s really — it’s using the internet for you.”

ChatGPT Atlas is available to macOS users worldwide, though agent mode is only available to Plus and Pro users. OpenAI said experiences for Windows, iOS and Android are coming soon.

“There’s a lot more to add. This is still early days for this project,” Altman said.

WATCH: Google adds Gemini to Chrome for all users in push to bolster AI search