With ChatGPT AtlasOpenAI opens a new front with its conversational browser. Sixteen years after the launch of Chrome, OpenAI adopts the same logic as Google by making its flagship product, ChatGPT, the heart of the online experience. During a livestream last night, Sam Altman and his team presented their very refined browser which leaves room for a single field “Ask ChatGPT or type a URL” and a clean welcome area displaying suggestions, news links or automatable tasks.
The application is currently available on macOS and computers equipped with M chipsAtlas will soon arrive on Windows and mobile. A deployment which testifies to Sam Altman’s desire to maintain the initiative over all tech players, OpenAI is now developing at full speed and has not slowed down the pace since the start of the year.
📢 AI is profoundly transforming research and professional uses. Brands must now be visible in AI engines to remain present in decision-making processes.
✅ We offer an activatable offer to support your GEO strategy, relying on the authority and reliability of our media to strengthen your visibility.
Develop an offensive strategy today to strengthen your presence in AI!
👉 To receive our special offer, click here
A conversational and contextual browser
Ergonomically, Atlas retains the fundamentals of a classic browser with its tabs, favorites, autofill, but it introduces three new features:
- Cat everywhere : the “Ask ChatGPT” sidebar accompanies the user on each page to summarize, compare or explain the content displayed.
- In-line edition : in Gmail or any text field, ChatGPT can correct, reformulate or translate without copying and pasting.
- Navigation memory : the user can activate “browser memories” which remember the pages read, searches or subjects consulted to later find information or obtain personalized suggestions.
Agent Mode: the AI that acts in your tabs
The most spectacular function, the Fashion Agent allows ChatGPT to act like a human user : click, enter, transfer data between web applications.
During the demonstration, the agent:
- converted tasks from a Google Doc to the tool Linear,
- automatically added the ingredients of a recipe in a Instacart cart,
- left the final purchasing decision up to the user.
The agent is currently operating only in the browserwithout running any code or accessing the file system. The user can define whether he acts in mode connected or disconnectedand regain control at any time. Faced with this automation, OpenAI insists on security and human supervisionwhile recognizing that this delegation marks a turning point.
The crack in the Google model
Atlas threatens Google on its three pillars:
- The search : the conversational response replaces the results page.
- Attention : the user remains in the Atlas environment, without going through an external engine.
- The data : OpenAI directly captures navigation intentions and context.
For Google, this disintermediation represents a systemic threat, if consultation and action take place via ChatGPT, ad clicks, display and attribution collapse.
Atlas reprograms the web and making its site understandable by AI will be the new challenge for brands
With AtlasOpenAI is reshuffling the cards for brands. The challenge is no longer just to attract Internet users, but to make their sites readable by artificial intelligence whose reading grid is no longer based on navigation, but on the semantic understanding of the content. After spending years over-optimizing their interfaces, whether it be button placement, contrast, user journey, micro-interactions, designers will have to integrate a new dimension of design with machine interpretability.
A market already changing
Microsoft has integrated Copilot into Edge, Google is preparing a version of Chrome with Gemini agentic featuresand emerging players like Perplexity Or Bow are already testing AI-first interfaces.
But OpenAI has an advantage with 700 million weekly active users of ChatGPT who will not fail to test and if necessary adopt Atlas, which will become their natural gateway to the web.
