The advent of agentic AI opens an unprecedented era where uses, economic models and digital sovereignty are completely redesigned.
This summer Openai has crossed a new border with the launch of Chatgpt Agent: a unified system capable not only to reason, but also to act. With him, artificial intelligence goes from an assistant role to that of an autonomous operator, capable of carrying out complex missions from start to finish. What DSIs and business directions still perceived yesterday as science fiction is now in the deployment phase.
If this innovation has gone relatively unnoticed, it nevertheless marks a rocking which redefines the entire technological landscape. The agentic AI is no longer content to answer: it executes. It thus upsets the economic models of the web based on traffic and advertising, by directly providing users with the responses they formerly came to seek elsewhere. A deep transformation, already underway, and which promises to impact both uses, business models and power balances between tech giants.
An unprecedented technological merger
Chatgpt Agent results from the combination of three major technological bricks developed by OpenAI:
Deep Research, launched in February 2025, which planned and synthesized data from hundreds of web sources to create complete reports.
Operator, presented in January, which allows the AI to act directly instead of the user to book, navigate or even write code.
Chatgpt, the powerful conversational agent of Openai.
This convergence creates a radically new experience where the user can make complex requests in natural language and see AI execute them from start to finish without human intervention. This paradigm shift announces the end of passive chatbots and the arrival of truly autonomous and operational assistants.
A strategic confrontation between tech giants
With Chatgpt Agent, Openai is not content to offer a technological innovation, she heads head -on Google. According to Reuters, Openai is preparing the launch in the coming weeks on its own AI web browser. Founded on Chromium, this browser intends to transform access to information. It could capture a significant part of the time spent online, to the detriment of the Google economic model based on advertising and traffic generated by Chrome. With more than 500 million weekly active users of chatgpt, the adoption potential is colossal.
But the battle is also played on another front. In this race for agentics, X (formerly Twitter) of Elon Musk could have a considerable advantage with Grok Heavy which explores an alternative approach: that of multi-agents. Several AI work in parallel on the same problem, compare their answers, and retain the best.
This innovative architecture could constitute a more robust alternative to the unified openai system. Grok Heavy establishes a new performance standard, reaching 50 % in the Humanity’s Last Exam test. But this technological advance comes up against two major obstacles: a subscription cost ten times higher than that of Chatgpt plus, and an image tarnished by several controversies, ranging from conspiracy to anti -Semitic remarks.
To a new digital order
Beyond technological performance, economic impacts are considerable. According to McKinsey, integration of AI into online research could increase user productivity by 20 to 30 %. Openai says that his agent can now carry out the work of a beginner analyst for a 500 fortune company.
But this transition is not limited to the company: it impacts the entire digital ecosystem. If traditional navigation is erased, it is the set of media, e-commerce platforms, and online services that will have to rethink their model.
The central question then becomes: who will control access to information tomorrow? The answer to this question conditions our digital sovereignty. And in this field, Europe is a worrying delay. Between prudent regulations (AI Act) and absence of technological champions, it risks becoming dependent on solutions that it will have neither designed nor supervised.
