Claude Cowork arrives on mobile and the web

Anthropic expands Claude Cowork to mobile and the web, a strong signal for the new generation of AI agents

Anthropic is broadening access to Claude Cowork on mobile and the web, according to a report by The Verge. Behind this announcement, it is not just a new entry point for the American publisher’s assistant. The move is more structural: it confirms Anthropic’s ambition to push Claude beyond the classic conversational chatbot, toward a continuous work tool capable of handling long tasks, operating asynchronously, and resuming a workflow from one device to another.

The issue goes far beyond ergonomics alone. Since the rise of mass-market conversational interfaces, competition between labs and software giants has gradually shifted. The challenge is no longer just to answer a question quickly, draft an email, or summarize a document, but to place an AI assistant at the heart of a real workday, with its interruptions, human approvals, context changes, and multi-screen usage. By opening Claude Cowork on mobile and the web, Anthropic is explicitly positioning itself within this transition.

The wording used by The Verge is important: Claude Cowork is moving beyond a more limited initial use to become accessible in much more everyday environments. This change brings the product closer to ordinary professional use, where people constantly move from a desktop workstation to a smartphone, from a moment of concentration to a quick check-in, from a task launched in the morning to picking it back up at the end of the day. The interest is therefore not only technical. It is also organizational.

For the AI agent market, this expansion comes at a time of sharply intensifying competition. OpenAI is pushing productivity-oriented uses, Google is multiplying Gemini integrations across its tools, and Microsoft continues to make Copilot a pillar of its professional offering. In this context, Anthropic is seeking to assert a specific proposition: an assistant designed to work over time, with a stronger continuity logic than a simple one-off interaction.

For French-speaking companies, this development deserves attention. The promise of a persistent assistant, available across multiple surfaces and suited to long tasks, addresses very concrete needs: tracking files, preparing summaries, coordinating writing work, resuming interrupted analyses, or progressively processing information spread over time. If conversational AI has already found its place in many exploratory uses, the next step is to become a true work companion. That is precisely the ground Anthropic is now seeking to advance on.

From chatbot to work agent: the context of the announcement

To measure the significance of Claude Cowork’s arrival on mobile and the web, we need to go back to the broader evolution of AI assistants since the end of 2022. The first wave of mass adoption was based on a simple pattern: a dialogue box, a prompt, a response. This model was enough to transform usage, particularly in writing, information retrieval, ideation, and programming support. But it also revealed its limits as soon as one moves beyond the short, isolated use case.

In the reality of work, tasks are rarely linear. They stretch over several hours, sometimes several days. They involve back-and-forth, intermediate approvals, interruptions, changes of medium, and multiple contexts. An employee may start an analysis on a computer, review it on a smartphone while commuting, then resume it later from a browser. Yet the classic chatbot interface remains poorly suited to this continuity, unless one stacks up discussion threads that eventually lose readability.

It is in this context that the notion of an agent or a more persistent work assistant emerged. The term is often used very broadly in the industry, sometimes in a marketing sense, but here it refers to a simple idea: a system that does not merely respond to an instant command, and can instead continue, organize, or resume work over time. The key point is not autonomy alone, but continuity.

Anthropic, founded by former OpenAI executives, quickly established itself as one of the central players in this new phase. The company became known for its Claude family of models, with positioning often associated with safety, alignment reliability, and the quality of responses on complex tasks, particularly in text analysis and professional use. Over time, Claude has gained visibility among companies, product teams, developers, and knowledge workers.

The move of Claude Cowork to more universal surfaces such as the web and mobile therefore fits into a logical trajectory. If Anthropic wants its product to truly support daily work, it cannot remain confined to too narrow a framework. Multi-device availability is an almost minimal condition for an assistant to become a production tool rather than a technological demonstrator.

The Verge’s reading also highlights two particularly revealing elements: the focus on long and asynchronous tasks, and the ability to resume work from one device to another. These two dimensions alone summarize the paradigm shift. We are no longer talking only about a model that can generate text, but about a product designed to fit into the real rhythm of digital work.

This development also echoes a shift in the center of gravity of competition. The first months of the generative AI battle were dominated by the raw quality of models and the novelty effect. Now, questions of interface, contextual memory, task persistence, human-machine collaboration, and integration into tools are becoming decisive. The announcement around Claude Cowork should be read in this light: as a product step aimed at making the assistant more livable on a daily basis.

What Anthropic is specifically announcing, according to The Verge

According to The Verge, Anthropic is launching Claude Cowork on mobile and the web. That is the core of the information: the product, previously used in a more limited setting, is becoming available on platforms that are much more natural for professional and personal use. This expansion is presented as a way to make Claude Cowork more accessible and more consistent with how users actually work.

The functional positioning being highlighted is clear. Claude Cowork is geared toward long and asynchronous tasks. This detail is essential, because it distinguishes the product from a simple chat session. An asynchronous task, in this context, is a task that does not require continuous and immediate interaction between the user and the AI. It can be launched, interrupted, resumed, enriched, corrected, or finalized later. This is a mode of operation much closer to office practices, project management, or distributed intellectual work.

Another promise being emphasized is the possibility of resuming work from one device to another. Here again, the announcement is less trivial than it may seem. Synchronization across platforms addresses a strong expectation among professional users. A productivity tool that does not follow the user through movement or context changes remains partially stuck in a demonstration logic. By contrast, an assistant that preserves the thread of a task between mobile and browser begins to behave like a persistent work layer.

It should be noted that The Verge presents this development as an expansion beyond a more restricted initial use. Even without overinterpreting this point, it means that Anthropic is seeking to move Claude Cowork out of an experimental or more tightly controlled perimeter and place it in a broader usage environment. This is often an important stage in the maturation of an AI product: moving from a demonstration or limited access to more general availability, with stronger expectations around ergonomics, reliability, and continuity.

This announcement is not described as a simple interface variation. It carries a product vision. The web and mobile are not secondary channels here, but the two most decisive surfaces for establishing a habit. The browser remains the main workspace for a large share of knowledge professions, while the smartphone has become the tool of choice for consultation, validation, monitoring, and transition. If Claude Cowork wants to support long tasks, it must exist on both of these planes.

The choice of the name itself deserves attention. The term Cowork suggests a logic of collaboration, co-execution, or shared work with the assistant. Without extrapolating beyond what The Verge reports, the product intent seems clear: to make Claude a work partner that can keep moving forward with the user, rather than a simple engine for one-off answers.

For Anthropic, the challenge is twofold. On one hand, the company is seeking to make its offering more concrete and more everyday. On the other, it is seeking to differentiate itself in a landscape where almost all major players now promise agents, copilots, or intelligent assistants. In this competition, the ability to support multi-step, multi-device work may become a selection criterion as important as the linguistic quality of the model itself.

Why this expansion matters in the AI agent battle against OpenAI, Google, and Microsoft

The opening of Claude Cowork on mobile and the web does not come in a competitive vacuum. It is part of a sequence in which all the major generative AI players are trying to move beyond the standalone chatbot format to build productivity assistants more deeply rooted in workflows. On this front, Anthropic is neither alone nor unquestionably ahead, but the announced expansion constitutes a clear strategic signal.

OpenAI played a central role in the global spread of conversational interfaces, before evolving its offering toward increasingly work-oriented uses, software assistance, and task automation. Google, for its part, quickly repositioned Gemini as a cross-cutting building block of its products, with a logic of integration into its office and information ecosystem. Microsoft, finally, understood very early the value of making Copilot a layer of professional assistance connected to the tools companies already use every day.

In this landscape, the question is no longer only which model responds best, but which product fits best into professional life. Announcements around AI agents are multiplying, but many still remain abstract for the end user. What gives value to an assistant is not only its ability to produce convincing text in one go. It is its ability to follow a file, resume work, remain useful despite interruptions, and exist where the user is.

From this point of view, Anthropic’s announcement is consistent with a market evolution. Mobile, long seen as secondary for advanced generative AI uses, is becoming an essential link. Not because it would be the ideal environment for producing long content, but because it is the tool of continuity. It is on mobile that people reread, validate, check progress, relaunch a task, or capture an idea while on the move. The web, meanwhile, remains the main production space. Together, the two form the minimal duo of a modern work assistant.

Anthropic is playing a different card here from that of a historic office software player like Microsoft. Where Microsoft has an immense distribution advantage through its enterprise software, Anthropic must persuade through the quality of the experience and the relevance of the product itself. Opening Claude Cowork on universal surfaces moves in that direction: it reduces access friction and brings Claude usage closer to ordinary digital habits.

Against OpenAI, the issue is also sensitive. The two companies are often compared on the quality of their models, their safety strategy, and their ability to attract developers as well as businesses. But the competition is now also being played out in the form of the product. An assistant that can move across contexts, preserve a thread of work, and adapt to professional rhythms can become more valuable than a simple conversational engine, even an excellent one. By opening Claude Cowork more broadly, Anthropic is showing that it intends to compete on this ground.

Google, finally, is pushing another adoption model: deep integration into tools that are already massively used. That gives Gemini considerable distribution power. For Anthropic, the challenge is therefore to offer an experience distinctive enough to compensate for the absence of such a mass-market software base. The implicit bet is that the quality of work companionship, particularly on long and asynchronous tasks, can become a competitive advantage in its own right.

Another often underestimated point should also be emphasized: the battle of agents is not only technological, it is cognitive. Users must learn to delegate differently. They are no longer just asking for an answer, they are entrusting a process. The more a product facilitates this shift, the more it increases its chances of adoption. By making Claude Cowork available on mobile and the web, Anthropic is narrowing the gap between the promise of the agent and its concrete use. That is precisely what providers of productivity-oriented AI are seeking today.

What this changes for French-speaking companies and professionals

For the French-speaking professional audience, the interest of this announcement is immediate, even if its exact scope will depend on access conditions, the languages actually supported in everyday use, and the internal policies of each organization. The central point lies elsewhere: the expansion of Claude Cowork to mobile and the web brings the tool closer to a realistic enterprise scenario, in which AI is no longer used only to test ideas, but to support continuous work.

In many French and European organizations, generative AI was first adopted cautiously, often for peripheral uses: rewording, summarization, brainstorming, translation, writing assistance, note preparation. This first phase helped familiarize teams with conversational models, but it also showed their limits. An isolated chatbot remains useful, but it struggles to become a true process tool if it cannot follow the user over time and across devices.

The promise of a persistent assistant is therefore particularly relevant for knowledge professions: consulting, marketing, communications, legal, HR, product, research, analysis, support, documentation, technical writing. In these environments, work is rarely built in a single interaction. It is made up of iterations, trade-offs, and resumptions. The ability to launch a task, let it live asynchronously, then find it again on another medium corresponds to a very concrete reality.

For French-speaking companies, this also raises the question of usage governance. The more persistent an AI assistant becomes, the closer it comes to an operational role. That implies stronger requirements in terms of confidentiality, data management, action traceability, human validation, and compliance. Even if the announcement relayed by The Verge focuses on mobile and web availability and on long tasks, it is part of a dynamic that directly concerns IT departments and innovation leaders.

In France as elsewhere in Europe, the adoption of a work agent is not determined solely by its performance. It also depends on its ability to fit into a demanding regulatory and organizational framework. Companies want productivity gains, but they also want to avoid opaque uses, context errors, or the scattering of sensitive information across multiple tools. The more continuous AI becomes, the more the question of control becomes structural.

There is also a cultural issue. The French-speaking market has sometimes approached AI chatbots with a mix of enthusiasm and skepticism, particularly in large organizations where proof of value must be tangible. A tool like Claude Cowork, if it delivers on its promise of continuity across devices and management of long tasks, may speak more to decision-makers than an assistant limited to spectacular but occasional interactions. The gain is not only in the quality of an answer, but in the reduction of friction over the course of a workday.

Another important point: mobile changes the nature of the relationship with the assistant. In many professional environments, AI use is still associated with a dedicated moment in front of a computer. Opening up to smartphones instead makes it possible to integrate the assistant into in-between moments: travel, waiting, quick validation, checking a progress status, collecting an idea, reading a summary. This granularity of use is decisive for a product to become truly everyday.

For SMEs as well as large groups, the potential interest is therefore less spectacular than operational. An assistant that supports a file over several hours, follows the user from one medium to another, and makes it possible to pick up the thread without starting from scratch meets a simple need: saving time without breaking work continuity. That is precisely what distinguishes a conversational gadget from a credible productivity tool.

A step toward persistent assistants, with long-term implications for the whole sector

The announcement around Claude Cowork on mobile and the web should finally be read as a long-term marker. The generative AI sector is entering a phase in which value will no longer be captured only by the best models, but by the best usage systems. In other words, the question becomes: how do you turn conversational intelligence into personal or collective work infrastructure?

On this front, the approach highlighted by Anthropic is revealing. By emphasizing long tasks, asynchrony, and cross-device resumption, the company seems to be betting on a gradual but deep evolution of behaviors. The AI assistant is no longer summoned only when there is a question. It becomes a software presence to which part of the follow-up, organization, and intellectual execution is delegated. This is a more significant shift than it may appear.

If this logic takes hold, several consequences are likely for the market as a whole. First, interfaces will have to evolve. The single-thread discussion model quickly shows its limits when it comes to managing several tasks in parallel, tracking progress states, or resuming old work. Next, synchronization across devices will become a standard requirement, not a differentiating feature. Finally, the boundary between personal assistant, collaborative tool, and productivity platform could become increasingly blurred.

For Anthropic, the challenge is to turn Claude into a credible work environment before the competition locks in usage. OpenAI, Google, and Microsoft each have considerable strengths, whether in brand recognition, distribution, or software ecosystems. Opening Claude Cowork on universal surfaces is therefore also a way of occupying the field, of showing that Anthropic does not want to remain confined to the role of provider of a high-quality model, but intends to exist as a publisher of work products.

For users, the promise is appealing, but it also raises a question of cognitive dependence. The more persistent an assistant becomes, the more it can shape the way one structures tasks, notes, priorities, and trade-offs. The benefit in fluidity may be real, but it comes with a subtle shift in the center of gravity of intellectual work. Companies will therefore have to arbitrate between efficiency, control, and team autonomy.

In the French-speaking context, this perspective is particularly important. European organizations often take a more cautious approach to digital infrastructures, especially when they touch on data, compliance, or sovereignty. If persistent assistants become the standard for AI-assisted work, the selection criteria will not concern only the power of the model or the quality of responses in French. They will also concern governance, portability of uses, integration into existing environments, and the ability to maintain clear human oversight.

The significance of the announcement relayed by The Verge therefore lies in what it reveals about the direction the market is taking. Claude Cowork on mobile and the web is not a simple cosmetic extension. It is another indication that the next AI battle will be fought in the continuity of work, not only in the brilliance of demonstrations. The products that win will probably be those that can support the user over time, without trapping them in a one-off interface or forcing them to constantly restart the same contexts.

For Anthropic, this step can strengthen Claude’s credibility as a next-generation productivity assistant. For competitors, it confirms that the race for agents is shifting toward much more everyday uses. And for French and European companies, it offers a fairly concrete glimpse of what AI in the office could become in the coming years: less a chatbot consulted on demand, more a persistent software colleague, present on the web as well as on mobile, capable of picking up the thread of work where the human left it.