When Anthropic launched Claude Cowork as a desktop application in January 2026, it was billed primarily as a tool for software developers. Six months later, the usage data tells a very different story — and Anthropic is responding accordingly.

On July 7, Anthropic announced it is expanding Claude Cowork beyond the desktop app to both web and mobile, bringing persistent AI agent sessions to wherever business users actually work. The rollout begins for Max plan subscribers and will expand to additional plans in the weeks ahead.

The catalyst for the expansion is striking. Anthropic pulled data from 1.2 million anonymised Cowork sessions across more than 600,000 organisations and found that more than 90 percent of sessions had nothing to do with software development. Users were running research tasks, drafting documents, preparing reports, managing workflows, and handling the kind of repetitive business operations that pile up across every team in every company.

What Changed

The web and mobile expansion does more than just change the screen you use to start a Cowork session. It changes what AI agency actually means in practice.

Cross-device continuity. A task started at a desk can now be checked on a phone and the finished output retrieved from any device — without any part of that workflow requiring someone to be sitting in front of their laptop. This closes one of the biggest gaps in how AI agents have actually been used: tasks would stall the moment the initiating device went offline.

True background execution. Cowork can now run tasks when no device is online at all. Users can schedule work for a specific time and have Claude execute it autonomously, returning completed results when they log back in. This shifts the dynamic from “AI assistant that needs supervision” to “AI agent that handles work while you focus elsewhere.”

Extended usage. Anthropic is doubling Cowork usage limits through August 5, giving enterprise users room to explore more ambitious use cases during the rollout period.

Why the Shift Matters

The original framing of tools like Cowork as “coding agents” reflected the early adopter base: developers building software. But that was always a narrow window into a much broader capability.

What most businesses actually need is something simpler and more valuable: the ability to hand off structured, repetitive work to an agent and trust it will be completed without constant check-ins. Research summaries. Compliance document reviews. Competitive analysis. Meeting preparation. Invoice processing. These are the tasks that consume hours every week and produce nothing that requires human creativity to execute.

The 90 percent non-coding figure from Anthropic’s own data makes this concrete. It also explains why expanding to web and mobile is the obvious next step. Business users do not sit at a desktop all day. They move between meetings, check results on their phones, and need work to continue in the background without them being tethered to a specific machine.

What This Means for Business

For companies evaluating how to get real value from AI in 2026, Claude Cowork’s expansion signals something important about where enterprise AI is heading.

The first generation of AI tools required users to actively prompt and supervise every step. The next generation — which Cowork now represents — involves AI agents that accept a task brief, execute it independently, and return results. The human stays in the loop for decisions, not for supervision of execution.

This is the pattern that creates actual time savings. Not AI that helps you write faster, but AI that completes work while you are doing something else entirely.

For organisations that have been watching AI deployment closely, the right question to ask is not “what can our AI tools do when we use them?” It is “what can our AI agents accomplish while we are not using them?”

The gap between those two questions is where the real productivity gains live.

Background on Claude Cowork

Claude Cowork launched as a desktop-only product in January 2026, part of Anthropic’s broader push to move Claude beyond conversational AI into persistent agentic work. The product allows users to assign multi-step tasks to Claude, which then plans and executes them autonomously rather than responding turn-by-turn.

The web and mobile rollout is the first major platform expansion since launch. Anthropic’s decision to lead with usage data — showing the breadth of non-developer use cases — suggests the company sees the business automation market as a primary growth vector, not just a secondary audience beyond the developer base that adopted it first.

Enterprise DNA’s own work building and deploying AI agent systems for businesses confirms this pattern. The organisations making the most progress with AI are not the ones with the most sophisticated technical teams. They are the ones that have identified the repeatable, structured tasks where agents can take ownership and let their people focus on the work that actually requires human judgment.

If you are ready to explore what AI agents can handle in your business, we help organisations map those opportunities and put the right systems in place. Start with a conversation here.