
Claude Cowork arrives on web and mobile
Anthropic is bringing Claude Cowork to web and mobile, so work you start at your desk finishes on your phone. Here is what it means for UK SME teams.


Claude Cowork arrives on web and mobile
On 7 July 2026, Anthropic started rolling out Claude Cowork to web and mobile. You can hand Claude a task at your desk and collect the finished work on your phone.
Cowork is Anthropic's tool for handing whole tasks to Claude, rather than chatting message by message. Until now it lived on the desktop. This release lets the work follow you.
What changed
Three things are new. Together they change where and when the work happens.
First, your work follows you across devices. Sessions and files move with you, so you start on a laptop and check progress from a phone.
Second, the work continues in the background. Scheduled tasks now run even with no device online. Set client prep for 6am. Claude works through the threads and transcripts, builds the brief, and leaves the follow-up drafted but unsent.
Third, decisions still come to you. Nothing ships until you review and approve it. When a task needs your judgement, the question lands on your phone.
Chat and Cowork now share one home on web and desktop. Your projects and artifacts sit in one place across both.
The beta rolls out over the next few weeks. It starts with the Max plan, with more plans to follow. Anthropic has also doubled Cowork usage limits until 5 August.
Most Cowork users are not developers
Anthropic reports that more than 90% of Cowork use is not software development. The largest categories are business operations and content creation.
That matters for smaller businesses. This is a tool for your finance lead, your office manager, and your marketer. It is not only for technical teams.
The doubled usage limits are a deliberate nudge. Anthropic wants people to build the habit of delegating real work. Taking them up on it is sensible.
The harder question is about teams
For one person, this is a real gain. You delegate at your desk and collect on your phone. The friction drops.
Most work in a small business is shared, not solo. So the harder question is how a team uses Cowork together.
If every person builds their own private way of working, you get silos. Each person syncs their own files and runs their own sessions. Shared capability never forms.
My view is simple. AI should help a team work as one, not split it into private silos. Anthropic has not fully answered the team question yet.
This is the 80/20 point gecco makes on every engagement. The tool is the 20%. How your team agrees to use it is the 80%.
What this looks like in practice
Picture a 30-person firm. The operations lead schedules a Monday brief for 7am. Claude reads the weekend emails, updates the project notes, and drafts replies for review.
By the time the team logs in, the groundwork is done. A person checks each draft and sends it. The tool did the gathering; the people kept the judgement.
Now picture the same firm with no shared standard. Five people run five private setups. Nobody can pick up another's work. That is the outcome to avoid.
What to do this week
A note of realism
Background work is only as good as the context you give it. Vague instructions produce vague drafts.
Unattended tasks raise a governance question. Decide clearly what Claude may access and what it may not.
Shared use needs a shared standard, or it fragments. That is a people decision, not a software setting.
Next step
gecco helps UK SMEs put tools like Cowork to work across a whole team. The aim is shared capability, not a scattering of private habits. We start with your people and your process, then set the platform up around them. If you want a clear read on where your team sits with AI adoption, take the AI readiness assessment.

ChatGPT ads reach the UK this year
OpenAI is bringing ads to ChatGPT, with the UK named in the next wave of its pilot. They hit the free and Go tiers, so they land on the staff most teams rely on.

Anthropic pulls top AI models after US government order
Anthropic switched off its two most powerful AI models after a US government order. We explain what happened and why UK SMEs should not build critical work on the newest model.

