Anthropic gave AI access to your desktop

Anthropic has launched Claude Cowork, a general-purpose AI agent that can read, analyse, organise, and create files on a user’s computer.

It’s available as a research preview for Max subscribers on $100 or $200 per month plans.

Anthropic describes Cowork as “Claude Code for the rest of your work.”

It builds on Claude Code, but is designed for non-technical users rather than developers.

While Claude Code already works beyond coding tasks, its name and developer-focused interface have made it less approachable for everyday use.

Cowork is designed to work more autonomously, handling tasks like organising downloads, turning receipt screenshots into spreadsheets, and drafting documents from notes stored across a desktop.

Anthropic frames it as assigning work to a colleague, rather than chatting back and forth.

The tool was reportedly built in just over a week, largely using Claude Code itself.

Its launch places Anthropic in more direct competition with enterprise productivity tools such as Microsoft Copilot, with expectations that rivals like OpenAI and Gemini will follow a similar path.

Like other autonomous AI agents, Cowork carries security risks, particularly prompt injection attacks.

Anthropic has acknowledged these risks, advising users to limit access to trusted sources while noting that agent safety remains an active area of development.

The release has also raised concerns among startups, as Cowork overlaps with tools built to handle file organisation, document creation, and data extraction.

Some founders argue that specialised workflows and stronger user experiences may still offer a competitive edge.

Every startup that raised money for “file automation” just flinched. - MG