Anthropic dropped Cowork on Monday, and it’s the kind of release that makes you stop and think about where this whole AI agent thing is actually heading.
Cowork is essentially <a href="https://video.allwinchina.org/ai-tools/claude-code/" title="Claude Code review”>Claude Code stripped of the terminal and given a friendly face. For the past year, Claude Code has been a hit with developers who use it to automate coding grunt work. But Anthropic noticed something odd: people were using it for everything except coding. Vacation research. Slide decks. Email cleanup. Canceling subscriptions. Recovering wedding photos from a hard drive. Monitoring plant growth. Controlling an oven.
Boris Cherny, an engineer at Anthropic, laid this out on X. The common thread? The underlying Claude Agent is good enough that people wanted to use it for tasks that have nothing to do with software development. So Anthropic did the obvious thing: they built a version that doesn’t require you to know your way around a command line.
Cowork is a research preview, exclusive to Claude Max subscribers — that’s the $100 to $200 per month tier — and only available on macOS for now. You give it access to a folder on your machine, and it can read, edit, and create files within that sandbox. That’s it. No coding required.
What makes this interesting is the architecture. It’s not a chat interface where you paste text and get a response. Cowork operates in what Anthropic calls an “agentic loop.” You give it a task — say, “organize my Downloads folder” or “turn these receipt screenshots into an expense spreadsheet” — and it formulates a plan, executes steps in parallel, checks its own work, and asks for clarification if it gets stuck. You can queue multiple tasks and let it process them simultaneously. Anthropic describes the experience as “much less like a back-and-forth and much more like leaving messages for a coworker.”
That’s a fundamentally different interaction model from what most people are used to with AI. It’s not about generating text. It’s about getting stuff done in your file system.
Here’s the part that I find genuinely wild: the team reportedly built Cowork in about a week and a half, largely using Claude Code itself. That’s a recursive loop where AI tools are building better AI tools. It’s not science fiction — it’s happening right now, and it’s happening fast.
Now, the obvious question is trust. Giving an AI agent read and write access to a folder on your machine requires a level of confidence that most people don’t have yet. Anthropic is betting that the utility outweighs the risk, and they’ve limited the scope to a single designated folder to keep things contained. But if you’ve ever watched an AI hallucinate a convincing-looking but completely wrong answer, you understand why this makes people nervous.
Still, the direction is clear. The industry has spent the past year talking about LLMs that can write poetry or debug code. With Cowork, Anthropic is saying the real value is in an AI that can open a folder, read a messy pile of receipts, and generate a structured expense report without human hand-holding. That’s a bet on practical productivity over flashy demos.
Whether it works reliably enough for mainstream users is an open question. But the fact that they built it in ten days using their own tool tells you something about the pace of development in this space. It’s moving faster than most people realize.
Comments (0)
Login Log in to comment.
Be the first to comment!