Your AI Assistant Just Stopped Needing Your Laptop

What actually happened

Cowork is the mode inside Claude that's built for real work tasks like organizing files, drafting documents, researching, and running multi-step projects, as opposed to just chatting back and forth.

Until now, Cowork lived mostly on desktop. You'd open it on your computer, kick off a task, and it would run there. Close the laptop, and the session paused with it. The update changes that in two ways:

• Cowork now works on web and mobile, not just desktop. You can start a task on your computer and check on it from your phone later. Same account, same session, same files.
• Sessions run remotely, in beta. Once you kick something off, it keeps going even after you close your laptop. Scheduled tasks, the kind you set up to run every morning or every week, now run whether or not any of your devices are actually on.

Anthropic is rolling this out gradually, starting with people on the Max plan, with other plans expected to follow.

Why this matters more than it sounds like it does

Here's the thing about most AI updates: they're incremental. A slightly better model, a slightly cheaper price, a new button somewhere in a settings menu. This one is different because it changes a category of task you could reasonably ask AI to do for you.

Before this, a recurring AI task mostly meant something you had to remember to trigger. Even if you'd set it up to run automatically, it often needed something to be on: a computer left running, a tab open, a service pinging a server somewhere.

Now the task itself lives on Anthropic's servers, not your laptop. That's a small architectural detail with a real-world consequence: the AI stops being something you operate and starts being something that operates for you, on its own schedule, in the background, without you thinking about it.

Think about what that unlocks. A morning summary of your inbox that's actually waiting for you when you wake up, not something you have to remember to ask for. A weekly competitor check that runs every Monday whether you're at your desk or at your kid's soccer game. A report that's ready before your 9am, because it didn't need you to be awake to generate it.

Is this hype, or is it actually important?

I'll be honest about the limits here, because I think that matters more than the excitement. This is a genuinely useful infrastructure change, not a new brain. The underlying AI didn't get smarter this week. It just got freed from needing your device to stay on. That's meaningfully different from a new model release that changes what the AI can reason through or write. This is about reliability and reach, not capability.

It's also rolling out gradually. If you're not on a Max-tier plan yet, you may not have access right away, and beta means Anthropic itself is still working out the rough edges. I wouldn't build a mission-critical process around it just yet, but it's worth watching, and worth testing on something low-stakes. So: real, but not revolutionary. Useful, but not a reason to panic if you missed it.

What business owners should actually do with this

You don't need to overhaul anything. But if you've been putting off setting up a recurring AI task because "it needs my computer to be on" felt like an annoying catch, that excuse is gone. A few places this is worth trying first:

• A daily or weekly summary. Inbox, calendar, or a specific project's status, something you currently check manually every morning.
• Routine research. Competitor pricing, industry news relevant to your business, or a weekly roundup of a topic you care about.
• A recurring report someone on your team currently builds by hand. Not to replace them, but to see if the boring, repeatable 80% of it can run itself, freeing them for the part that actually needs judgment.

Start with one task. See how it goes for two weeks before you trust it with anything that matters.