AI News Today
← All episodes
Episode 1 · March 26, 2026 · 12:54

Claude AI's Computer Takeover Just Started

Claude AI Can Now Control Your Computer: The End of Boring Work?


Anthropic just released a massive update allowing Claude to directly control your mouse, keyboard, and browser to complete tasks. Learn how the new Computer Use and Dispatch features turn your phone into a command center for your AI assistant while keeping your files secure.


00:00 - Intro: AI Computer Takeover

00:39 - How Claude Controls Your Screen

01:09 - Dispatch: Control Claude From Your Phone

03:25 - Current Limitations & Pricing

05:50 - The Big Picture: AI Agent Trends

06:52 - Best Use Cases for Creators & Agencies

09:05 - Claude vs. OpenClaw: Which is Best?

10:34 - Self-Improving AI: The Next Frontier

Full transcript

Claude's AI computer takeover just started, and I don't mean in a scary way, I mean it in a your whole workflow just changed overnight kind of way. Yesterday, March 23rd, Anthropic dropped something that most people are still sleeping on. Claude can now use your computer to complete tasks. It points, clicks, and navigates your screen on your behalf.

It opens up files, it uses the browser, it runs tools, no setup required. That's it. That's a shift. You're not chatting with an AI anymore, you're handing it the keys to your computer.

Here's how it actually works. Claude will reach for the most precise tool first. It starts with connectors to services like Slack or Canada. But when there isn't a connector, it's actually going to go straight to directly controlling your browser, mouse, keyboard, and screen to complete the task with Claude co-work.

So it doesn't just guess, it tries the smart way first, and if that doesn't work, it looks at the screen and then it figures it out. This is an AI employee working 24-7 for you. And here's the part that made my jaw drop. There's a second piece to this called dispatch.

Dispatch lets you assign Claude tasks from your phone. You can assign a task on your phone and then turn your attention to something else and then open up the finished work on your computer. Think about what that means. You're on the train.

You send a message to Claude. You get off the train. The work is done. A content agency could, for example, text Claude, pull last week's analytics, write a performance summary, and then drop it in the client folder.

And then they come back to a finished document. And that's what just dropped yesterday. Now, let me give you the real context here, because this matters. A year ago, Claude was just a chat window.

You typed in, you got text back. That was it. Then came co-work. Claude running on your desktop, managing files, reading documents, doing longer tasks.

That was step two. Now we have computer use plus dispatch. So you can tell Claude to automatically check your emails every morning or pull some metrics every week. And you could, for example, spin up a session to create a report or, for example, pull a request whilst you're not there.

This is a totally different category of tool. It went from answer my questions to do my work. Step one, step two, step three, each one bigger than the last. And here's the thing nobody's talking about.

This is not just for developers. In fact, it's the opposite. Computer users describe their tasks. Claude autonomously plans the steps, opening applications, operating browsers, and handling files, including mouse clicks and keyboard inputs.

And before each execution, the interface displays the operation plan, waits for user confirmation. You can stop it at any time. You stay in control. It asks before it touches anything new.

You can pause it. You can stop it. It's not just running wild on your machine like a AI agent lunatic out of control. It's more like a very fast, very capable AI assistant who checks in with you before making big moves.

Now, I want to be honest with you about the limitations because this is still early. Anthropic said computer use is still early compared to Claude's ability to code or interact with text. Claude can make mistakes. Its tasks might require multiple attempts.

Its screen-based operations are slower than direct API integrations. And right now, it's available on Mac OS only for Claude and Mac subscribers. Windows users are still waiting for this. Pro is $20 a month.

Max runs from $100 to $200 per month. So it's not for everyone yet, but the direction is obvious. Every month, this thing gets better. Every month, the price comes down and every month, more people get access.

The people who understand this now are the ones who will move fastest when it opens up fully. Which brings me to this. If you actually want to get ahead of this and not just watch from the slide lines, but actually join and interact with this stuff, then join the AI Profit Boardroom. It's a community of 2,600 business owners learning exactly how to use tools like this in their real work.

You get four weekly coaching calls, daily step-by-step video tutorials, 30-day roadmaps and prompts for everything, plus a local map so you can connect with people in your city who are doing the same thing as you. Support 24-7 too, because there's always someone online. And you can check it out, link in the comments description or go to the AIProfitBoardroom.com. Back to the story.

Cloud Dispatch enables simultaneous management of multiple workflows, email automation, data organization, content projects, all from your phone. And here's what makes this different from everything else out there. The work still runs on the desktop computer. Local files still stay there.

The phone just becomes the place where work gets designed, checked and picked up later. The files never leave your machine. No cloud processing for sensitive stuff. That matters for anyone building and running a real business.

A freelancer managing multiple clients, an ecommerce store pulling daily reports, a creator batching content research. These are the people this was built for. Now let's talk about the bigger picture here. Because this isn't happening in isolation.

Anthropic has shipped a major update roughly every two weeks in 2026. Sometimes even more frequent than that. Each one has disrupted a different industry. First came Opus 4.6.

Then came co-op plugins, doing the work of attorneys and financial analysts. Then a cyber security tool, inside cloud code, that outperforms anything else on the market. And now this. Computer use, dispatch, your phone as a command center for the most powerful AI on earth.

A company owner predicts 40% of enterprise applications will include task specific AI agents by the end of 2026. And we're not even halfway through the year yet. The companies and freelancers and creators who are learning how to direct these agents right now, they're not going to be replaced by AI. They're going to be the ones running it.

So what do you actually do with this information? If you're a creator or agency owner, start thinking about which repetitive tasks eat your time every week. That could be client reports, inbox management, pulling data, organizing files. Those are exactly what this is built for.

If you're on a team, start paying attention to which tools you use every day that don't have good integrations. Claude can now operate those tools directly with computer use without waiting for someone to build a connector. And if you're on Windows, keep watching. Anthropic says it will continue to update based on use of feedback.

Bear in mind, Microsoft have just released Claude co-work Microsoft 365. And this basically does the same thing, but on Microsoft. So I would expect Claude co-work and computer usage to become even more powerful on Windows 2. Mac is the research preview.

The Windows version is coming soon. The question isn't whether this changes how works gets done. The question is whether you're ready when it fully lands. This is the takeover, not of jobs, not of your life, of the boring, repetitive, low value stuff that fills your day and stops you doing the work that actually matters.

Claude is sitting on your desk right now waiting to start. The only question is, are you going to let it? Now, if you want to be the type of person who knows how to use this stuff before everyone else catches up, come join us in the AI Profit Boardroom at AIProfitBoardroom.com 2600 members are already inside their daily tutorials, weekly coaching, 30 day roadmaps, real support, real people. There's a lot of great people inside there.

Now, what I would also recommend here is use this with caution because whatever you point computer use at, it can, for example, log into your accounts on your browser. It can use your social media accounts if you're not careful. So just be very careful where you post the permissions. And bear in mind, this is a much simpler, less technical version of something like OpenClaw.

Now OpenClaw is basically, I would say you've got Claude, which is the more enterprise level, more secure, less customizable version, but more safe version of using something like Claude and OpenClaw. But then you've got AI agents like OpenClaw and they have the fun factor. They're far more customizable, they're more technical to set up, but that's almost part of the reward when you're using these tools. Now, if you have time, check them both out.

If you don't have time and you're someone who's non-technical, then go with using Claude co-work and computer use at. But if you are technical and you don't have much time, then I would actually go with OpenClaw. So it really depends where you want to go on this. The one thing that I would say here is, if you look at all of this, right, OpenAI, you've got Antigravity, you've got Google Gemini, you've got all these other AIs that are coming out.

But in terms of leading the pack from everything I've seen, all the updates that are coming out, Claude co-work is by far the best. It's by far the one that's leading the pack in terms of creating new updates, shipping new updates. And I think one of the biggest improvements that I've seen with Claude this year is that they're now self-iterating and most of the code they're creating is not from their developers. So the constraint is how fast the AI can improve, not how fast a team can iterate.

Let me give you an example. So Claude co-work itself, most of the code for that was actually created with Claude code, the AI. And so the AI is now self-improving. And we've seen this with Minimax as well recently.

If you've seen my other videos on Minimax M2.7, this is another self-improving agent. And so the feedback loops are getting faster. The improvements are coming quicker. And that means AI is getting more and more powerful.

If you look over the last 90 days, the trend and the trajectory is the Claude co-work is only improving. It's only getting better. And I would expect to see that directly with Claude this year. I think they're going to overtake OpenClaude.

I think more people are going to adopt it. It's already valued at $380 billion. I would only expect Claude's value to go up now. They've already become like the number one app that's downloaded in the app store in the US.

And so the trajectory of Claude, I think you can see it right now. It's like they're leading the pack. If you look right now, most people have, you know, a lot of people, especially on the consumer side, have kind of abandoned chat GPT. You know, they've been let down with GPT 5.

There was a huge quit GPT movement where 2.5 million people actually left chat GPT. And so Claude is going like this, chat GPT is going down. Gemini seems to focus more on like coding, not like day to day work. And also the other thing I was going to say about Claude co-work is like he's focused on non-technical people.

He's actually built for people who are like, some people might call them normies, you know, people who don't use AI day to day, people who've kind of skipped the last couple of years of evolution in AI, who don't watch videos like this. And so the trajectory of Claude is unbelievable. And if they've already done this in the first 90 days of the year, right, they've already launched mobile assets. They've already launched scheduled tasks.

They've already launched Claude co-work computer use. What is coming over the next nine months? That's what's terrifying. And that's what I'll leave you with is like, how fast is this going to move now?

Because it seems like it's speeding up. It seems like the feedback improvements are getting better and better. And it's only going one direction, right? It's not going to get worse.

It's only going to become better. So I would definitely keep tabs on Claude. I would definitely keep tabs on OpenCore as well, because that seems to be releasing new updates all the time. And this is one of the most exciting times that I've ever seen in AI.

So let me know what you think inside the comments. I'd love to know what you think, what you're expecting in the future and which AI you actually use. Let me know. Cheers.

More episodes

Browse all episodes →