Claude Dispatch: Claude's Biggest AI Update This Week

Claude Dispatch, Claude's biggest AI update this week.

Imagine you're sitting at your desk on a Tuesday morning. You've got a meeting in 20 minutes. Your inbox is 47 unread emails. You've got a report due by the end of the day, a spreadsheet you haven't touched, a presentation that needs to be built from scratch, and three client files that need to be sorted and organized before your call this afternoon. That's a normal Tuesday for most people who are running a business, growing a brand, managing a team, or just trying to stay on top of the work.

Now, imagine something different. Imagine you pick up your phone while you're still in bed at 7 a.m. You type four sentences into an app. You put your phone down. You make your coffee. You take a shower. Maybe you go for a walk. And by the time you sit down at your desk, the report is written. The spreadsheets built. The presentation is done. The files are organized. The emails are triaged and drafted. You didn't do any of that. Your AI did. Not just answering your questions, not just giving you suggestions, not just generating text you have to copy and paste somewhere else. Actually doing your work on your computer in your folders, creating real files that are sitting there waiting for you.

This is what happened today. Anthropic just shipped something called Dispatch for Claude Co-work. I want to talk to you about why this is one of the most significant AI updates we've ever seen in 2026, and why it changes things for, and what it means for the way every single one of us works going forward. Let's get into it.

First, let me tell you exactly what Claude Dispatch is, because the name alone doesn't do it justice, right? Dispatch is a new feature inside Claude Co-work. It gives you one persistent, continuous conversation with Claude that runs on your computer. You can message it from your phone at any time. You can come back to finish work later. Read that again. You message it from your phone and you come back to finish work later. This is not a chatbot. This is not a back-and-forth conversation where you get a reply and then have to do something with the reply. You assign Claude a task, go do something else, and then come back to the finished work.

Claude runs on your computer with access to your local files, connectors, and plugins. And the messages you... the result when it's done. That's the thing that most people are going to miss when they first hear this. It runs on your computer. Your files stay on your machine. You're not uploading anything to the cloud. You're not giving external server access to some private documents. Claude runs code in the sandbox on your machine. Your files stay local. You approve what Claude touches before it acts. And that last part matters too, right? You approve what Claude touches. This is not an AI that just runs wild through your whole computer. You tell it what folders it can access. You tell it what to touch. You're in control. It just does the execution.

Now, let me zoom out for a second, because some of you might be hearing this and thinking, "Okay, cool. Another AI tool. I already use ChatGPT. I already use Claude. What's different here?" Here's what's different. And there are three types of AI tools in the world right now.

The first is a conversation tool. You ask a question, you get an answer. You ask it to write something, it writes it. You copy and paste the output, you go use it somewhere else. That's ChatGPT. That's regular Claude in the browser. That's Gemini. They're useful, but they're advisers, not workers.

The second type is a coding tool. So, Claude Code, GitHub Copilot, Cursor. These are built for developers. They can actually implement code, talk to your computer's file system, write software, but they require a terminal. They require a technical level of confidence that most people don't have. Most of the people watching this, most business owners, content creators, marketers, coaches, freelancers, they're not living in a command line.

Claude Co-work sits in the middle. It's a desktop agent designed for non-programmers who want an AI system with files, documents, and workflows. Think of it as having an AI co-worker who can actually do your busy work. Not just advise you on it. Not just generate text you have to manually implement. Actually go into your folders, read your files, build the deliverables, organize a mess, deposit the finished work into the folder you told it to use. Claude doesn't just give you text responses in the chat window. It writes directly into your folders. When you ask for a presentation, you get an actual PowerPoint file sitting in your directory. When you need some sort of model, an Excel spreadsheet appears. That's a distinction that most people don't fully understand. The output isn't a chat message you copy and paste. It's an actual file, a real PowerPoint, a real spreadsheet, a real Word document sitting in the folder you told it to put things in. Ready to open, ready to edit, ready to send.

And now with Dispatch, you don't even need to be at your computer to start this process. Let me give you a really concrete picture of what this looks like in practice. You open up Co-work and dispatch five tasks in roughly five minutes. Then you walk away, meditate, make breakfast, do a yoga workout, and return to find deliverables waiting. That's from a doctor at Duke University, an orthopedic surgeon, right, who's running this company. And he said he dispatches five tasks before 6:15 a.m. and comes back to finished work. Now, that guy has more on his plate than anything, you know, all sorts of responsibilities, a company to build, academic public papers to publish, research to review, and he's using this tool to run five parallel workflows before breakfast.

If that sounds extreme to you, let me bring it down to earth. Let's say you're a freelancer or a small business owner or someone who's running a content business. Your morning might look like this. You wake up, you pick up your phone, you open up the Claude app, you type, "I need you to pull the data from the spreadsheet in my reports folder, write a summary of last week's numbers, and format it into a one-page PDF I can send to a client." You put your phone down, you make some coffee. By the time you're back, Claude has pulled data from your local spreadsheet, compiled a summary report, and delivered it. And if you want, it can also search your Slack messages and email, then draft a briefing document or build a formatted presentation built from files in your Google Drive. That's not a demo. That's what this tool actually does.

And look, I want to be honest with you, because I always do, right? This is still a research preview. That means it's early. Dispatch is currently slow, and it's about a 50/50 shot whether what you try will work. That's not good enough to rely on when you're away from your desk. So, I'm not going to stand here and tell you this is perfect. It's not. It's an early version. Anthropic themselves said as much. But here's what I'll tell you. Every transformative product that exists today started as a research preview. It was a 50/50 shot before, right? Gmail was invite-only for two years and didn't even work half the time. The iPhone didn't have copy and paste in its first version. The first version of Uber was breaking constantly and was only available in San Francisco. The question is never, is it perfect today? The question is, what does the direction of travel tell us about where this is going? And the direction of travel is unmistakable.

Co-work launched January 12th, 2026, as a research preview for Mac subscribers. Four days later, Pro users got access. A week after that, Team and Enterprise. Then plugins came out on January 30th. Now, Windows support, all in under a month. That is the pace of iteration. That is how fast Anthropic is shipping on this product. And now, less than 70 days after launch, they've shipped Dispatch, mobile control of a desktop AI agent. That's not a company that's slowing down. This is a company that's sprinting.

Now, here's the thing I keep saying, and I'll say it again because it matters. Most people are thinking about AI wrong. They're thinking about it as a search engine that's just gotten better. They're using it to answer questions. They're using it to generate text. They manually implement. They're treating it like a smarter version of Google. That's not what this is anymore.

What Dispatch represents is the shift from AI as a tool you use to AI as a worker you manage. Think about what that actually means for your business. If you have a team of four people, and one of them is spending six hours a week on reporting, organization, formatting, document creation, administrative work, what happens to your output when AI handles all of that? You don't fire that person, you free them, right? You free six hours of their week. You free their brain for the work that actually requires a human brain. That's a game. It's not about replacing humans. It's about what becomes possible when the drudgery is automated.

You can dispatch multiple tasks simultaneously. Have Claude organize your downloads folder whilst it drafts email responses, whilst it researches a topic, and each one runs independently. Parallel workflows running simultaneously whilst you do other things. That is not the same category of productivity tool as a better word processor or faster research engine. This is a different kind of leverage.

Now, let me put some real numbers on this so it doesn't feel abstract. The average knowledge worker spends two and a half hours a day on email, another hour and a half on document creation, another 45 minutes organizing and searching for files. That's close to five hours a day on work that produces no direct value. Admin work, not output work, right? Five hours a day. That's more than half a standard day that most people spend on things that an AI agent can do.

Now, and I know somebody's watching and thinking, "Okay, but can it really do that? My situation is complicated. My files are a mess. My workflow is unique." Here's what I'll say to that. Co-work shines when your outputs and your inputs are incomplete or disorganized. If you have rough notes spread across text files, markdown documents, or Word drafts, Claude can synthesize them into a first-pass report. It identifies overlapping ideas, organizes sections logically, and produces a readable draft you can refine further. It is specifically good at messy stuff. It is specifically designed for the scenario where your stuff is everywhere and you need someone to make sense of it. That's not a flaw in how you work. That's a use case this was built for.

Now, let me talk about Dispatch specifically for a second, because I want to make sure you understand the specific upgrade that happened today. Before Dispatch, Co-work already existed, and Co-work was already powerful. You could open your desktop, give Claude a task, let it run, and come back to finished work. But here was the problem. You have to be at your computer to start it. You had to sit down, open the app, type in the task, and then step away. That meant the only time you could set work in motion was when you were already at your desk, which means you're already in work mode, which means the benefit of walking away was smaller because you'd already started your workday.

What Dispatch does is break that constraint entirely. From your phone, you can dispatch tasks that use everything on your desktop, including things you can't open on your phone. Your phone becomes the remote control for your entire AI-powered desktop setup. You're on the train. You remember you need a report built. You pull out your phone, open up Claude, type a message, boom, task, it's done. Right? By the time you get to the office, the report is ready. You're at the gym. You think of three things you need organized before your afternoon meeting. You message Claude from your phone whilst you're on the treadmill. By the time you've showered and dressed, it's done.

Right now, here's something interesting as well. So, Felix Reeseberg, the engineer at Anthropic who shipped this, said it himself. It feels magical to give Claude a mission on your computer and get occasional updates like creating reports from internal dashboards or finding a better seat on your next flight. Everything Claude can do on your computer files, browser tools, is reachable wherever you are. Finding a better seat on your next flight, for example. Or, you know, that one example jumped out at me because of how mundane it is. That's not a big, dramatic business use case. That's just a personal assistant task. A task that has no business eating up five minutes of your brain, but has always required you to be sitting at a computer actively doing it. Now, Claude does it whilst you're doing something else.

And this is where I want to introduce a concept that I think is important for understanding where all of this is heading. The bottleneck in your business right now is almost certainly not ideas. You probably have loads of ideas, right? The bottleneck is execution time. The bottleneck is the hours in your day. The bottleneck is how many things you can physically touch and move forward in any given week. AI agents like Dispatch are systematically dismantling that bottleneck. Not by making you faster, by multiplying what gets done whilst you're doing other things. This is a genuinely different kind of leverage than anything has ever existed before.

Right? In the history of work, think about it like this. Before computers, the leverage available to a small business owner was basically just hiring more people. If you wanted more output, you needed more hands. The ratio of people to output was roughly linear. Then computers arrived, and a single bookkeeper could handle the work of, [snorts] ten bookkeepers with ledgers. A single designer with Photoshop could produce what an entire art department used to produce. The leverage improved, but it still required a human to be actively operating the computer.

Then the internet arrived, right? And suddenly a single person with a website could reach millions of customers, sell products around the clock, automate transactions, and do things that used to require entire departments. But even then, someone had to be online. Someone had to be actively managing the website, writing the emails, responding to customers, and creating the content.

What AI agents like Dispatch represent is the next step in the progression. Now, the computer can initiate work, implement work, and deliver finished output without a human operating it. The human sets a mission. The AI implements. The human reviews and approves. That is not a marginal improvement in productivity. That is a structural change in what a solo operator or small team can produce.

And here's where it gets uncomfortable for a lot of people, because this is not evenly distributed. The people who understand how to use these tools, who learn the workflows, who figure out how to give good instructions, who build systems around AI implementation, those people are going to pull away from the people who don't. Not slightly, by a lot. I keep saying this, I've been saying it for a while now. The gap between people who use AI well and people who don't is not a productivity gap, it's a leverage gap, right? And it's growing every single week. When someone who uses AI agents effectively can complete in one morning what used to take three days, and they're compounding that advantage week after week, the distance between them and someone working without these tools becomes enormous, fast.

And here's the part that I want you to really sit with. The people who are going to lose ground are not like the lazy ones. They're not like the, you know, the people who don't understand this stuff. They're not even the ones who refuse to learn things. They're the people who are so busy implementing the current workflows that they never have time to upgrade it. Their heads down, working hard, producing solid output, and every day they do that without upping their tools. So the gap gets a little bit wider. And that's not judgment, that's just physics, right?

Which brings me to something I want to mention briefly, because I think it's genuinely relevant to what we're talking about inside the AI Profit Boardroom. We spend a lot of time specifically on this, right? The question of how do you build workflows that use AI leverage, not just AI assistants. But there's a difference. AI assistance is using ChatGPT to help you write an email. It's useful, saves a few minutes. AI leverage is building a system where Claude automatically processes your inbox and drafts your responses, and organizes your files, and builds your weekly reports, and delivers finished work to you every morning before you've even touched a keyboard. There's a completely different order of magnitude. And the people inside the AI Profit Boardroom are figuring out how to build that second thing. So, if you want to go into that room and learn how to actually apply this stuff to your business, your career, your future, the link is inside the comments in the description, or you can just go to AIProfitBoardroom.com to get access. We'll come back to more of what Dispatch means for you practically, but first, I want to talk about something that I think is even more interesting than the feature itself.

What I want to talk about is the existence of this feature and how this tells us where AI is heading over the next 12 months. Because Dispatch didn't appear out of nowhere. It's the latest step in a very clear progression that Anthropic has been implementing, and understanding the progression helps you see where this ends up.

Step one was Claude the chatbot: question and answer, smart, capable, and useful. Step two was Claude Code, right? Developers could now give Claude code access to their computer and have it actually write and run code autonomously. This was a significant leap, but it was still aimed at technical users.

Step three was Co-work. Anthropic took the same underlying technology that powers Claude Code, the agentic feature, the ability to plan and implement multi-step tasks, and wrapped it in an interface that didn't require technical expertise. The entire Co-work feature was developed in approximately 10 days as a research preview, responding to users who were already pushing Claude Code beyond its intended use cases for tasks like, for example, vacation planning, document organization, and email management. Think about that. Users were already bending a developer tool to handle their life admin. They were so hungry for this capability that they were using a terminal-based coding tool to plan vacations. Anthropic saw that, built Co-work in 10 days, and shipped it.

Step four is Dispatch: mobile control of the desktop agent. The ability to set AI in motion from anywhere at any time without being physically present at your computer. What's step five? I'll tell you what I think it is. Scheduled autonomous work. With the introduction of scheduled tasks, Claude can complete something for you automatically, something that isn't possible in regular chats outside of Co-work. And that's already there. You can already schedule recurring tasks in Co-work. You can tell Claude, "Every Monday morning at 7:00 a.m., pull last week's data from my reports folder, compile it, and drop the summary in my briefing document." It's not theoretical, that exists today. The next step after that is AI that doesn't just respond when you trigger it, but monitors, updates, and asks on its own judgment within parameters you set. Imagine telling Claude, "If you see an urgent email from any of my top five clients, summarize it and send a text immediately." That's where this is going. We are moving from tools you use to workers you manage, to systems you run. And the businesses that are building on top of these systems right now, while most people are still treating AI as a better search engine, those businesses are going to be operating at a completely different level of efficiency in 12 months.

Now, let me talk proactively about this, and because I know some of you are listening to all of this and thinking, "Okay, I'm convinced this is important, but what do I actually do with it?" Let me break down this by who you are, right?

So, if you're a solopreneur or a freelancer, here's where Dispatch immediately moves the needle for you. The biggest problem is usually not capability, it's capacity. Right? You're great at your work. You just don't have enough hours to do the work, do the admin, do the business development, do the marketing, and do the client management all at once. Dispatch starts eating the admin category. You can create daily news or work briefs, draft email replies, delete unused subscriptions, manage to-do lists via tools like Notion, generate reports and presentations, and organize messy files. All of that from your phone whilst you're not at your desk. The practical move right now is to identify your top three most time-consuming admin tasks, the things that eat hours but don't generate anything directly, and test whether Co-work can handle them. Start there. Don't try to automate everything at once. Pick the three simplest, most repetitive tasks you do every week. Test Dispatch on those.

If you're a business owner managing a team, here's where this matters. The question I'd be asking is, what are my team members spending time on that an AI agent could implement instead? If someone on your team is spending two hours a week building reports, that's two hours of human work spent on robot work. That's not a knock on your team. That's just an inefficiency that now has a solution. One user built an email priority run skill that categorizes emails by work pillar, drafts responses, and produces a dashboard with copy-and-pasteable replies. That's a repeatable workflow. Save it once, run it every day. The skill does the same work every morning without anyone having to remember it. Think about how many of those workflows exist inside your business. The weekly summary, the monthly report, the new client onboarding document, the competitor monitoring, the content calendar, the follow-ups. Every single one of those is a candidate for an AI workflow. Now with Dispatch, your team can trigger those workflows from their phones whilst they're doing other things.

If you're a content creator, this one is personal for me personally. Content creation is genuinely one of the best use cases for Co-work and Dispatch. Right now, you're working across multiple platform formats, right? YouTube scripts, blog posts, email newsletters, social media content, podcasts. You have research scattered across dozens of tabs and documents. You have notes that never become articles. You have ideas that just drop off in a notebook somewhere. If you have rough notes spread across text files, markdown documents, or Word drafts, Claude can synthesize all of them into a first-pass report, identifies overlapping ideas, organizes sections logically, and produces a readable draft you can refine further. That right there is a content engine. Your job becomes creating, refining, and putting your voice on the work. The raw assembly, the first draft, the structural organization. Claude actually does that. You're still the creative director. You're not being replaced. You're being freed from the part of content creation that doesn't require your voice, the research, the outlining, the formatting, the cross-posting, the repurposing. Claude handles all of that. What's left for you is the thinking and the delivery. That's a great trade.

If you're a managerial implement, here's the lens I'd use, right? The question isn't, "Can my company use this?" The question is, "What happens if your competitors start using this before you do?" Think about two companies competing in the same market. Company A is running traditional workflows. Their people spend five hours a day on admin, reporting, document creation, and organization. Company B starts using Co-work and Dispatch across their team. The admin hours drop to one. Their team is spending four more hours a day on work that actually moves the business forward. That difference compounds week over week. After six months, Company A hasn't changed, but Company B has effectively added the equivalent of two or three full-time employees' worth of productive output without hiring anyone. That's not a hypothetical. That's basic math on [snorts] what happens when you reduce the friction between ideas and implementation at scale.

Now, here's something I want to address directly, because I know it's sitting in the back of some of your minds. What about the people who work for you? What does this mean for them? This is the honest question that a lot of AI content creators dodge, and I'm not going to do that. The truth is complicated. For most knowledge workers in the near term, the biggest risk is not being replaced. The biggest risk is being managed by someone who knows how to use AI and they don't, right? Because a gap in output between a person using AI well and a person not using it at all is going to make it very hard for organizations to justify paying them the same. That's the uncomfortable reality. The answer is not to be afraid of the tools. The answer is to become someone who knows how to use them.

The people who thrive in this environment are not necessarily the most technically skilled. They're the ones who learn how to use AI directly, who understand what tasks to delegate, what instructions to give, how to quality check the output, when to trust a machine, and when to override it. Those are learnable skills. They're not complicated. They don't require an engineering degree. They require curiosity and the willingness to spend a few hours a week actually using these tools. And that's the whole point.

There's a version of 2026 where you are spending five hours a day on admin. Your competitors are spending one. And you wonder why you're falling behind even though you're working so hard. And there's a version of 2026 where you've learned how to use these tools. You've built the workflows. You've freed up your capacity. And you're operating at a level of output that would have required a team three times your size two years ago. That second version doesn't happen by accident. It happens because you decided to invest the time to understand this stuff before it becomes a crisis.

Now, let me bring this back to Dispatch specifically on what I think the next few weeks will look like. This is still an early version, right? And you can expect more to come within the next few days. It's rolling now to Mac subscribers with Pro coming in the next few days. So, if you're on a Pro plan, you'll have access to this within a few days of when I'm recording this.

When you get access, here's what I'd suggest. First thing, update your Claude desktop app. The Dispatch option shows up in the Co-work section after you update. You'll see it in the left panel. You pair your phone by scanning a QR code. Takes about two minutes. After updating the Claude Mac app, a new Dispatch option appears in Co-work that will prompt you to scan a QR code to pair the session with your iPhone or other mobile devices.

Number two, start with one small test. Not like your most important workflow, something low stakes. Ask it to organize a folder. Ask it to pull data from a spreadsheet and write a summary. Ask it to search your emails for messages you haven't replied to. Something simple where it gets slightly wrong. Doesn't cost you anything to try. Right.

Third, pay attention to what it can't do yet, because the limitations right now are actually a preview of what's coming. The things that are hard today will be easy in six months. The things that are slightly clunky right now will be seamless by the end of the year. Understanding the edges of the tool helps you understand where to watch for improvements.

There are some real limitations that I want to be upfront. Right now, your desktop must be active. Claude works on your desktop computer, right? So, if your computer is asleep or the Claude desktop app is closed, Claude can't work on tasks. So, if you want to message Claude from your phone at 2 a.m. and have it work on something important overnight, your computer has to be awake and the app has to be open. There's a real constraint. It means you have to think about your setup in advance.

There are no notifications when tasks complete. Right now, all messages live inside a single continuous thread. There's no way to start a new thread or manage multiple threads. That will change, but for now, it's one conversation. You have to check back to see if it's done. And as I mentioned, it's still early. Some tasks will work perfectly, some won't. That's a deal with a research preview.

But here's what I want you to hold on to. Every single capability that exists inside the final polished version of a product existed first in a research preview that was clunky and unreliable and got mixed reviews, right? The people who got familiar with it early, who learned its quirks, who built their workflows around its capabilities before anyone else. Those are the people who had a head start when the polished version arrived. We are at that moment with Dispatch.

The question isn't whether this technology is going to be reliable and powerful and everywhere in 12 months. It will be, right? The question is whether you're familiar with it when that moment arrives, or whether you're scrambling to catch up.

Let me tell you what I think the world looks like in 12 months if this trajectory continues. Right now, the default workflow for most knowledge workers is: sit down at computer, open apps, do work, close computer, and go home. In 12 months, I think the default for people who are paying attention starts to look like this. You set your AI's agenda for the day when you wake up. You review what it completed overnight. You approve, reject, or redirect. You spend the first 90 minutes of your day on the work that only you can do. The thinking, the relationships, the creative directions. And your AI handles everything else. That's not a fantasy. That is a direct extrapolation from what Dispatch does today. The infrastructure for that workflow is being built right now, piece by piece.

Microsoft has also seen what Anthropic is doing with Co-work. Announced a new Copilot Co-work product built on top of Anthropic technology. Microsoft actually fell in value like 14% since Anthropic debuted in Co-work mid-January. Let that land for a moment. This is one of the largest companies in the world responding to an AI agent product from a company that is a fraction of its size by building their own version of it because they understand what's happening. Right? The desktop, the file system, the work computer. Those are not small categories. Those are at the center of the knowledge economy. And whoever controls the AI layer on top of that infrastructure is going to have enormous influence over how work gets done. Anthropic planted a flag there in January. Dispatch them deeply in that position. And it's why I keep coming back to the same message week after week. People who are building fluency with these tools right now, not just awareness, actual fluency, are positioning themselves for a completely different level of productivity, output, and leverage than the people who are watching from the sidelines. That's not hype. That's what is actually happening.

Anthropic shipped a major AI product update in January. They iterated every single week. They added Windows support. They added scheduled tasks. They added a plugin marketplace. And together, and today, they added Dispatch, the ability to run your AI agent from your phone. That is 70 days of shipping, right? Think about what 70 days of this pace looks like in one year.

I'll leave you with this. Every major productivity shift in history: electricity in factories, computers in offices, the internet in businesses, was met first with confusion, then skepticism, then cautious adoption, then a mad scramble by the people who waited too long. The people who won in each of those transitions were not necessarily the smartest. They were the ones who got curious early, who learned the tools before the tools were polished, who built the habits and the workflows whilst everyone else was still debating whether it was real. This is that moment.

Dispatch is a feature, but what it represents is a direction. And the direction is a world where AI is not a tool you pick up when you need it, but a worker you manage around the clock. Where your phone is the interface to an AI that's always on, always capable, and always implemented. Where the limiting factor in your output is not how many hours you can work, but how good you are at giving instructions and reviewing results. That world is being built right now, not in a lab somewhere, in a product you can download today. The question is simple: Are you going to be ready for it?

If the answer is yes, and you want to actually learn how to build these AI workflows, how to use tools like Co-work and Dispatch properly, how to apply AI at leverage to your specific business or career, come join us in the AI Profit Boardroom. Link in the comments, description, or just go to AIProfitBoardroom.com. We are doing the work in there, right? Real workflows, real results, real applications for real businesses. That's where the work gets done.

All right, that's today's episode. And if you found this useful, share it with someone who needs to hear it. Because the biggest reason people fall behind on this stuff is not that they're not smart, is that no one explained to them how it works in plain English. Be the person who passes along. And I will see you on the next one.