Master it with Mary

Anthropic just dropped something that should have every business owner, tech leader, and AI-curious professional paying attention. What if your AI didn’t stop working just because you stepped away from your desk? That’s the idea behind Dispatch inside Claude Cowork. It’s a new research preview that lets you assign work to an AI agent running on your Mac, then check in and guide it from your phone. In other words… your AI can keep moving through tasks on your computer while you’re away. Anthropic says this mobile control of Cowork is rolling out first to Max users, then to Pro users right after, and Cowork itself runs locally in an isolated VM on macOS. Why does this matter? Because this is another big shift in how we use AI. We’re moving from: “Help me with this task right now” to “Take this task, keep working on it, and I’ll check back in later.” That’s a huge mindset change. This isn’t just AI as a chatbot. This is AI as a working partner. And for people who manage a lot of moving pieces, that opens up some very real possibilities: • Researching and organizing information • Handling multi-step workflows • Working through files and local tasks • Letting you monitor progress without being glued to your desk What makes this especially interesting is the local control piece. Anthropic says Cowork runs on your computer with access to local files and integrations, which is exactly why this will catch the attention of knowledge workers, developers, and businesses that care about privacy and keeping work close to home. My take? This is where AI starts feeling less like a tool you open… and more like a team member you delegate to. We’re still early, and it’s a research preview, so this is not the final form. But the direction is clear: AI is becoming more persistent, more agentic, and more woven into how real work gets done. Anthropic’s March 17, 2026 release note describes it as controlling Cowork from your phone with a persistent thread. The question is no longer just: “Can AI help me?” Now it’s: “What work am I comfortable handing off?” That’s the conversation smart businesses should be having right now.

Big ask… and an even bigger impact. 💙 I’m supporting a friend and the incredible work of HELP Now AZ at their Inaugural Pickleball Community Event — and this one matters. Every swing, every match, every ticket helps support families in crisis through HELP Now AZ. 🏓 Come play. 🎟️ Grab a ticket. 🙌 Or simply show up and support. 📅 April 25 | 10AM–2PM 🍳 Breakfast + 🌮 Lunch + 🎶 Live DJ + 🎁 Raffle If you’ve been looking for a way to give back AND have a great time doing it… this is it. 👉 Get your tickets now! Let’s show up, support, and make a real impact. 💥

Someone just sold their house using AI. No real estate agent. No marketing team. Just prompts. After living in his home for 15 years, a Florida homeowner decided to experiment: What if AI handled the entire process of selling the house? So he used AI for almost everything: • creating a timeline for preparing the house • suggesting improvements that would increase resale value • determining pricing strategy • writing the listing description • designing open house marketing materials • explaining how to list the property on MLS • recommending the best day to list • coordinating showings • even helping generate the sales contract The results? Within 72 hours of listing the home, he had five offers. Five days later, the home was under contract. He still hired a lawyer to review the paperwork (smart move), but he estimates using AI saved him about 3% of the sale price. Now before any real estate agents panic… AI isn’t replacing professionals. But it is becoming the assistant behind everyday decisions. People are already using AI to: • plan vacations • write resumes • build meal plans • create workout routines • write marketing content • research purchases • organize projects • manage email • plan events The real shift isn't AI doing everything. The shift is that people who learn how to use AI well suddenly have a powerful assistant. And that assistant works 24/7. The only part AI couldn’t help him with? Packing the moving boxes. Yet. Curious… Would you trust AI to help you sell your house?

Your personal data is probably floating around the internet right now. 😬 Not on one site… On hundreds. Data brokers. Sketchy directories. Sites you’ve never even heard of. That’s exactly what scammers and identity thieves rely on. Enter: Incogni 👇 This tool goes out and finds + removes your personal data from data broker websites — automatically. No manual requests. No chasing companies. No endless forms. Just cleaner digital footprints and way less exposure. Why this matters: The more your data is out there… the easier it is to target you with scams, phishing, and identity theft. This flips that. Less data online = less risk. Simple. If you’re serious about digital security, this is one of those tools that just makes sense. Worth checking out.

Big move in the AI ecosystem today. 👀 Anthropic just launched a limited preview of the Claude Marketplace — and the strategy behind it is smart. Instead of forcing companies to open new budgets or go through procurement again, businesses can use their existing Claude spend to access partner tools. That includes platforms like: • GitLab • Harvey • plus additional partners rolling out Why this matters: For most enterprises, procurement is the real bottleneck to AI adoption. Not the tech. So if teams can tap into tools using budget they’ve already approved… adoption speeds up fast. This is the bigger shift happening in AI right now: The winners won’t just be the companies with the best models. They’ll be the ones building ecosystems around those models... • Marketplaces • Partners • Integrations • Distribution And Anthropic just made a serious move in that direction. The AI platform race just got a little more interesting.

🍀 Happy St. Patrick’s Day! 🍀 Being Irish and Scottish, this holiday is basically in my DNA. ☘️🏴 Today is about a few important things: ☘️ Wearing green so you don’t get pinched ☘️ Believing a little extra in luck ☘️ And maybe enjoying a drink that’s suspiciously green But honestly, I kind of love the reminder behind it all… A little more joy. A little more laughter. And maybe taking a chance on something lucky. So here’s to good vibes, good people, and a little bit of Irish luck finding its way to you today. 🍀✨ Sláinte! 🍻

Ohhhh LinkedIn 👀👏 The hammer is quietly falling… LinkedIn just announced Automated Comment Detection — and it’s a big one. If a comment is submitted using: • A browser extension • A script • A third-party automation tool …and there’s no human actually clicking “comment”? 👉 It’s not allowed. What happens? • The comment can be removed from “Most Relevant.” • It may disappear outside the commenter’s network. • Continued use could lead to restrictions. Now let me be clear… I LOVE AI. I teach AI. I build AI Assistants. I help people use AI to grow faster and smarter. But I like real people more. There’s a difference between: ✔️ Using AI to help you think or refine your words ❌ Using automation to fake engagement AI should amplify you — not replace you. I am curious though… What about tools that help you draft a comment, but you still review it and click submit yourself? I know several colleagues who use those as support, not automation. Will LinkedIn draw a clear line between assistance and autopilot? That’s going to be interesting to watch. At the end of the day… Real connection > fake scale. Authentic conversation > automated noise. Where do YOU think we draw the line? 💛

Ohhhh LinkedIn 👀👏 The hammer is quietly falling… LinkedIn just announced Automated Comment Detection — and it’s a big one. If a comment is submitted using: • A browser extension • A script • A third-party automation tool …and there’s no human actually clicking “comment”? 👉 It’s not allowed. What happens? • The comment can be removed from “Most Relevant.” • It may disappear outside the commenter’s network. • Continued use could lead to restrictions. Now let me be clear… I LOVE AI. I teach AI. I build AI Assistants. I help people use AI to grow faster and smarter. But I like real people more. There’s a difference between: ✔️ Using AI to help you think or refine your words ❌ Using automation to fake engagement AI should amplify you — not replace you. I am curious though… What about tools that help you draft a comment, but you still review it and click submit yourself? I know several colleagues who use those as support, not automation. Will LinkedIn draw a clear line between assistance and autopilot? That’s going to be interesting to watch. At the end of the day… Real connection \ fake scale. Authentic conversation \ automated noise. Where do YOU think we draw the line? 💛

Doctors didn’t go to medical school to become data entry specialists. And yet… for years, that’s exactly what’s happened. Now AI-powered ambient documentation tools like Dragon Copilot by Microsoft are changing that. Here’s what’s happening: AI listens to the doctor-patient conversation. It automatically transcribes it. It generates structured clinical notes. It drafts documentation directly inside the health record. In real time. That means: • Less time charting at night • Less screen time during appointments • More eye contact • More presence • Less burnout This isn’t AI replacing doctors. It’s AI removing the administrative weight that’s been crushing healthcare for years. When AI handles repetitive documentation, physicians can focus on what actually matters: Listening. Diagnosing. Caring. Connecting. And honestly? That’s the kind of AI we should all be paying attention to. We don’t need more hype. We need more real-world solutions that give humans their time back. Would you feel comfortable knowing your visit was transcribed by AI if it meant your doctor was fully present with you? Curious to hear your thoughts. 👇

AI just got a little more responsible — and that’s a big deal. OpenAI is rolling out serious security upgrades in ChatGPT with Lockdown Mode and Elevated Risk labels, and I am here for it. 👇 🔐 What’s new? ✨ Lockdown Mode — an optional advanced setting built for high-security situations (think execs, security teams, sensitive data environments). This mode tightens how ChatGPT interacts with external apps and the web, disabling or restricting tools that bad actors might exploit to leak data. It’s like putting your AI in a digital safe with a keypad lock — less flexibility, more trust. ⚠️ Elevated Risk labels — clearly visible cues in the UI that tell you when a feature might introduce additional exposure. They’re designed to help teams make smarter decisions about when to enable powerful capabilities that touch external systems. 💡 Why it matters: As AI gets more connected — browsing the web, interacting with apps, automating workflows — the risk surface grows. Prompt injection attacks (where a crafty input tricks the AI into doing or revealing something harmful) are becoming more than just academic theory — they’re a real security challenge. These new tools show a shift in AI product design: security isn’t just buried in the engineering basement anymore — it’s front and center. We’re starting to get controls that let organizations balance usability with safety without throwing the baby out with the bathwater. If you’re building with AI, using it in your org, or just thinking about what “trusted AI” really means — this is the kind of feature update worth paying attention to. 🙌