Carl Rabeler

The Mindless VA Is Dead. Long Live the Specialist. As of today across our companies, we stopped drawing artificial lines between “VAs” and “specialists.” Doesn't matter whether you're in the U.S. or abroad, whether you're young or experienced, or whether English is your first language. What matters is your competency. Can you: ** Make and process video content that tells stories and drives action? ** Understand client needs and communicate them clearly? ** Speak empathetically with AIs to get quality outputs? Jack Wendt and I have been traveling through 8 countries over the last 3 weeks—spotty Wi-Fi, 9 time zones ahead of our usual rhythm—and yet we’re still delivering for clients. Because we understand the work, the context, and how to leverage AI to multiply our impact. Fiverr-style tasks can be done by the AI, supervised by A Players. Smaller teams of A Players are using AI to 100X their value. If you're an A Player, you love this shift that's happening. Why shouldn't a rockstar in Manila earn like a rockstar in Milwaukee?

38

7 Comments

When it comes to selecting the right DXP, things can go off the rails. But don’t lose your caboose. Dan Drapeau has a two-track evaluation model that can keep you rolling – full steam ahead. In today's Guest Critic article, Dan proves why he’s a masterful conductor on the digital experience train. As he explains, the DXP competitive landscape has converged across architectural models, and evaluation approaches haven’t evolved at the same pace. Dan’s observation is that single-track processes are still chugging along – focusing on feature breadth and demo cohesion and favoring pre-composed suites. In his experience, what we really need is a more balanced approach that separates platform evaluation from ecosystem architecture. How do you leave the station? By running two structured tracks, which he outlines in detail. This helps vet not only current capabilities but also long-term durability relative to core considerations like vendor lock-in, CDP and DAM strategies, and the ever-evolving target of AI. In his post, Dan details the operational requirements, where the complexities exist, and how to synthesize results. He even digs into the critical topics of validation and stress testing, and distills it all into a handy checklist to help guide your DXP evaluation. As he says, composable and pre-composed DXPs represent architectural strategies along a spectrum, and these pointers can help keep you on track. “A structured two-track evaluation approach ensures that whichever model is selected, it is chosen for structural alignment and long-term outcomes rather than surface cohesion.” If you’re currently evaluating DXPs – or considering it in the future – this is a fantastic read from a sharp practitioner. Check it out on CMS Critic: 🔶 https://lnkd.in/e-XB3Pig #dxp #digitalexperience #digitalexperienceplatform #softwareevaluation #dam #digitalassetmanagement #cdp #customerdata #customerdataplatform #cmp #contentmarketingplatform #ai #artificialintelligence

15

1 Comment

New quarter, new Lenny's List on Maven I'm seeing a shift from people wanting to learn how to use AI to write PRDs and brainstorm ideas, to answering more technical questions like: How do I build reliable agentic systems? What do I need to know about evals? How do I use AI even more in my product and design workflows? So I updated Lenny's List to reflect this evolution by adding a few more technical courses: - "Prototype to Production: The AI PM Playbook" with Aman Khan - "Agentic AI System Design for PMs" with Gabriela de Queiroz and Hamza Farooq - "Systematically Improving RAG Applications" with Jason Liu - "Level Up with Figma" with Joey Banks A few reasons why I continue to recommend Maven for learning: 1. Real practitioners - Every instructor has shipped many real products at scale. 2. Project-based - Each course makes you ship something. 3. Fresh - With the pace of AI innovation, live courses are the most up-to-date way to learn. Check out all of the courses at maven.com/lenny and use code LENNYSLIST to get a whopping 15% off any of the courses.

304

20 Comments

I post this with a few caveats. * I am an AI skeptic. I believe that generative AI (genAI) is a useful tool in specific contexts, and a dangerous crutch in others. * I believe that genAI is not intelligence except at the very weakest definitions for intelligence - it represents a marker that we are beginning to get a handle on cognitive computing, but we still have a long way to go before. * I believe that once the novelty of genAI wears off, we will realise that it is often a very expensive hammer to do many things that could be done with a cheaper hammer more effectively, and that there is a concerted effort to keep people from this realisation until a sufficient amount of money has been extracted from them. * If your job is in the field of cognitive intelligence, you should be paying very close attention to what is going on right now, but if it's not, then you can be overwhelmed with what are, at their core, academic arguments. * I always ask myself: "In five years time, will this be relevant?" Mostly the answer is no, but occasionally it will be yes; educate yourself to the point where you can (usually) tell which is which. * A good magician practices the art of misdirection - focusing the eyes of the audience on something that is flashy in order to disguise what they are really doing. There is a lot of misdirection in the genAI industry right now, which is why it's worth looking at it with a criticial eye. * Most genAI companies are failing. Most genAI initiatives are failing. The ones that succeed are in areas where you would expect them to succeed - in generating content that doesn't require a high degree of veracity or consistency. * Similarly, beware of survivor bias. Amazon and Google succeeded because there was need for their products and few competitor at the time, and even then it took a couple of decades for that to happen. There is surprisingly little need for GenAI outside of the niches that they are already doing well in. * FOMO plays primarily upon greed. It's at the heart of nearly every scam and ponzi scheme out there - if you don't do this, you'll not get rich, which is fallacious since there are any number of people who didn't do that who nonetheless became rich. There's also the implication (much less often stated) that if you do this, you will become rich. Most often, that is false as well, as anyone who plays the lottery can tell you. Someone will likely get rich, yes, but everyone else gets just a little bit poorer). So, staying informed with what's happening with GenAI is reasonable, but if you look at GenAI as being a creative but unreliable service, you can then better evaluate whether it is useful for you. https://lnkd.in/g2VQ9kV9

69

24 Comments

📣 How Axios is navigating a post-news era... America’s information ecosystem is broken, polluted and increasingly dangerous. Too many people don’t know what or whom to trust — or what’s even real. AI will make this worse before it gets better. 📌 Why it matters: Axios is hellbent on being part of the solution. We've succeeded, in our first decade, in making vital topics easier to consume and understand for smart professionals. But all of us, Axios included, need to help clean up how realities are formed — and informed. 🧠 The big picture: We've entered the post-news era. Your reality — how you see the world — is no longer defined by "the news." Instead, it's shaped by the videos you watch, the podcasts you listen to, the people you follow on social media and know in person, and the reporting you consume. What traditional news media companies can do is be useful, trusted, illuminating sources of vital information. That calling is more important than ever. Now, we want Axios to grow even more useful, more clinical and more trusted by more people. So in 2026, we're committed to four big goals: 1️⃣ Focus on the three once-in-a-generation changes unfolding in real-time: The dramatic effects of AI on business, society & work; the dramatic shift in politics and governance; and the equally dramatic shift in how our realities form. 2️⃣ Help readers navigate this historic change. 3️⃣ Fight fake fire with factual fire. 4️⃣ Expand Axios Local to more cities and towns. 🙏 Our ask of you: Please be quick to share feedback and ideas. We serve a premium audience and deeply appreciate your input. If you find us useful and trustworthy, please tell your friends and share our content. 💡We realize this is an epic moment, with politics, technology and our realities changing with rare velocity. We want to help you navigate this with clinical, useful and illuminating coverage. 📰 Read my full column on this new era: https://lnkd.in/eYmRt3XP

130

6 Comments

If you're at #AutoCon4, you'll have an opportunity to chat with the folks at Forward Networks, Inc.--and you should. If you don't know Forward Networks, they are the original digital twin. I don't mean "we stand up some images in ContainerLab and try to replicate your network" sort of digital twin. I mean they model your network mathematically and bring your actual network state into that model as often as you need to facilitate network operations. From an automation perspective, run your proposed change against the Forward model to make sure that you aren’t breaking prod, and do it from two perspectives. 1️⃣ Use Forward not only for "Will this work?" questions, but also... 2️⃣ "Did this work?" questions--both pre- and post-change. Really useful for your network automation platform. There is of course a lovely Forward Networks GUI, but for the network automation folks walking the halls of AutoCon, you'll be more interested in the API. If you swing by the booth, they can show off what they can do, and answer your technical questions.

65

3 Comments

On April 4, 1975, Bill Gates and Paul Allen founded Microsoft Corporation in Albuquerque, New Mexico. Initially named "Micro-Soft," the company was a portmanteau of "microcomputer" and "software," reflecting its focus on developing software for emerging microcomputer technology. The catalyst for Microsoft's founding was the Altair 8800, the first commercially successful personal computer. Gates and Allen saw an opportunity to create a BASIC interpreter for the Altair, which became their first major product. This early success was pivotal in establishing Microsoft as a key player in the burgeoning personal computer industry. Bill Gates's decision to drop out of Harvard University was a significant turning point in his career. He left to pursue Microsoft full-time, joining Paul Allen in Albuquerque. This commitment allowed them to focus on developing software that would eventually propel Microsoft to global prominence. The company's early success with Altair BASIC led to further opportunities, including the development of MS-DOS for IBM's personal computer in 1980, which became a cornerstone of Microsoft's dominance in the personal computer market. Over time, Microsoft expanded its product line to include Windows, Office, and other iconic software solutions, cementing its position as one of the world's leading technology companies. #FunFacts #history #interestingfacts #facts #trivia #BillGates #PaulAllen #microsoft #microcomputer #software #company #develop #harvard #fulltime

22

2 Comments

AI industry horrified to face largest copyright class action that could financially destroy it entirely! Perhaps they should’ve been ethical?👇🏽 The AI company Anthropic and industry groups are reportedly begging an appeals court to reverse last month’s decision made by a federal judge who ruled that potentially millions of writers can join a copyright infringement lawsuit brought against Anthropic, which is a serious blow to the entire industry. Ars Technica reports that AI industry representatives are pleading it would be the largest copyright class action in history that could "financially ruin" the entire AI industry. These AI industry groups, led by Anthropic, are urging the appeals court to block it completely because, as Anthropic humbly argues, it can't be allowed to happen because AI is simply too important to fail 🙄 So ruining the lives of authors and creators by stealing their work is not that important, apparently. And, yes, let’s call it for what it is: stealing! Regardless, according to the report, Anthropic argued that it now "faces hundreds of billions of dollars in potential damages liability at trial in four months" based on a class certification rushed at "warp speed" that involves "up to seven million potential claimants, whose works span a century of publishing history," each possibly triggering a $150,000 fine. But I have zero sympathy because the industry brought this situation on to itself, despite warnings from industry insiders like me and many others. Here’s an insider secret for you. Just over a decade ago, certainly from my own experience at Google Campus, there was a common etiquette (not a rule or law) that if you train your AI models with copyrighted content, it would only be used for internal experiments, or you pay copyright licenses to owners where appropriate. Hence, the idea that any company would just take other people’s work to train AI models and sell them to consumers was simply unthinkable. You just don’t do that 🤷 My company at the time built science-based language models and we worked closely with academic institutions and publishers to ensure full transparency and collaboration and worked well. So for those who claim that it is not possible are either lying or clueless! However, few VCs told me directly when I was telling them my plans about building an ethics-first AI company, Nebuli, in 2018, they see ethics as “bullshit” and “an obstacle”! Were they lying or just clueless? 🙈 So this attitude was already festering for years, mostly amongst the VC/investor community. I know, not all of them are the same, but if look at the current state of the industry, you can see it’s all about prioritising shareholders, not innovation, and definitely not ethics or people’s interests. Hence, I walked away from all of this crap, and I know others have done the same. Because we saw where this industry was heading. I warned about it since 2014, and yet here we are 💀 #ceo #ceor #ai #copyright #ethics

347

121 Comments

What happens when we build AI capabilities that become an extension of ourselves? The sci-fi version: You embed a Augentic AI and related tech into your brain. It's your personal assistant, therapist, and all-around agent—enhancing your human capability. In this model, everything it sees, hears, processes, or stores becomes part of your digital self. Now let’s bring that same power to something as accessible as your iPhone. It listens to every conversation. Labels the speakers. Pulls from your data—calendar, apps, emails, chats. Does real-time internet research mid-meeting. Add in smart glasses with facial recognition? Now it knows who you’re talking to, compares the discussion to psychological profiles, and feeds you real-time insights on how to land your point better—or just stores it for later. It’s not far-fetched. It’s technically possible now. But here’s the thing: we don’t have legal, regulatory, security, or privacy frameworks that truly recognize what this means—especially if all data is used solely for personal augmentation and becomes part of your digital identity. This kind of human augmentation will be adopted—regardless of whether frameworks are ready for it. Which is why security and privacy need to get there first. Not to slow us down, but to help us do it right. #AI #HumanAugmentation #DigitalSelf

17

1 Comment

Every product tells a story nobody wrote down. Most people see mistakes. I see archaeology. Last month I found a button labeled "Export (OLD)." It was the only export button. There was no new one. The person who named it "OLD" left two years ago. Nobody knew why it was called that. Nobody wanted to touch it. That button had become sacred through fear. I find these artifacts everywhere: A settings page with a design language from 2019. A three-step flow that used to be seven, someone optimized it, left no notes. A dropdown with options that reference a feature that no longer exists. Every strange decision made sense to someone. A deadline that couldn't move. A workaround for a bug that got fixed. A feature cut halfway through. The context left with the person who built it. Products don't get messy because teams are careless. They get messy because decisions accumulate faster than documentation. This is what I think about on quiet weekends. How every product is really just layers. Sediment from every team, every pivot, every compromise. My job isn't to judge the layers. It's to understand them well enough to know: What stays. What goes. What finally gets named. The best systems don't erase history. They organize it. Starting from scratch feels cleaner. Working with what exists is harder. And almost always smarter. Most products don't need a revolution. They need an editor. What's the strangest artifact you've found in a product you inherited?

110

63 Comments

Let’s be honest: MACH (Microservices, APIs, Cloud-native and Headless) isn’t shiny and new anymore. It’s out of the honeymoon phase. And that’s a good thing—because now we get to talk about what works. We learned from real conversations in Chicago at the MACH Alliance's Composable Conference this week from awesome martech practitioners from Altitude-sports.com, Diageo and James Hardie: composability is no longer a theory. It’s a discipline. And for marketing technologists, it’s a tough one. You’re not just buying flexibility—you’re inheriting complexity, integration debt, vendor orchestration and organizational growing pains. We asked the hard questions: Has MACH drifted into buzzword territory? Are companies really seeing the benefits? What does meaningful modularity look like in practice? We got the answers—and they aren’t always romantic. Like Pamela Anderson and marriage. It works; just not all the time. You’ll see the cracks (vendor roadblocks for developers), the real breakthroughs (modular stacks that reduce IT dependency) and the gray areas where AI, governance and agility still collide. I've come to realize this is the nature of the beast. It's tech. There really is no paradise. Martech is an airport hotel. Nothing sexy. No four-star breakfasts. Just some cereal boxes and a waffle machine if you're lucky. Gets the job done, if you have the right mindset. But this is also true: some big brands are buying into composability. They told me. MACH can enable better experiences, but only if you lead the orchestration. The architecture won’t do it for you. You will need to align teams, train differently, think modular but act deliberate, and, above all, resist the urge to overbuild. And we heard all of this straight from the practitioners: Yann Boisclair-Roy, Danson Huang and Sophia Zlatin, with an assist from KPMG's Nat Gross. https://lnkd.in/ezes4cWR

18

Building agents in Copilot Studio? Here are 6 frameworks that are helping me ship agents that actually work in production.I've watched dozens of agent projects kick off, but not land. Not because the technology was wrong. I was wrong. The first agents I helped design looked perfect in demos. Then they hit production- and users abandoned them in weeks. The problem wasn't Copilot Studio. It was us. We jumped straight to building without asking the right questions first. Here are 6 frameworks I use when designing agents in Copilot Studio. Each framework helps you: • Make better architectural decisions upfront • Avoid common mistakes that kill adoption • Build agents that scale 𝟭. 𝗢𝗿𝗰𝗵𝗲𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻 𝗠𝗼𝗱𝗲𝗹 Classic vs. Generative orchestration. ↳ Classic: Trigger phrases match to topics (predictable, rigid) ↳ Generative: Agent autonomously selects topics/tools based on descriptions and context (flexible, intelligent) Decision: How much autonomy does your use case require? 𝟮. 𝗕𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗕𝗹𝗼𝗰𝗸 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 Four primitives: Topics, Tools, Knowledge, Connectors. ↳ Topics = conversation flows ↳ Tools = actions the agent can take ↳ Knowledge = grounding data ↳ Connectors = integrations to external systems Decision: What does your agent need to know vs. do? 𝟯. 𝗗𝗲𝗽𝗹𝗼𝘆𝗺𝗲𝗻𝘁 𝗣𝗮𝘁𝘁𝗲𝗿𝗻𝘀 Three paths: Standalone, M365 Extension, or Embedded. ↳ Standalone: Full control, custom branding, any channel ↳ M365 Extension: Surface inside Microsoft 365 Copilot ↳ Embedded: Inside your apps, websites, or workflows Decision: Where do your users already work? 𝟰. 𝗔𝘂𝘁𝗵𝗼𝗿𝗶𝗻𝗴 𝗔𝗽𝗽𝗿𝗼𝗮𝗰𝗵 Manual design vs. AI-assisted authoring. ↳ Manual: Full control, topic-by-topic configuration ↳ AI-assisted: Describe what you want in natural language, Copilot builds it Decision: Speed vs. precision. Most teams need both. 𝟱. 𝗟𝗶𝗳𝗲𝗰𝘆𝗰𝗹𝗲 𝗙𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸 Plan → Implement → Manage → Improve → Extend. ↳ Plan: Define objectives, success criteria, risks ↳ Implement: Build topics, tools, knowledge integrations ↳ Manage: Governance, security, ALM ↳ Improve: Analytics, KPIs, prompt optimization ↳ Extend: SDK integrations, child agents, MCP Decision: Where is your team spending time? Where should they be? 𝟲. 𝗛𝘂𝗺𝗮𝗻-𝗔𝗴𝗲𝗻𝘁 𝗜𝗻𝘁𝗲𝗿𝗮𝗰𝘁𝗶𝗼𝗻 𝗠𝗼𝗱𝗲𝗹 Who decides what—and when does handoff happen? ↳ Full automation: Agent handles end-to-end ↳ Human-in-the-loop: Agent drafts, human approves ↳ Escalation paths: Agent knows when to hand off Decision: What's the consequence of a wrong answer? Most agent projects fail not because of bad technology. They fail because teams skip the framework decisions and jump straight to building. Start with the "why" and "who" before the "how." --- Which framework does your team need most right now? Follow for more on building agents in Microsoft Copilot Studio.

88

As we celebrate APIHM, let’s listen to the voices of the Gameheads Hawai‘i students.   A few years ago, Lili’uokalani Trust approached Gameheads with a vision to bring their program to Hawai‘i, to teach video game design, development, DevOps, and mixed media to low-income students and Native Hawaiian youth, to carry on the legacy of Queen Liliʻuokalani (Hawai‘i’s only reigning queen) and help Hawaiian children realize their greatest potential. Today, their vision has become a reality, and now Gameheads and Liliʻuokalani Trust, are working together with students on educational development, workforce training, STEAM learning, college prep, and social & emotional growth. This gives Native Hawaiian students skills to thrive and succeed in tech and video game industries . . . a career path where they could build a meaningful future without having to leave their native lands.   “Representation is powerful. It tells us that we belong. As a Native Hawaiian, I rarely saw people who looked like me in games or in the industry behind them. That’s why mentoring with Gameheads Hawai‘i means so much to me. These students aren’t just learning to make games . . . they’re learning that their voices, their stories, and their culture matter. When young people see someone who shares their background thriving in this space, it opens a door. It says, ‘You can be here too.’ And for our community, that visibility is everything. We have so many rich mo‘olelo (or stories) to share, and the next generation deserves the chance to tell them.” - Corinne Kaulia Kimball, Xbox Game Studios - Accessibility Quality Engineer    Thank you to Damon Packwood at Gameheads for creating such an amazing nonprofit program for underrepresented students of Hawai‘i and 16 other states. Hats off to Marcus Montgomery and Corinne Kuʻulei Kaulia Kimball of Xbox Game Studios for mentoring the next generation of Native Hawaiian game devs and helping them share their beautiful stories, culture, heritage, and aloha with the world. 🌺💚   Learn more about the Gameheads Hawai‘i program, and check out the students’ games: https://lnkd.in/gjTcTgPU

146

15 Comments

NYTimes Opinion - About the impact of AI on Critical Thinking. https://lnkd.in/g3ZAyPWT My concern is broader: the use of AI and basic expectations of rational thought. It seems that the magic of AI and non-sequitur magical thinking go hand in hand.

12

4 Comments

Value propositions are 10x harder in govtech because governments don't exist to make money. A healthy HALF or so of founders and early-career salespeople I know (myself included) have stalled out at some version of "time savings." That's shallow and intellectually dishonest. Do your customers tell you they're clocking out at 4:17 today because you saved them 43 minutes? What is actually going wrong in their organization because a process is time-consuming? Here are 10 better value propositions downstream of "time savings": 1. Improved resident satisfaction in service delivery (which, I'll argue, is the highest ROI lever for government agencies to build public trust. And top of mind for a lot of CIOs) 2. Address strategic priority 3. Reduce backlogs 4. Increase capacity to perform this work in-house rather than hire out consultants 5. Address compliance gaps that stem from spread-thin staff 6. Improve turnaround / response time (your customer may not know the term SLA, but what's their version of it? Same-day inspections? Firetruck arrives in five minutes?) 7. Improve accuracy and consistency of output 8. Augment staff shortages 9. Do something that can't be done right now (e.g. reporting or coordinating across multiple systems and departments) 10. Risk mitigation - preventative maintenance, emergency preparedness, etc. What did I miss?

45

26 Comments