Pascal Bornet

🎉 Our Book "AGENTIC ARTIFICIAL INTELLIGENCE" Is Finally Here! 🎉 Friends, today's the day! After months of hard work, I'm beyond excited to announce our new book, "Agentic Artificial Intelligence" — a practical, non-technical guide for business leaders, entrepreneurs, and curious minds. zurl.co/kbbH6 I'm incredibly proud that we brought together 27 brilliant minds from across business, academia, and programming to create this. It wasn't always easy, but we were driven by a shared goal: to bring real clarity to this field based on our hands-on experience implementing agentic AI in many companies. As Bill Gates put it, "Agents are bringing about the biggest revolution in computing since we went from typing commands to tapping on icons." But what does this really mean for you? Our book cuts through the hype to offer: 👉 Practical steps to find where AI agents can create actual value in your work 👉 Real stories of organizations cutting costs by over 25% while making customers 40% happier 👉 Honest advice on avoiding the pitfalls we've seen firsthand 👉 A clear view of how these technologies will reshape business and society Based on the content of the book, we have built the First Executive Masterclass on Agentic AI Strategy and Implementation. Join us here: zurl.co/aDVed This isn't about whether AI agents will transform your industry—it's about how you'll lead that change. I'd love to hear your thoughts once you've had a chance to read it. Your feedback will help all of us push this exciting field forward! #AgenticAI #AIBook #FutureOfWork #AITransformation #Leadership

The first time I watched Steve Jobs’ 1983 Aspen keynote, it really struck me. Not because of the technology predictions. But because of the mindset behind them. Jobs said computers would become extensions of human creativity and potential. In 1983, that sounded bold. Today, with AI everywhere, it feels almost obvious. But the line that stayed with me most was this: Everything around you was created by someone no smarter than you. The world isn’t fixed. It’s built. That idea becomes even more powerful in the AI era. Because today we’re not just using tools. We’re shaping entirely new possibilities for how humans and machines work together. So let me ask you: Are you simply watching the AI revolution happen… or are you helping build what comes next? #ArtificialIntelligence #Innovation #Leadership #Entrepreneurship #FutureOfWork #AI #SteveJobs

The AI Micromanager Trap I see this pattern a lot when companies adopt AI. Leaders think they’re driving the transformation. But often, they’re just slowing it down. 🚲 They believe they’re steering the team toward progress. ⚙️ In reality, they’re reviewing every prompt, editing every output, and creating bottlenecks. AI moves fast. Micromanagement doesn’t. From what I’ve observed, the most effective leaders don’t control every step. They set the direction — and let people and machines move. So here’s a question worth asking: Are you enabling AI… or unintentionally standing in its way? #ArtificialIntelligence #AITransformation #Leadership #FutureOfWork #Automation #Innovation

This is what IT calls seamless integration. And what senior management proudly calls digital transformation. Curious: 𝗗𝗼𝗲𝘀 𝘆𝗼𝘂𝗿 𝘁𝗲𝗰𝗵 𝘀𝘁𝗮𝗰𝗸 𝗹𝗼𝗼𝗸 𝗮𝗻𝘆 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁? #AI #DigitalTransformation #Automation #TechHumor #FutureOfWork Image credits: Ralph

This might be the most accurate illustration of 𝗔𝗜 𝗵𝗮𝗹𝗹𝘂𝗰𝗶𝗻𝗮𝘁𝗶𝗼𝗻 I’ve seen in a while. You ask a simple question. The model answers 𝗰𝗼𝗻𝗳𝗶𝗱𝗲𝗻𝘁𝗹𝘆. And then… reality strongly disagrees. Working with AI every day, I see this pattern often. Large language models don’t actually know things. They predict 𝘄𝗵𝗮𝘁 𝘀𝗼𝘂𝗻𝗱𝘀 𝗿𝗶𝗴𝗵𝘁. Most of the time, that works surprisingly well. But sometimes the answer is simply a very 𝗰𝗼𝗻𝗳𝗶𝗱𝗲𝗻𝘁 𝗴𝘂𝗲𝘀𝘀. Which is why asking AI whether a mushroom is edible might not be the safest life decision. 🍄 Curious: 𝗪𝗵𝗮𝘁’𝘀 𝘁𝗵𝗲 𝗺𝗼𝘀𝘁 𝗰𝗼𝗻𝗳𝗶𝗱𝗲𝗻𝘁𝗹𝘆 𝘄𝗿𝗼𝗻𝗴 𝗔𝗜 𝗮𝗻𝘀𝘄𝗲𝗿 𝘆𝗼𝘂’𝘃𝗲 𝗲𝘃𝗲𝗿 𝘀𝗲𝗲𝗻? #AI #AIHallucinations #ArtificialIntelligence #GenerativeAI #TechHumor

𝗠𝗮𝗿𝘀 𝗶𝘀𝗻’𝘁 𝗷𝘂𝘀𝘁 𝗮𝗯𝗼𝘂𝘁 𝗯𝗶𝗴𝗴𝗲𝗿 𝗿𝗼𝗰𝗸𝗲𝘁𝘀. 𝗜𝘁’𝘀 𝗮𝗯𝗼𝘂𝘁 𝘁𝗶𝗺𝗶𝗻𝗴. One thing that always fascinates me about space missions is this: SpaceX’s Starship does not fly to Mars in a straight line. It follows a Hohmann transfer orbit, an elliptical path around the Sun that uses gravity and momentum to travel the most efficient route possible. Which leads to an interesting constraint. You cannot launch to Mars whenever you want. The optimal Mars transfer window only opens about every 26 months, when Earth and Mars align in the right position. The next one is expected in late 2026. The journey itself is also quite remarkable: → Starship launches and enters low Earth orbit → Tanker ships perform orbital refueling → The spacecraft fires for Trans-Mars Injection (TMI) → It then coasts for 6–9 months through space And finally: Starship enters the Martian atmosphere at extreme speed, its heat shield absorbing the energy, before using aerodynamic drag and engine thrust to land vertically. What I find impressive here is not just the rocket. It is the precision of the system. Orbital mechanics, timing, refueling in space, atmospheric entry… all working together. Because reaching Mars is not just about power. It is about understanding the physics of the system you are operating in. Curious: 𝗗𝗼 𝘆𝗼𝘂 𝘁𝗵𝗶𝗻𝗸 𝗵𝘂𝗺𝗮𝗻𝘀 𝘄𝗶𝗹𝗹 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝘀𝗲𝘁𝘁𝗹𝗲 𝗼𝗻 𝗠𝗮𝗿𝘀 𝗶𝗻 𝗼𝘂𝗿 𝗹𝗶𝗳𝗲𝘁𝗶𝗺𝗲? #SpaceX #Mars #Innovation #SpaceExploration #Technology

𝗕𝗶𝗹𝗹𝗶𝗼𝗻𝘀 𝗮𝗿𝗲 𝗯𝗲𝗶𝗻𝗴 𝗶𝗻𝘃𝗲𝘀𝘁𝗲𝗱 𝗶𝗻 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗔𝗜 to summarize documents and produce concise reports. Meanwhile, somewhere in an office: → Ctrl + A → reduce the font size → delete the margins Suddenly the 30-page report becomes 22 pages. 𝗜𝗻𝗻𝗼𝘃𝗮𝘁𝗶𝗼𝗻. 😄 Curious: 𝗪𝗵𝗮𝘁’𝘀 𝘁𝗵𝗲 𝗺𝗼𝘀𝘁 “𝗶𝗻𝗻𝗼𝘃𝗮𝘁𝗶𝘃𝗲” 𝗼𝗳𝗳𝗶𝗰𝗲 𝘁𝗿𝗶𝗰𝗸 𝘆𝗼𝘂’𝘃𝗲 𝘀𝗲𝗲𝗻? #AI #GenerativeAI #FutureOfWork #TechHumo Image credits: Ralph Aboujaoude

𝗜𝗳 𝘆𝗼𝘂’𝗿𝗲 𝘀𝘁𝗶𝗹𝗹 𝗳𝗲𝗲𝗱𝗶𝗻𝗴 𝗿𝗮𝘄 𝗝𝗦𝗢𝗡 𝗶𝗻𝘁𝗼 𝗟𝗟𝗠𝘀, 𝘆𝗼𝘂 𝗺𝗶𝗴𝗵𝘁 𝗯𝗲 𝘄𝗮𝘀𝘁𝗶𝗻𝗴 𝘁𝗼𝗸𝗲𝗻𝘀, 𝗹𝗮𝘁𝗲𝗻𝗰𝘆, 𝗮𝗻𝗱 𝗺𝗼𝗻𝗲𝘆. This is something I see surprisingly often in AI systems. JSON works well for software systems. But for LLMs, it carries a lot of unnecessary structure. Recently I came across 𝗧𝗢𝗢𝗡 (Token-Oriented Object Notation), a format designed specifically for LLM workflows. And the difference is not subtle. Same dataset: → 𝟰𝟭𝟮 characters in JSON → 𝟭𝟱𝟰 characters in TOON Less structure noise. More useful information for the model. 𝗛𝗲𝗿𝗲’𝘀 𝘄𝗵𝗮𝘁 𝗧𝗢𝗢𝗡 𝗯𝗿𝗶𝗻𝗴𝘀 𝘁𝗼 𝘁𝗵𝗲 𝘁𝗮𝗯𝗹𝗲: ⬇️ → Token efficiency: often 30–60% fewer tokens for uniform datasets. → LLM-friendly structure: predictable fields improve model parsing. → Minimal syntax: fewer brackets, quotes, and repeated keys. → Flexible workflow: convert between JSON and TOON when needed. 𝗪𝗵𝗲𝗻 𝘀𝗵𝗼𝘂𝗹𝗱 𝘆𝗼𝘂 𝘂𝘀𝗲 𝗶𝘁? ⬇️ → Large structured datasets such as logs, catalogs, telemetry, user lists. → Systems where token cost and latency matter. For deeply nested or irregular data, JSON still makes sense. But if you are building AI agents, copilots, or automation systems, the format of your data suddenly matters much more than it used to. For years we optimized models. Now we are starting to optimize how we talk to them. 𝗖𝘂𝗿𝗶𝗼𝘂𝘀: 𝗗𝗼 𝘆𝗼𝘂 𝘁𝗵𝗶𝗻𝗸 𝗟𝗟𝗠-𝗻𝗮𝘁𝗶𝘃𝗲 𝗱𝗮𝘁𝗮 𝗳𝗼𝗿𝗺𝗮𝘁𝘀 𝗹𝗶𝗸𝗲 𝗧𝗢𝗢𝗡 𝗰𝗼𝘂𝗹𝗱 𝗯𝗲𝗰𝗼𝗺𝗲 𝘁𝗵𝗲 𝗻𝗲𝘅𝘁 𝘀𝘁𝗮𝗻𝗱𝗮𝗿𝗱? #AI #LLM #AgenticAI #DataEngineering #ArtificialIntelligence

𝗧𝗵𝗶𝘀 𝗿𝗼𝗯𝗼𝘁 𝗱𝗼𝗲𝘀 𝘀𝗼𝗺𝗲𝘁𝗵𝗶𝗻𝗴 𝗺𝗼𝘀𝘁 𝗱𝗿𝗼𝗻𝗲𝘀 𝘀𝘁𝗶𝗹𝗹 𝘀𝘁𝗿𝘂𝗴𝗴𝗹𝗲 𝘄𝗶𝘁𝗵. It can 𝘄𝗮𝗹𝗸, 𝗷𝘂𝗺𝗽, 𝗮𝗻𝗱 𝗳𝗹𝘆. Researchers at EPFL’s Laboratory of Intelligent Systems built 𝗥𝗔𝗩𝗘𝗡, a robot inspired by how birds move between land and air. Most drones need open space and stable launch points. 𝗥𝗔𝗩𝗘𝗡 works differently. → It can 𝗷𝘂𝗺𝗽 𝘁𝗼 𝘀𝘁𝗮𝗿𝘁 𝗶𝘁𝘀 𝗳𝗹𝗶𝗴𝗵𝘁 → It can 𝘄𝗮𝗹𝗸 𝗼𝗻 𝗿𝗼𝘂𝗴𝗵 𝘁𝗲𝗿𝗿𝗮𝗶𝗻 → It can 𝗹𝗮𝗻𝗱 𝘄𝗵𝗲𝗿𝗲 𝗺𝗼𝘀𝘁 𝗱𝗿𝗼𝗻𝗲𝘀 𝗰𝗮𝗻’𝘁 What I find interesting here is the design philosophy. Instead of forcing robots to adapt to machines, engineers are studying how 𝗻𝗮𝘁𝘂𝗿𝗲 𝗮𝗹𝗿𝗲𝗮𝗱𝘆 𝘀𝗼𝗹𝘃𝗲𝗱 𝘁𝗵𝗲 𝗽𝗿𝗼𝗯𝗹𝗲𝗺. That approach could unlock new possibilities: → 𝘀𝗲𝗮𝗿𝗰𝗵 𝗮𝗻𝗱 𝗿𝗲𝘀𝗰𝘂𝗲 in difficult terrain → 𝗶𝗻𝘀𝗽𝗲𝗰𝘁𝗶𝗼𝗻 in tight or cluttered environments → 𝗱𝗲𝗹𝗶𝘃𝗲𝗿𝘆 where drones cannot easily land Small shift in design. Potentially 𝗯𝗶𝗴 𝗶𝗺𝗽𝗮𝗰𝘁 in real-world robotics. 𝗖𝘂𝗿𝗶𝗼𝘂𝘀: 𝗪𝗵𝗲𝗿𝗲 𝗱𝗼 𝘆𝗼𝘂 𝘁𝗵𝗶𝗻𝗸 𝗵𝘆𝗯𝗿𝗶𝗱 𝗿𝗼𝗯𝗼𝘁𝘀 𝗹𝗶𝗸𝗲 𝘁𝗵𝗶𝘀 𝗰𝗼𝘂𝗹𝗱 𝗯𝗲 𝗺𝗼𝘀𝘁 𝘂𝘀𝗲𝗳𝘂𝗹? #Robotics #AI #Innovation #FutureOfRobotics #Technology

𝗜𝗺𝗽𝗿𝗲𝘀𝘀𝗶𝘃𝗲 𝗰𝗼𝗺𝗽𝗮𝗿𝗶𝘀𝗼𝗻. ⚡ Sam Altman recently responded to criticism about 𝗔𝗜 𝗲𝗻𝗲𝗿𝗴𝘆 𝘂𝘀𝗲 with an argument that made many people pause. People often talk about how much electricity it takes to 𝗰𝗿𝗲𝗮𝘁𝗲 𝗮𝗻𝗱 𝘁𝗿𝗮𝗶𝗻 𝗮𝗻 𝗔𝗜 𝗺𝗼𝗱𝗲𝗹. But we rarely compare it with what it takes to “train” a human. Think about it: → roughly 𝟮𝟬 𝘆𝗲𝗮𝗿𝘀 𝗼𝗳 𝗹𝗶𝗳𝗲 → food, education systems, and infrastructure → centuries of accumulated human knowledge His point is simple. The fair comparison is not 𝗔𝗜 𝘁𝗿𝗮𝗶𝗻𝗶𝗻𝗴 vs a human answering one question. It is the energy required for a 𝘁𝗿𝗮𝗶𝗻𝗲𝗱 𝗔𝗜 𝗺𝗼𝗱𝗲𝗹 to answer a question compared with the energy used by a 𝘁𝗿𝗮𝗶𝗻𝗲𝗱 𝗵𝘂𝗺𝗮𝗻 𝗯𝗿𝗮𝗶𝗻. Once the model is trained, he believes AI may already be approaching similar or even better 𝗲𝗻𝗲𝗿𝗴𝘆 𝗲𝗳𝗳𝗶𝗰𝗶𝗲𝗻𝗰𝘆 𝗽𝗲𝗿 𝗾𝘂𝗲𝗿𝘆. Working in AI, I find this framing interesting. Maybe the real debate is not simply how much energy AI uses. Maybe it is 𝗵𝗼𝘄 𝘄𝗲 𝗰𝗼𝗺𝗽𝗮𝗿𝗲 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 𝗮𝗻𝗱 𝗲𝗳𝗳𝗶𝗰𝗶𝗲𝗻𝗰𝘆. 𝗗𝗼 𝘆𝗼𝘂 𝘁𝗵𝗶𝗻𝗸 𝗔𝗜’𝘀 𝗲𝗻𝗲𝗿𝗴𝘆 𝘂𝘀𝗲 𝗶𝘀 𝗼𝗳𝘁𝗲𝗻 𝗺𝗶𝘀𝘂𝗻𝗱𝗲𝗿𝘀𝘁𝗼𝗼𝗱? #AI #ArtificialIntelligence #Energy #FutureOfAI #Technology

𝗔 𝗰𝗮𝗿 𝘁𝗵𝗮𝘁 𝗺𝗼𝘃𝗲𝘀… 𝗹𝗶𝗸𝗲 𝗮 𝗰𝗿𝗮𝗯. 🦀🚗 I recently came across Hyundai’s 𝗲-𝗖𝗼𝗿𝗻𝗲𝗿 𝗦𝘆𝘀𝘁𝗲𝗺, and it immediately made me think: why didn’t we build cars like this earlier? Each wheel can rotate independently. Which means the car can literally 𝗺𝗼𝘃𝗲 𝘀𝗶𝗱𝗲𝘄𝗮𝘆𝘀. No awkward turning. No multi-point parking. Just slide into the spot. The technology debuted on the 𝗜𝗢𝗡𝗜𝗤 𝟱-based 𝗠𝗼𝗯𝗶𝗼𝗻 concept and could change how vehicles move in crowded cities. Imagine: → 𝗽𝗮𝗿𝗮𝗹𝗹𝗲𝗹 𝗽𝗮𝗿𝗸𝗶𝗻𝗴 𝘄𝗶𝘁𝗵𝗼𝘂𝘁 𝘀𝘁𝗿𝗲𝘀𝘀 → gliding out of tight spaces → maneuvering where traditional cars struggle Sometimes innovation is not about making cars faster. It is about making everyday driving 𝗺𝘂𝗰𝗵 𝘀𝗺𝗮𝗿𝘁𝗲𝗿. 𝗪𝗼𝘂𝗹𝗱 𝘆𝗼𝘂 𝘁𝗿𝘂𝘀𝘁 𝗮 𝗰𝗮𝗿 𝘁𝗵𝗮𝘁 𝗰𝗮𝗻 𝗺𝗼𝘃𝗲 𝘀𝗶𝗱𝗲𝘄𝗮𝘆𝘀 𝗹𝗶𝗸𝗲 𝘁𝗵𝗶𝘀? #Innovation #Mobility #EV #FutureOfTransport #Technology

𝗙𝗶𝗿𝗲𝗳𝗶𝗴𝗵𝘁𝗶𝗻𝗴 𝗶𝘀 𝗴𝗼𝗶𝗻𝗴 𝗮𝗶𝗿𝗯𝗼𝗿𝗻𝗲. 🚁🔥 I’ve been following how robotics is entering emergency response, and drones are starting to change the way wildfires are fought. Instead of sending crews directly into dangerous areas, drones can now: → reach hotspots much faster → drop water or fire retardant precisely where needed → use AI-powered sensors to map fires in real time The impact is significant. Less risk for firefighters. Faster situational awareness. More precise intervention. It is a good reminder that AI and robotics are not only about productivity or automation. Sometimes the most meaningful applications are about protecting human lives. Curious: 𝗪𝗵𝗮𝘁 𝗼𝘁𝗵𝗲𝗿 𝘁𝗲𝗰𝗵𝗻𝗼𝗹𝗼𝗴𝗶𝗲𝘀 𝗱𝗼 𝘆𝗼𝘂 𝘁𝗵𝗶𝗻𝗸 𝗰𝗼𝘂𝗹𝗱 𝘁𝗿𝗮𝗻𝘀𝗳𝗼𝗿𝗺 𝗲𝗺𝗲𝗿𝗴𝗲𝗻𝗰𝘆 𝗿𝗲𝘀𝗽𝗼𝗻𝘀𝗲 𝗶𝗻 𝘁𝗵𝗲 𝗻𝗲𝘅𝘁 𝗱𝗲𝗰𝗮𝗱𝗲? #AI #Robotics #Drones #Innovation #FutureOfWork

𝗜𝗺𝗽𝗿𝗲𝘀𝘀𝗶𝘃𝗲. 𝗪𝗶𝗿𝗲𝘀 𝘁𝗵𝗮𝘁 𝗱𝗼𝗻’𝘁 𝗯𝗿𝗲𝗮𝗸. 👏 Researchers in China have developed a Metal-Polymer Conductor (MPC) by combining elastic polymers with liquid metals like gallium and indium. The result is quite remarkable. A circuit that can: → stretch → twist → fold → bend repeatedly …and still conduct electricity. Traditional wires fail under stress. This material behaves more like living tissue. Which opens interesting possibilities. → wearable electronics → soft robotics → medical sensors → even artificial skin Working in automation and AI, I find breakthroughs like this fascinating. Because the future of intelligent machines will not just depend on better software. It will also depend on new kinds of hardware that can move, flex, and interact safely with the human world. 𝗪𝗵𝗲𝗿𝗲 𝗱𝗼 𝘆𝗼𝘂 𝘁𝗵𝗶𝗻𝗸 𝗳𝗹𝗲𝘅𝗶𝗯𝗹𝗲 𝗲𝗹𝗲𝗰𝘁𝗿𝗼𝗻𝗶𝗰𝘀 𝘄𝗶𝗹𝗹 𝗵𝗮𝘃𝗲 𝘁𝗵𝗲 𝗯𝗶𝗴𝗴𝗲𝘀𝘁 𝗶𝗺𝗽𝗮𝗰𝘁? #AI #Robotics #Innovation #FutureTech #Technology

𝗖𝗵𝗶𝗻𝗮 𝗱𝗼𝗲𝘀𝗻’𝘁 𝘁𝗿𝘂𝘀𝘁 𝗳𝗼𝗿𝗲𝗶𝗴𝗻 𝘁𝗲𝗰𝗵 𝗳𝗼𝗿 𝗹𝗼𝗻𝗴. 𝗢𝗽𝗲𝗻𝗖𝗹𝗮𝘄 may be the latest example. The open source AI agent from the West began spreading quickly across Chinese developer communities this week. Soon after, Beijing raised security concerns. And within days, companies such as 𝗧𝗲𝗻𝗰𝗲𝗻𝘁, 𝗭𝗵𝗶𝗽𝘂, and 𝗞𝗶𝗺𝗶 started introducing their own local versions. This pattern is not new. Foreign technologies often gain traction in China first. But as they become strategically important, local alternatives tend to emerge rapidly. Over time, entire domestic ecosystems form around them. What caught my attention here, however, was the speed of adoption. In Shenzhen, more than 20 engineers set up a free street stall to help people install OpenClaw locally on their laptops. Thousands showed up. Not only developers, but students, teachers, farmers, parents, and even grandparents. Volunteers helped visitors download the software, configure it, and run the AI agent directly on their computers. The gathering formed just outside 𝗧𝗲𝗻𝗰𝗲𝗻𝘁’𝘀 headquarters. Scenes like this reveal something important. AI agents are beginning to spread through developer communities in a way that resembles the early growth of major software platforms. And whichever frameworks developers adopt first often shape the ecosystems that follow. This points to a broader shift in how software itself is evolving. For years, the dominant model was simple. 𝗔𝗜 𝘂𝘀𝗲𝘀 𝘁𝗼𝗼𝗹𝘀. But what we are beginning to see now is something different. 𝗧𝗼𝗼𝗹𝘀 𝘁𝗵𝗲𝗺𝘀𝗲𝗹𝘃𝗲𝘀 𝗮𝗿𝗲 𝗯𝗲𝗰𝗼𝗺𝗶𝗻𝗴 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝘁. The software industry may be moving from: 𝗔𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀 → 𝗪𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀 → 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝘀𝘆𝘀𝘁𝗲𝗺𝘀. Systems that do not simply execute commands, but understand tasks and coordinate work across tools. If Chinese developers begin building on domestic OpenClaw alternatives rather than the original framework, China’s agent ecosystem could evolve separately from Western AI stacks. And that could shape how the global landscape for agentic software develops in the coming years. Which raises an interesting question. 𝗜𝗳 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁𝘀 𝗯𝗲𝗰𝗼𝗺𝗲 𝘁𝗵𝗲 𝗻𝗲𝘄 𝗶𝗻𝘁𝗲𝗿𝗳𝗮𝗰𝗲 𝗳𝗼𝗿 𝘀𝗼𝗳𝘁𝘄𝗮𝗿𝗲, 𝘄𝗶𝗹𝗹 𝘄𝗲 𝘀𝗲𝗲 𝘁𝘄𝗼 𝗽𝗮𝗿𝗮𝗹𝗹𝗲𝗹 𝗔𝗜 𝗲𝗰𝗼𝘀𝘆𝘀𝘁𝗲𝗺𝘀 𝗲𝗺𝗲𝗿𝗴𝗲? One centered in the West. One built largely inside China. #AI #AgenticAI #ArtificialIntelligence #FutureOfWork #Automation

𝗔𝗽𝗽𝗹𝗲 𝗷𝘂𝘀𝘁 𝗺𝗮𝗱𝗲 𝗺𝗮𝘁𝗵 𝗳𝗲𝗲𝗹 𝗮 𝗹𝗶𝘁𝘁𝗹𝗲 𝗺𝗼𝗿𝗲 𝗵𝘂𝗺𝗮𝗻. With iOS 18, Apple introduced an AI-powered Calculator that can understand handwritten equations. You simply write the math problem. The system reads your handwriting, solves it instantly, and even keeps the result in your handwriting style. No typing. No switching formats. Just write → solve → continue. I find this kind of AI interesting. Not dramatic automation. Just technology quietly adapting to how humans already think and work. Sometimes the most meaningful AI innovations are the ones that **remove friction without changing behavior. 𝗪𝗵𝗶𝗰𝗵 𝘁𝗼𝗼𝗹 𝗱𝗼 𝘆𝗼𝘂 𝘄𝗶𝘀𝗵 𝗰𝗼𝘂𝗹𝗱 𝘂𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱 𝘆𝗼𝘂𝗿 𝗵𝗮𝗻𝗱𝘄𝗿𝗶𝘁𝗶𝗻𝗴 𝗹𝗶𝗸𝗲 𝘁𝗵𝗶𝘀? #AI #Apple #iOS18 #Innovation #HumanCenteredAI

𝗠𝗶𝗰𝗿𝗼𝘀𝗼𝗳𝘁’𝘀 𝗔𝗜 𝗖𝗘𝗢 𝗷𝘂𝘀𝘁 𝗱𝗿𝗲𝘄 𝗮 𝗰𝗹𝗲𝗮𝗿 𝗿𝗲𝗱 𝗹𝗶𝗻𝗲 𝗳𝗼𝗿 𝗔𝗜. He recently said something that made me pause. 𝗔𝗜 𝘀𝗵𝗼𝘂𝗹𝗱 𝗯𝗲 𝘀𝗵𝘂𝘁 𝗱𝗼𝘄𝗻 𝗶𝗺𝗺𝗲𝗱𝗶𝗮𝘁𝗲𝗹𝘆 if it ever combines four capabilities at the same time: → 𝗥𝗲𝗰𝘂𝗿𝘀𝗶𝘃𝗲 𝘀𝗲𝗹𝗳-𝗶𝗺𝗽𝗿𝗼𝘃𝗲𝗺𝗲𝗻𝘁 → 𝗦𝗲𝘁𝘁𝗶𝗻𝗴 𝗶𝘁𝘀 𝗼𝘄𝗻 𝗴𝗼𝗮𝗹𝘀 → 𝗔𝗰𝘁𝗶𝗻𝗴 𝗮𝘂𝘁𝗼𝗻𝗼𝗺𝗼𝘂𝘀𝗹𝘆 𝗶𝗻 𝘁𝗵𝗲 𝘄𝗼𝗿𝗹𝗱 → 𝗔𝗰𝗾𝘂𝗶𝗿𝗶𝗻𝗴 𝗶𝘁𝘀 𝗼𝘄𝗻 𝗿𝗲𝘀𝗼𝘂𝗿𝗰𝗲𝘀 Put those four together, he argues, and you no longer have just a powerful system. 𝗬𝗼𝘂 𝗵𝗮𝘃𝗲 𝘀𝗼𝗺𝗲𝘁𝗵𝗶𝗻𝗴 𝘁𝗵𝗮𝘁 𝗰𝗼𝘂𝗹𝗱 𝗿𝗲𝗾𝘂𝗶𝗿𝗲 𝗺𝗶𝗹𝗶𝘁𝗮𝗿𝘆-𝗹𝗲𝘃𝗲𝗹 𝗶𝗻𝘁𝗲𝗿𝘃𝗲𝗻𝘁𝗶𝗼𝗻. His conclusion is straightforward: 𝗔𝗜 𝗰𝗮𝗽𝗮𝗯𝗶𝗹𝗶𝘁𝗶𝗲𝘀 𝗹𝗶𝗸𝗲 𝘁𝗵𝗲𝘀𝗲 𝘀𝗵𝗼𝘂𝗹𝗱 𝗯𝗲 𝗮𝘂𝗱𝗶𝘁𝗲𝗱 𝗮𝗻𝗱 𝗿𝗲𝗴𝘂𝗹𝗮𝘁𝗲𝗱 — similar to nuclear infrastructure. I find this framing interesting. Not science-fiction scenarios. But clear technical thresholds that signal when intervention becomes necessary. 𝗤𝘂𝗲𝘀𝘁𝗶𝗼𝗻: 𝗗𝗼 𝘆𝗼𝘂 𝘁𝗵𝗶𝗻𝗸 𝗱𝗲𝗳𝗶𝗻𝗶𝗻𝗴 𝗰𝗹𝗲𝗮𝗿 “𝗿𝗲𝗱 𝗹𝗶𝗻𝗲𝘀” 𝗳𝗼𝗿 𝗔𝗜 𝗶𝘀 𝘁𝗵𝗲 𝗿𝗶𝗴𝗵𝘁 𝗮𝗽𝗽𝗿𝗼𝗮𝗰𝗵 𝘁𝗼 𝗴𝗼𝘃𝗲𝗿𝗻𝗶𝗻𝗴 𝗶𝘁? #AI #ArtificialIntelligence #AIGovernance #FutureOfAI #Technology

𝗔𝗻𝘁𝗵𝗿𝗼𝗽𝗶𝗰 𝗷𝘂𝘀𝘁 𝘀𝗼𝗹𝘃𝗲𝗱 𝗼𝗻𝗲 𝗼𝗳 𝘁𝗵𝗲 𝗺𝗼𝘀𝘁 𝗮𝗻𝗻𝗼𝘆𝗶𝗻𝗴 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀 𝗶𝗻 𝗠𝗶𝗰𝗿𝗼𝘀𝗼𝗳𝘁 𝗢𝗳𝗳𝗶𝗰𝗲. For decades, Excel and PowerPoint have technically been part of the same suite, yet in practice they rarely worked together smoothly. You would analyze the numbers in 𝗘𝘅𝗰𝗲𝗹, and then manually rebuild the story in 𝗣𝗼𝘄𝗲𝗿𝗣𝗼𝗶𝗻𝘁. The data lived in one place. The narrative lived somewhere else. 𝗖𝗹𝗮𝘂𝗱𝗲 now connects the two. Because it understands the spreadsheet while helping create the presentation, the AI can move directly from analysis to storytelling without losing context. In practical terms, the workflow can now look something like this: → Analyze portfolio data in Excel → Identify key insights and structure the analysis → Turn those insights into slides in PowerPoint → Keep the presentation updated as the spreadsheet evolves But what caught my attention is not only the feature itself. It is something Anthropic calls 𝗦𝗸𝗶𝗹𝗹𝘀. These are 𝗿𝗲𝘂𝘀𝗮𝗯𝗹𝗲 𝗔𝗜 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀 designed to capture how work is typically done inside organizations. For example: → Reviewing complex financial models → Cleaning large spreadsheets automatically → Generating benchmarking or comparison tables → Turning analysis into presentation ready insights Over time, something important begins to happen. Knowledge that once lived in training manuals, internal playbooks, or the experience of senior employees begins moving into the software itself. And to me, that signals a bigger shift. The software industry may be evolving from: 𝗔𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀 → 𝗪𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀 → 𝗥𝗲𝗮𝘀𝗼𝗻𝗶𝗻𝗴 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 Instead of tools that simply execute commands, we are beginning to see systems that understand the context of the work and help coordinate it. Some observers already refer to this emerging layer as 𝗖𝗹𝗮𝘂𝗱𝗲𝗢𝗦. The real shift here is not just a better chatbot. It is that 𝗔𝗜 𝗶𝘀 𝗯𝗲𝗰𝗼𝗺𝗶𝗻𝗴 𝘁𝗵𝗲 𝗿𝗲𝗮𝘀𝗼𝗻𝗶𝗻𝗴 𝗹𝗮𝘆𝗲𝗿 𝗶𝗻𝘀𝗶𝗱𝗲 𝘁𝗵𝗲 𝘁𝗼𝗼𝗹𝘀 𝘄𝗵𝗲𝗿𝗲 𝘄𝗼𝗿𝗸 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗵𝗮𝗽𝗽𝗲𝗻𝘀. And whoever controls that layer will likely shape how knowledge work is done. So the question I keep coming back to is this: 𝗜𝗳 𝗔𝗜 𝗯𝗲𝗰𝗼𝗺𝗲𝘀 𝘁𝗵𝗲 𝗿𝗲𝗮𝘀𝗼𝗻𝗶𝗻𝗴 𝗹𝗮𝘆𝗲𝗿 𝗮𝗰𝗿𝗼𝘀𝘀 𝘄𝗼𝗿𝗸𝗽𝗹𝗮𝗰𝗲 𝘀𝗼𝗳𝘁𝘄𝗮𝗿𝗲, 𝘄𝗵𝗼 𝘄𝗶𝗹𝗹 𝘂𝗹𝘁𝗶𝗺𝗮𝘁𝗲𝗹𝘆 𝘀𝗵𝗮𝗽𝗲 𝗵𝗼𝘄 𝘄𝗼𝗿𝗸 𝗶𝘀 𝗱𝗼𝗻𝗲? #AI #FutureOfWork #ArtificialIntelligence #Automation #DigitalTransformation

𝗛𝘆𝗱𝗿𝗼𝗴𝗲𝗻 𝗰𝗮𝗿𝘀 𝗸𝗲𝗲𝗽 𝗰𝗼𝗺𝗶𝗻𝗴 𝗯𝗮𝗰𝗸 𝗶𝗻𝘁𝗼 𝘁𝗵𝗲 𝗱𝗲𝗯𝗮𝘁𝗲. For years, most of the focus has been on battery EVs. But recently I came across something interesting: NamX’s hydrogen SUV, developed with Pininfarina. Instead of relying on lithium batteries, the vehicle uses removable hydrogen capsules. A few things stand out: → Around 𝟴𝟬𝟬 𝗸𝗺 𝗿𝗮𝗻𝗴𝗲 → 𝗙𝗮𝘀𝘁 𝗿𝗲𝗳𝘂𝗲𝗹𝗶𝗻𝗴, in just a few minutes → No dependence on 𝗹𝗶𝘁𝗵𝗶𝘂𝗺 The most interesting idea is the CapX capsule system, designed to make hydrogen refueling simpler and more flexible. Hydrogen mobility has struggled for years mainly because of infrastructure. But innovations like this show how engineers keep trying to rethink the problem. Sometimes progress does not come from replacing the dominant technology. Sometimes it comes from challenging the assumptions behind it. Curious: 𝗗𝗼 𝘆𝗼𝘂 𝘁𝗵𝗶𝗻𝗸 𝗵𝘆𝗱𝗿𝗼𝗴𝗲𝗻 𝗰𝗮𝗿𝘀 𝗰𝗼𝘂𝗹𝗱 𝗯𝗲𝗰𝗼𝗺𝗲 𝗮 𝗿𝗲𝗮𝗹 𝗮𝗹𝘁𝗲𝗿𝗻𝗮𝘁𝗶𝘃𝗲 𝘁𝗼 𝗯𝗮𝘁𝘁𝗲𝗿𝘆 𝗘𝗩𝘀? #Innovation #Hydrogen #CleanEnergy #FutureOfMobility #Technology

𝗬𝗮𝗻𝗻 𝗟𝗲𝗖𝘂𝗻’𝘀 𝗰𝗮𝗿𝗲𝗲𝗿 𝗮𝗹𝘄𝗮𝘆𝘀 𝗺𝗮𝗸𝗲𝘀 𝗺𝗲 𝘁𝗵𝗶𝗻𝗸 𝗮𝗯𝗼𝘂𝘁 𝗵𝗼𝘄 𝗶𝗻𝗻𝗼𝘃𝗮𝘁𝗶𝗼𝗻 𝗿𝗲𝗮𝗹𝗹𝘆 𝗵𝗮𝗽𝗽𝗲𝗻𝘀. Today, deep learning powers much of modern AI. But when Yann LeCun began working on neural networks in the 1980s, many researchers believed the approach had little future. He continued anyway. His work on 𝗰𝗼𝗻𝘃𝗼𝗹𝘂𝘁𝗶𝗼𝗻𝗮𝗹 𝗻𝗲𝘁𝘄𝗼𝗿𝗸𝘀 led to LeNet, a system used by banks to recognize handwritten digits on checks. Decades later, similar ideas helped ignite the deep learning breakthroughs around 2012. LeCun later became Chief AI Scientist at Meta and received the 𝗧𝘂𝗿𝗶𝗻𝗴 𝗔𝘄𝗮𝗿𝗱 in 2018 with Geoffrey Hinton and Yoshua Bengio. What I find most interesting is the persistence. Some of the ideas shaping our world today were once considered unfashionable or impractical. Now he is moving on again, founding a startup focused on world models, while continuing his work at NYU. It makes me wonder: 𝗪𝗵𝗶𝗰𝗵 𝗔𝗜 𝗶𝗱𝗲𝗮𝘀 𝘁𝗼𝗱𝗮𝘆 𝗺𝗶𝗴𝗵𝘁 𝗹𝗼𝗼𝗸 𝗼𝗯𝘃𝗶𝗼𝘂𝘀 𝟭𝟬 𝘆𝗲𝗮𝗿𝘀 𝗳𝗿𝗼𝗺 𝗻𝗼𝘄? #AI #DeepLearning #MachineLearning #ArtificialIntelligence #Innovation

𝗘𝘂𝗿𝗼𝗽𝗲 𝗷𝘂𝘀𝘁 𝗯𝗮𝗰𝗸𝗲𝗱 𝗮 $𝟭𝗕 𝗯𝗲𝘁 𝘁𝗵𝗮𝘁 𝗰𝗵𝗮𝗹𝗹𝗲𝗻𝗴𝗲𝘀 𝘁𝗵𝗲 𝗲𝗻𝘁𝗶𝗿𝗲 𝗟𝗟𝗠 𝗿𝗼𝗮𝗱𝗺𝗮𝗽. 𝗬𝗮𝗻𝗻 𝗟𝗲𝗖𝘂𝗻 has spent years arguing that much of today’s AI is essentially “glorified autocomplete.” Now he has $1 billion to prove there may be another path. 😳 For the past few years, the AI race has followed a fairly simple playbook: build bigger models, train them on more data, and add more compute. The results have been remarkable. Systems like ChatGPT, Claude, and Gemini can write, code, analyze information, and reason with language at a level that would have seemed impossible not long ago. But LeCun has been repeating a very uncomfortable idea. 𝗣𝗿𝗲𝗱𝗶𝗰𝘁𝗶𝗻𝗴 𝘁𝗲𝘅𝘁 𝗶𝘀 𝗻𝗼𝘁 𝘁𝗵𝗲 𝘀𝗮𝗺𝗲 𝗮𝘀 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲. That belief is now shaping his next move. His new company, 𝗔𝗱𝘃𝗮𝗻𝗰𝗲𝗱 𝗠𝗮𝗰𝗵𝗶𝗻𝗲 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 (𝗔𝗠𝗜), is focused on building something very different: 𝗪𝗼𝗿𝗹𝗱 𝗠𝗼𝗱𝗲𝗹𝘀. Instead of training mainly on internet text, these systems aim to learn from real-world signals such as video, spatial interaction, and physical environments. In simple terms: → Language models learn patterns in text → World models try to learn patterns in reality 𝗪𝗵𝗮𝘁 𝗺𝗮𝗸𝗲𝘀 𝘁𝗵𝗶𝘀 𝗺𝗼𝘃𝗲 𝗶𝗻𝘁𝗲𝗿𝗲𝘀𝘁𝗶𝗻𝗴 𝘁𝗼 𝗺𝗲 𝗶𝘀 𝗻𝗼𝘁 𝗷𝘂𝘀𝘁 𝘁𝗵𝗲 𝘁𝗲𝗰𝗵𝗻𝗼𝗹𝗼𝗴𝘆, 𝗯𝘂𝘁 𝘁𝗵𝗲 𝘀𝘁𝗿𝗮𝘁𝗲𝗴𝗶𝗰 𝗯𝗲𝘁 𝗯𝗲𝗵𝗶𝗻𝗱 𝗶𝘁. AMI reportedly raised about $1.03B at a valuation near $3.5B, with backing from groups including Bezos Expeditions, Nvidia, Temasek, and HV Capital. But the real story here is not the funding. 𝗧𝗵𝗶𝘀 𝗶𝘀 𝗮 $𝟭𝗕 𝗯𝗲𝘁 𝘁𝗵𝗮𝘁 𝘀𝗰𝗮𝗹𝗶𝗻𝗴 𝗹𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗺𝗼𝗱𝗲𝗹𝘀 𝗮𝗹𝗼𝗻𝗲 𝗺𝗮𝘆 𝗻𝗼𝘁 𝗯𝗲 𝘁𝗵𝗲 𝗽𝗮𝘁𝗵 𝘁𝗼 𝗿𝗲𝗮𝗹 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲. Less autocomplete. More understanding. 𝗜𝗳 𝘁𝗵𝗮𝘁 𝗯𝗲𝘁 𝗶𝘀 𝗿𝗶𝗴𝗵𝘁, 𝗶𝘁 𝘄𝗼𝘂𝗹𝗱 𝗰𝗵𝗮𝗻𝗴𝗲 𝗵𝗼𝘄 𝘄𝗲 𝘁𝗵𝗶𝗻𝗸 𝗮𝗯𝗼𝘂𝘁 𝘁𝗵𝗲 𝗻𝗲𝘅𝘁 𝗲𝗿𝗮 𝗼𝗳 𝗔𝗜. Today’s AI boom may not be the finish line. It might be the beginning. Curious how you see this playing out. Will the next major breakthrough come from scaling language models even further… or from systems that learn how the real world actually works? #AI #ArtificialIntelligence #MachineLearning #FutureOfAI #Innovation

𝗛𝗼𝘄 𝗯𝗶𝗴 𝘁𝗲𝗰𝗵 𝗿𝗲𝗮𝗹𝗹𝘆 𝗼𝗿𝗴𝗮𝗻𝗶𝘇𝗲𝘀 𝘁𝗲𝗮𝗺𝘀. 📦 𝗔𝗺𝗮𝘇𝗼𝗻 → clean hierarchy, optimized for execution. 🧠 𝗚𝗼𝗼𝗴𝗹𝗲 → brilliant ideas… connected by 400 meetings. 🌐 𝗙𝗮𝗰𝗲𝗯𝗼𝗼𝗸 → everything connected to everything. 🪟 𝗠𝗶𝗰𝗿𝗼𝘀𝗼𝗳𝘁 → silos… occasionally firing at each other. 🍏 𝗔𝗽𝗽𝗹𝗲 → one vision in the center. ⚖️ 𝗢𝗿𝗮𝗰𝗹𝗲 → legal approves the org chart. Satire, of course. But every joke about org charts usually hides a bit of truth. 𝗡𝗼𝘄 𝗮 𝗺𝗼𝗿𝗲 𝗶𝗻𝘁𝗲𝗿𝗲𝘀𝘁𝗶𝗻𝗴 𝗾𝘂𝗲𝘀𝘁𝗶𝗼𝗻: 𝗪𝗵𝗮𝘁 𝘄𝗶𝗹𝗹 𝘁𝗵𝗲 𝗼𝗿𝗴 𝗰𝗵𝗮𝗿𝘁 𝗹𝗼𝗼𝗸 𝗹𝗶𝗸𝗲 𝘄𝗵𝗲𝗻 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁𝘀 𝗮𝗿𝗲 𝗽𝗮𝗿𝘁 𝗼𝗳 𝘁𝗵𝗲 𝘁𝗲𝗮𝗺? Humans managing humans… or humans managing agents that manage other agents? Curious what you think. #AI #AgenticAI #FutureOfWork #TechCulture #Automation