In the traditional organization, decision rights were implicit. Maybe your team refined them through time and office politics. Maybe your team pulled them out of a fortune cookie. But chances are that no one wrote them down. The AI-first organization has to write them down in a form that AI agents can read. The side effect is that decision rights become visible to people, which makes them negotiable. That is yet another uncomfortable way in which AI makes leadership harder in ways that aren't obvious when all you're thinking about is the tech... In this episode, I walk through three changes to how organizations need to be designed for the AI era. 📍 AI Advice From Anywhere: Watch to the end and see if you can guess the location before the reveal. Post the number of seconds it took you in the comments. (No spoilers.) #AAfA #DecisionIntelligence #AILeadership #FutureOfWork #ExecutiveDecisionMaking #AIAdoption

Rest in peace, prompt engineering. The prompt is only ever a symptom of clear thinking, not a skill in itself, which means one model update and your prompting edge is gone. And the model itself is becoming a commodity: what gets commoditized is anything most humans can do reliably at scale. For now that feels dazzling. But dazzling is not the same as scarce. 📍 In this episode of AI Advice From Anywhere, I walk through what's actually fragile, what's becoming a commodity, and what that leaves behind. Watch to the end and see if you can guess the city before the reveal. Post the number of seconds it took you in the comments. No spoilers. Invite Cassie to speak at your event: https://makecassietalk.com Learn directly from Cassie: https://decisiongptcourse.com Newsletter: https://decision.substack.com Intro 1:1s: https://intro.co/CassieKozyrkov

55% of employers admit to regretting their AI-driven layoffs. I keep hearing stories of astronomical offers made to, and often rejected by, ex-employees. Gartner projects that by 2027, half of enterprises without a people-centric AI plan will lose their top talent. And that's before you factor in the extra coordination burden that complex systems create. In this episode of AI Advice From Anywhere, I walk through what the data says about where this is heading, and what leaders who are thinking clearly about this need to understand before making a move they can't undo. 📍 Watch to the end and see if you can guess the city before the reveal. Post the number of seconds it took you in the comments. No spoilers. Invite Cassie to speak at your event: https://makecassietalk.com Learn directly from Cassie: https://decisiongptcourse.com Newsletter: https://decision.substack.com Intro 1:1s: https://intro.co/CassieKozyrkov

What do you call it when someone pours their heart out without knowing they’re talking to a bot? Business as usual. AI is making it easier to imitate the messy little signals we use to decide whether someone is real, safe, caring, charming, or in trouble. Best case? A disappointing date with someone whose messages had a little too much help from ChatGPT de Bergerac. Worst case? A panicked call from someone you love, except the voice isn’t theirs. In this episode of AI Advice From Anywhere, Cassie Kozyrkov walks through three relationship-based AI scams worth knowing about: from awkward to criminal to the kind every family should prepare for before anything goes wrong. 📍 AI Advice From Anywhere: Watch to the end and see if you can guess the city before the reveal. Post the number of seconds it took you in the comments. (No spoilers.) #AAfA #DecisionIntelligence #AILeadership #FutureOfWork #ExecutiveDecisionMaking #AIAdoption ********************************** Invite Cassie to speak at your event: https://makecassietalk.com Learn directly from Cassie: https://decisiongptcourse.com/ Newsletter: https://decision.substack.com Intro 1:1s: https://intro.co/CassieKozyrkov

Most AI ROI dashboards are grading the easy stuff. That should worry you. The easiest things to measure are often the easiest things to automate. And once AI is pointed at a metric, what if it just helps you do stupid things faster? Just like coffee! But what happens when what’s easiest to count is not what actually counts? In this episode of AI Advice From Anywhere, Cassie Kozyrkov walks through the metric trap hiding inside AI ROI, why Goodhart’s Law matters more than ever in the age of automation, and what leaders risk eroding when they optimize only for what shows up neatly on a dashboard. 📍 AI Advice From Anywhere: Watch to the end and see if you can guess the city before the reveal. Post the number of seconds it took you in the comments. (No spoilers.) #aafa #DecisionIntelligence #AILeadership #FutureOfWork #ExecutiveDecisionMaking #aiadoption ********************************** Invite Cassie to speak at your event: https://makecassietalk.com Learn directly from Cassie: https://decisiongptcourse.com/ Newsletter: https://decision.substack.com Intro 1:1s: https://intro.co/CassieKozyrkov

Your AI adoption number is probably lying to you, and not for the reason you think. 57% of employees globally conceal their AI use from managers. 80% admit to using unapproved tools. Nearly half have entered sensitive company data into public models their employer doesn't control. The standard response is to push harder on adoption. More training, better rollout, stricter policy. That response is solving the wrong problem. In this episode I walk through what the data shows and why you can't policy your way out of a trust problem. 📍 AI Advice From Anywhere: Watch to the end and see if you can guess the city before the reveal. Post the number of seconds it took you in the comments. (No spoilers.) #aafa #DecisionIntelligence #AILeadership #FutureOfWork #ExecutiveDecisionMaking #AIAdoption ********************************** Invite Cassie to speak at your event: https://makecassietalk.com Learn directly from Cassie: https://decisiongptcourse.com/ Newsletter: https://decision.substack.com Intro 1:1s: https://intro.co/CassieKozyrkov

Now that the machines do the doing, what’s left? In 2017, Google published the paper that introduced the Transformer architecture, the basis of every modern LLM (the “T” in ChatGPT). Its math is dry, but the title is accidental poetry: Attention Is All You Need. As execution gets cheaper, judgment gets more important. You now sit one promotion up the stack, closer to the work that was always hardest: deciding what deserves the machine’s next move. AI systems run on memories, constraints, and objectives, explicitly or from culture and habit seeping in. The question is not whether there is a steering force. The question is whether it’s applied on purpose. That force is attention. Tune a customer support AI for speed and the tickets close fast. Tune it for resolution quality and the agent slows down to listen. The customer, of course, may feel differently about what ‘closed’ means. The system follows your attention, whether you encode it cleanly or by chance. The defining characteristic of your leadership becomes the sheer naked quality of what you choose to focus on. Easy tasks let you skip the choice, but the future will have automated them all. You’ll have to prioritize. Ashby’s Law of Requisite Variety, an enduring classic from 1956 (the same year that gave us the term Artificial Intelligence), says that a control system must be at least as capable as the system it governs. Scale up what a system can do and you must scale up what steers it, or the steering stops being steering and becomes weather. You inherited a controller built for a smaller century. Attention has always been valuable. It is what turns time into progress. What is new is how fast it now compounds. We know what compound interest does to a dollar set aside in 1956. We’ve yet to learn the same respect for an hour of well-pointed attention today, though the term is shorter and the bank requires no signature. The leader who waits will be steered by the weather. To manage hybrid teams is to design attention scaffolding; the systems you do not choose will be chosen for you by people whose incentives are not yours. Build feedback loops that sharpen judgment. Resist repetition unless it’s an investment in less repetition next time; teach a machine instead so your attention can work for you while you sleep. Your role is not to do more. It is to see more clearly. To decide what matters. To guide complexity without pretending you control it completely. And to pay attention to your priorities. Literally. Not everyone will recognize the agentic era as an opportunity for compounding attention. Many will let theirs wander to whatever is loudest. That creates an opening for those who choose differently. The leader of the next decade is the one who knows where their attention is going, why it is going there, and what they would have to stop doing to get it back. Attention is all you need. 📍 AI Advice From Anywhere: Watch to the end and see if you can guess the city before the reveal. Post the number of seconds it took you in the comments. (No spoilers.) #aafa #DecisionIntelligence #AILeadership #FutureOfWork#ExecutiveDecisionMaking #aiadoption ********************************** Invite Cassie to speak at your event: http://makecassietalk.com Newsletter: http://decision.substack.com Learn directly from Cassie: https://decisiongptcourse.com/ Intro 1:1s: https://intro.co/CassieKozyrkov

You don’t have to suffer through writing that doesn’t work for you anymore. If you’re reading something mainly to get the information out of it, give tradition the boot and use an LLM to reshape it into a format that actually fits the way you think. Prefer audio? Listen to it. Prefer structure? Turn it into bullet points. Prefer visuals? Turn it into a comic or infographic. Prefer to interact? Ask questions about it. Prefer dragons? Ask for examples that involve dragons. 🐉 LLMs are amazing at converting between formats, including from writing that bores you to writing you can absorb with ease. This means you will learn more overall. Of course, these systems can still hallucinate. So if you're going to act on what you learn, verify first. If you weren't going to act on it, you'll probably forget the details anyway, so there's no harm done. Remember: the higher the stakes, the less you should take on faith. 📍 AI Advice From Anywhere: Watch to the end and see if you can guess the city before the reveal. Post the number of seconds it took you in the comments. (No spoilers.) P.S. I'm not saying don't read books. I'm saying don't read BORING books/articles/guides. If you love the prose, then revel in it!. But if you're just there for information, convert it to your preferred info style. And you're not even sure if it's worth reading in the first place, summarize it first! #aafa #DecisionIntelligence #AILeadership #FutureOfWork#ExecutiveDecisionMaking #aiadoption ********************************** Invite Cassie to speak at your event: http://makecassietalk.com Newsletter: http://decision.substack.com Intro 1:1s: https://intro.co/CassieKozyrkov

Claude Mythos is basically a metal detector for ancient landmines... except the beach is the entire internet. Anthropic's monster new model has been withheld from public release and in this episode of AI from Anywhere, let's talk about why. In a nutshell: it's likely too expensive for consumers. And while enterprises might be able to afford it, most are still struggling to extract value from the AI they already have. That gap has a name: capability overhang. The models in people’s hands are often far more capable than the value they’re getting out of them. That’s not a failure of AI. It’s a failure of leadership and imagination. So who *does* have the imagination for this? Who most wants a model this capable, and knows exactly what to do with it? Cyber defenders and cyber criminals. Which is why withholding Mythos from the general public and only sharing it with serious defenders in the Project Glasswing consortium is all about who gets the cyber capability jump first: attackers or defenders. 📍 AI Advice From Anywhere: Watch to the end and see if you can guess the city before the reveal. Post the number of seconds it took you in the comments. (No spoilers.) #aafa #DecisionIntelligence #AILeadership #FutureOfWork#ExecutiveDecisionMaking #aiadoption If AI-mediated systems are shaping what gets seen and selected, it’s worth being intentional about how we show up within them: Learn to use LLMs to upgrade your decision-making: decisiongptcourse.com Invite Cassie to speak at your event: http://makecassietalk.com Newsletter: http://decision.substack.com Intro 1:1s: https://intro.co/CassieKozyrkov

Turns out AI is biased... against humans and in favor of AI! A recent paper suggests something surprising: when AI systems evaluate content, they may be more likely to favor writing that looks like it was produced with AI tools. That might not sound like a big deal, until you remember how much of the digital world is now filtered by algorithms. AI systems increasingly rank, grade, shortlist, and recommend what gets seen. So if AI tends to favor AI-style content, an obvious question follows: What exactly is being rewarded? In this episode of AI Advice from Anywhere, Cassie Kozyrkov explores what this research might mean in practice, and why leaders should pay attention to how AI-mediated systems quietly shape outcomes behind the scenes. 📍 AI Advice From Anywhere: Watch to the end and see if you can guess the city before the reveal. Post the number of seconds it took you in the comments. (No spoilers.) #aafa #DecisionIntelligence #AILeadership #FutureOfWork#ExecutiveDecisionMaking #aiadoption If AI-mediated systems are shaping what gets seen and selected, it’s worth being intentional about how we show up within them: Learn to use LLMs to upgrade your decision-making: decisiongptcourse.com Invite Cassie to speak at your event: http://makecassietalk.com Newsletter: http://decision.substack.com Intro 1:1s: https://intro.co/CassieKozyrkov

NVIDIA CEO Jensen Huang described AI as a five-layer cake. 🍰 Chips, models, applications, energy, data centers. A full stack. A complete system. A solved problem. Delicious. 🍴 Except there’s a layer missing. When everyone has access to the same models, the same tools, and the same compute, it’s tempting to assume the playing field is level. It's not. The gap doesn’t disappear. It moves. In this episode of AI Advice from Anywhere, Cassie Kozyrkov talks about the part of the AI stack that isn’t hardware, software, or data, but ends up determining everything anyway. 📍 AI Advice From Anywhere: Watch to the end and see if you can guess the city before the reveal. Post the number of seconds it took you in the comments. (No spoilers.) (Oh, and if you haven’t seen Jensen's breakdown, here it is: https://blogs.nvidia.com/blog/ai-5-layer-cake/) #aafa #DecisionIntelligence #AILeadership #FutureOfWork#ExecutiveDecisionMaking #aiadoption ********************************** Invite Cassie to speak at your event: http://makecassietalk.com Newsletter: http://decision.substack.com Intro 1:1s: https://intro.co/CassieKozyrkov

Claude Computer Use is here. 💪 Still using AI in the browser like it's 2025? Then prepare to have your mind blown by what you've been missing out on! It's all much easier to use than you think. Let me show you in this candid demo and intro to some of my favorite HANDS-FREE AI tools, including: ⭐ Wispr Flow - type everywhere with your voice, throw away your keyboard ⭐ Claude Code - let AI do tasks that make changes to your desktop ⭐ Claude Cowork (new) - let AI do tasks that make changes to your desktop... in a more cuddly interface ⭐ Claude Dispatch (new) - let AI do tasks that make changes to your desktop... from your phone ⭐ Claude Computer Use (NEW!!) - exactly what it sounds like: let Claude move the mouse, type, and click around your Mac to operate apps like a human Plus a short intro to running agentic pipelines. (All of it friendly to non-technical folks!) In the demo, I've got Claude directly using - among other things - ChatGPT and Gemini (blasphemy!) and as usual, I'm chatting about AI while we wait for things to run. Unless otherwise indicated, everything happens at 1x speed so you get a real taste of what to expect. #claude #ai #computeruse If you liked it, share it with someone to encourage me! If you really liked it, subscribe and comment! ********************************** Invite Cassie to speak at your event: http://makecassietalk.com Newsletter: http://decision.substack.com Intro 1:1s: https://intro.co/CassieKozyrkov

What happens when AI agents try to do real professional work instead of benchmark tests? In this episode of AI Advice from Anywhere, Cassie Kozyrkov looks at new research comparing AI performance on controlled tasks vs. real freelance projects. The gap is bigger than most headlines suggest. If you’re making decisions about AI adoption, automation, or workforce strategy, this distinction matters more than you think. Topics in this episode: Why benchmarks can be misleading The difference between tasks and jobs Why real work breaks AI more often than demos do How to think about AI reliability in the real world 📍City challenge: How many seconds did it take you to realize this episode was filmed in [redacted]? Comment the number of seconds ONLY. Let others play too! 😏 Subscribe for more AI Advice from Anywhere. PS: If you're up for some long reading, you can check out the full papers here: APEX: https://arxiv.org/pdf/2601.14242 RLI: https://arxiv.org/pdf/2510.26787 #AAfA #DecisionIntelligence #AIAdoption #FutureOfWork #AIStrategy #LeadershipThinking #ResponsibleAI

Everyone wants to be AI-first. But what does that actually mean? In this episode of AI Advice from Anywhere, Cassie Kozyrkov explores a key distinction many leaders miss: the difference between being AI-first as an individual and AI-first as an organization. Those two ideas are often conflated, and when they are, companies end up making expensive mistakes. In this walk-and-talk episode, Cassie covers: • How individuals can use AI as a powerful advice engine • The skill most people overlook when asking AI for help • Why blindly trusting AI outputs creates new risks • Why organizations should not simply “add AI” to everything • The leadership mindset required to make AI genuinely valuable If you're a leader thinking about AI strategy, this episode will challenge some common assumptions. 📍AI Advice From Anywhere: can you guess where in the world I am before I reveal it? Subscribe for more insights on decision intelligence, AI strategy, and better thinking in the age of AI. ********************************** ⭐ Join Cassie's latest course! Decision-Making with ChatGPT: http://bit.ly/decisiongptcourse ********************************** Invite Cassie to speak at your event: http://makecassietalk.com ✨ Decision advising: http://decisionadvising.com Content in other formats: http://kozyr.com Intro 1:1s: https://intro.co/CassieKozyrkov #AAfA #aileadership #decisionintelligence #enterpriseai #innovation #aistrategy #aiadvicefromanywhere #businesstrategy #aiadoption #decisionmaking ********************************** Invite Cassie to speak at your event: http://makecassietalk.com Decision advising: http://decisionadvising.com Content in other formats: http://kozyr.com Intro 1:1s: https://intro.co/CassieKozyrkov

Can you ship a million lines of software without writing any code? OpenAI just did. Over five months, an internal product grew to ~1M lines — all generated by AI agents. Estimated build time: roughly 1/10th of a traditional approach. That’s the headline. The more interesting part is how they did it. From day one, they imposed a hard constraint: no manually written code. Not a single line. If something needed to change, the agent had to do it. That rule forced discipline. The team couldn’t “just fix it.” If the agent struggled, the answer wasn’t to edit the code... it was to improve the environment. Add guardrails. Expose better context. Clarify architecture. Encode standards. Strengthen feedback loops. ⚒️ This is harness engineering. Out of that constraint came two outcomes: 1. Proof that large-scale agent-driven development is viable 2. A set of practical patterns for making it reliable In the video, I distill 12 principles that made it work: principles around legibility, boundaries, documentation, throughput, cleanup, and escalation. The discipline didn’t disappear. It moved. From writing code to designing systems agents can operate within. If you care about where software engineering is heading, this shift in philosophy deserves your attention. 📍AI Advice From Anywhere: can you guess the city before I reveal it? Reference: https://openai.com/index/harness-engineering/ ********************************** ⭐ Join Cassie's latest course! Decision-Making with ChatGPT: http://bit.ly/decisiongptcourse ********************************** Invite Cassie to speak at your event: http://makecassietalk.com ✨ Decision advising: http://decisionadvising.com Content in other formats: http://kozyr.com Intro 1:1s: https://intro.co/CassieKozyrkov #AAfA #aileadership #decisionintelligence #enterpriseai #innovation #aistrategy

You can “win” your AI rollout and still lose your workforce. A recent study of 430 manufacturing employees found that even successful AI transformations increase emotional exhaustion. Translation? Your KPIs can trend up 📈, while motivation trends down 📉. The research identifies two distinct fears: • Replacement fear: “Will AI take my job?” • Learning fear: “Can I keep up?” They look similar. They are not. Each requires a different leadership response. Most organizations treat them as one. That’s where transformation quietly derails. AI success isn’t just about algorithms--it’s about emotional infrastructure. If you’re accountable for AI outcomes, this week’s AI Advice from Anywhere breaks down: – Why anxiety drains discretionary effort – Why service-oriented leadership isn’t enough – Why learning time must be explicitly protected – And why burnout should be tracked alongside AI KPIs 📍AI Advice From Anywhere: can you guess the city before I reveal it? If you’re leading AI transformation, this one matters. ********************************** ⭐ Join Cassie's latest course! Decision-Making with ChatGPT: http://bit.ly/decisiongptcourse ********************************** Invite Cassie to speak at your event: http://makecassietalk.com ✨ Decision advising: http://decisionadvising.com Content in other formats: http://kozyr.com Intro 1:1s: https://intro.co/CassieKozyrkov #AAfA #aileadership #decisionintelligence #enterpriseai #innovation #aistrategy

AI rarely fails because it’s inaccurate. It fails because leaders plan as if accuracy were the point. Most teams ask: “How good is the model?” Too few ask: “What happens when it’s wrong?” ⚠️ If 10% of outputs fail: • Who notices? • Who escalates? • Who absorbs the impact? If those answers aren’t clear before deployment, the gap isn’t technical. It’s leadership. AI at scale is not a tooling problem. It’s a decision-design problem. 📍 AI Advice From Anywhere: how many seconds did it take you to identify the city? (Extra credit if you can name the first clue that gave it away.) ********************************** ⭐ Join Cassie's latest course! Decision-Making with ChatGPT: http://bit.ly/decisiongptcourse ********************************** Invite Cassie to speak at your event: http://makecassietalk.com ✨ Decision advising: http://decisionadvising.com Content in other formats: http://kozyr.com Intro 1:1s: https://intro.co/CassieKozyrkov #AAfA #aileadership #decisionintelligence #enterpriseai #innovation #aistrategy

Using AI well is a modern version of Pascal's Wager. What's the downside of asking? You lose a sentence. Maybe a minute. What's the upside? You make progress on your biggest questions and problems that used to feel stuck. or the first time in history, you don't need technical fluency (or permission) to ask for help thinking something through. The limiting factor is no longer the tool. It's whether you're clear on what actually matters. And because asking is now essentially free, shallow questions aren't a tooling problem. They're a self-leadership one. Your prompt history reflects what you care about, what you're trying to fix, and what you've quietly decided isn't worth improving anymore. In this episode of AI Advice From Anywhere, I invite you to audit your prompt history and notice what it says about your priorities. I'd love to hear what surprised you. And yes, there's also a city to guess! If you want to build the skill of asking better questions and making better decisions with AI, that's what I teach in my Decision-Making with ChatGPT course: 👉 http://decisiongptcourse.com #AAfA #DecisionIntelligence #AILeadership