Latest / Beyond the Byte / Everyone Has the Same AI. Here's What Actually Gives You an Edge

Transcript

Everyone Has the Same AI. Here's What Actually Gives You an Edge · Beyond the Byte

• Mo Hafez: Hi. It's been a little while, so I wanted to make this video. The last time I sat down for Beyond the Byte, I was talking with Rebecca Krauthamer from QSecure about quantum computing. And since then, a lot has happened in technology, obviously, but also in the way that I've been thinking about this show. I've spent the last few months teaching and advising companies and working with emerging technologies and having lots of conversation with people trying to figure out the same basic thing. What does all of this actually mean for us? I mean us humans. Not just what AI can do, not which new model is better, but what are these technologies changing? What do they make possible? What do they make us dependent on? And what do we need to become capable of in response? And somewhere in all of that, I realize that's what Beyond the Byte is really about. So I'm bringing the show back a little differently, more conversations. More questions, more exploration beyond product announcements and the hype. And I wanted to start with an idea that has become pretty central to how I think about technology now. Technology is the easy part. Capability is the moat. So welcome back to Beyond the Byte. If you're not subscribed to the channel, please do subscribe. Sometimes the YouTube algorithm randomly unsubscribes you, so please check if you're subscribed. And now, let's get on with the show. You know what I think is funny about technology? We spend an enormous amount of time talking about technology, which sounds obvious. But think about it. Every few years some new thing arrives, and suddenly everybody becomes obsessed with that thing itself. Think about it, cloud, blockchain, VR, AI, agents, robots. Whatever happens to be having its moment. And immediately we start comparing all the features. Who has the better product? Whose is faster? Whose is cheaper? Whose ahead? What model has the better benchmark? What company raised the most money? What tool should I buy? And I understand why. Technology is tangible. You can demo it, you can put it on a slide, you can show somebody the interface. You can say, look what it does. But the more I've spent time around technology, the more I'm convinced that the technology is usually the easy part. Not easy to invent. I'm definitely not saying that. Some of the engineering happening right now is, I mean, just downright extraordinary. But what I mean is that once the technology exists, once people can buy it, once everyone has access to roughly the same tools, the real competition moves elsewhere. It moves into the people, the organizations, the habits, the judgment, the ability to actually do something useful with the thing. And that's where I keep coming back to this idea. Technology is the easy part. Capability is the moat. Look, take generative AI right now. Almost every company I'm talking to has access to roughly the same universe of tools: Chat GPT, Claude, Gemini, Copilot, open source models, APIs, automation platforms, agents, right? Whatever. There's no secret website where only the best companies are allowed to download artificial intelligence. Everybody can buy the shovel, so to speak. And so that's not the advantage anymore. The question is, what can your organization actually build with it? What problems can your people recognize? Can they tell a good use case from a bad one? Can they redesign a workflow? Can they challenge the output? Can they protect sensitive information? Can they decide what should be automated and what absolutely should not? And can they measure whether any of this is actually working? Can they operate when the vendor changes something? Can they learn faster than the technology changes? That's that's capability. And I think we're underestimating how important that distinction really is. You see, there's an old pattern in technology. So a new capability appears, and for a little while, the people who simply possess the technology, they're the ones that have the advantage. Then the technology spreads. Prices fall. Competitors just copy it. It becomes easier to access. Right? This happened with many technologies, the iPod, VR, you name it. Eventually the advantage shifts from simply having the tool to knowing what to do with it. Computers did this, the internet did this, cloud computing did this, smartphones did this. Think about websites. Like there was a time when simply having a website made your company look technologically sophisticated. Today, if a company says, our competitive advantage is that we have a website, I have some bad news for you. Because the technology became ubiquitous. The differentiator moved upward. What experience did you create? Right? How well do you understand your customer? How fast can you learn? How good are your operations? How strong is your brand? How trustworthy is your brand? The tool became infrastructure. The capability became the competitive advantage. And AI is going through that transition incredibly quickly. And this is why I get a little nervous when organizations start their AI strategy now by making a shopping list of tools that they need. we need Copilot, we need a Enterprise Chat GPT license, we need agents, we we need a chatbot. Okay, I mean, maybe, but why? Like, what are you actually trying to become capable of doing? And you see, that's a very different question. Because technology is, in a word, Seductive. It gives us something concrete to point at. Capability, now that's messier, right? Capability might mean that your finance team understands enough about AI to identify three processes worth automating. It might mean that your legal team knows when AI creates unacceptable risk. It might mean your executives know how to evaluate an AI proposal instead of simply approving whatever vendor sounds impressive. It might even mean your employees know when the machine is wrong. It might mean your IT team can swap models without rebuilding the entire business from the ground up. It might mean your organization has data that people actually find. None of that demos particularly well. Nobody walks into a conference booth and says, Would you like to see our phenomenal organizational judgment? Right? There's no blinking LED for that. But that's often the thing that determines whether the technology actually creates value. Think about Formula One. Right. You can give somebody the fastest car in the world. That doesn't make them a Formula One driver. Actually, it probably makes them dangerous. Because the machine amplifies whatever capability already exists inside. Right? The driver needs training, reflexes, judgment, an understanding of the car that they're driving. The engineers similarly need data. The pit crew needs coordination. The strategy team needs to understand tires, weather, com competitors, timing. The car matters enormously, but the car sits inside a system of capability. And without that system, you've basically bought a very expensive way to crash really fast. And I think a lot of companies are currently buying faster cars. And to be fair, this isn't new with AI. I mean we did this with the digital transformation. We spent years hearing companies announced that they were becoming digital. And sometimes what that meant was they bought software. That's not transformation, that's procurement. You can buy Salesforce and still have terrible sales operation. You can buy Slack and still communicate badly. You can buy a data platform and still make decisions based on whoever speaks loudest in the meeting. You can install Microsoft Teams and somehow still schedule a meeting to discuss why there are too many meetings. The presence of technology does not guarantee the presence of capability. And AI makes that mistake more dangerous because AI Feels intelligent. When software produces something fluent, when it talks confidently, when it generates strategy or a report or an analysis, it creates this temptation to confuse the machine's capability with our own. That's a very subtle mistake. Because there's a difference between having access to intelligence and becoming more intelligent as an organization. See, those are those are not the same things. I mean, I can give everyone in a company access to ChatGPT tomorrow. Congratulations, we now have AI. Do we make better decisions all of a sudden? Do customers have a better experience? Did we end up reducing more waste? Did we create something new? Are the employees in the company all of a sudden more capable? did we shorten the time between identifying a problem and solving it? Did we learn anything? Maybe. But maybe not. And if the answer is not really, well then the problem isn't the model. And this is where I think people misunderstand the word skill. Because when we talk about AI capability, everybody immediately thinks, you know, skills, AI skills like prompt engineering. How how do I write better prompts? A and yes, you know, right now, learn that. That's useful for now. But that's such a thin layer of what capability actually means. Right? The person who understands your customers deeply and knows how to use AI is probably more valuable than somebody who knows 97 clever prompting techniques, but has no idea what the customer really wants. The person who understands your operation, who recognized the weird bottlenecks that everybody else has just accepted as normal, and then sees how AI could remove it, that's capability. The executive who knows enough about the technology to ask, Why are we sending our proprietary data into that system? Right? That's capability. Like the employee who who gets a an AI-generated answer and then says, you know, that looks convincing, but I think it's totally wrong. That's capability. Or the team that knows how to experiment cheaply, learn, and then discard what doesn't work, also move on without turning every pilot into this political battle? Well, that's capability too. And the interesting thing is most of these capabilities aren't really about AI. They're about thinking, learning, good judgment, curiosity, systems. Great communication strategy. AI just makes them all the more valuable. And that's one of the reasons I'm actually pretty optimistic about this moment, because there's this narrative about AI that is going around that to make human capability less important, like we can rely more on the machines and the humans can just sit back. And I think, yes, in some areas, certain tasks will absolutely become less valuable. Like if your entire job is producing something the machine can reliably produce faster and cheaper and reliably, well, that's a real challenge. We shouldn't pretend otherwise. But zoom out, right? The cheaper that execution becomes, the more important judgment becomes. Right. If I can generate ten marketing campaigns in ten seconds, well the scarce skill is no longer creating the ten campaigns, it's knowing which ones deserve to exist, which ones should be followed up on and made. Right. If I can generate fifty product ideas, well the scarce skill now is knowing which problem matters to build solutions for. If I can produce code faster, the scarce skill becomes software architecture, prioritization. Security, right? Understanding the user. If every executive can generate a strategic plan, well the scarce skill now becomes knowing whether the plan makes any sense at all. Right? The bottleneck moves. Technology removes one constraint and reveals another. And increasingly I think the constraint we're revealing is human and organizational capability. You see, there's a a great temptation in business to assume that efficiency is the same thing as advantage. It's not. If everybody has the same efficiency gain, well, that's just the new baseline. I mean, imagine if AI makes everybody twenty percent faster. Great. Now everybody's twenty percent faster. What happens next? The competitive advantage moves somewhere else. Right? Who understands their customers better? Who makes better decisions? Who attracts better people? Who experiments faster? Who has trust? Who has proprietary knowledge? Who has relationships? Who can coordinate? Who knows when to slow down? Who can use all of this new speed without driving straight into a wall? That's where strategy lives. And I like to play chess, so I think of this a little bit like chess. Right? chess engines, became vastly better than humans years ago. But chess didn't disappear. People didn't say, well, you know, Stockfish, which is a AI model for playing chess, can play better than me. I guess that's it. You know, chess is over. No, humans now started learning with these engines, studying with these engines, testing ideas, analyzing mistakes. And something interesting happened. The tool changed how people trained, it raised the level of play. But owning stockfish wasn't the competitive advantage. I mean, everybody had access to stockfish. The advantage was what you learned from it, how you incorporated it into your preparation, what new patterns you recognized, how you performed when sitting across from another human being and the clock was running. Access became a commodity, but capability remained scarce. And I think organizations should be asking themselves the same question right now, right? If every competitor eventually gets access to roughly the same AI, What do we want to become unusually good at? Right? What does that organization or that individual want to become unusually good at with these tools? That's the question. Not what software should we own or you know, it's it's more like what capability should we own? Maybe that's customer intimacy, knowing your customer really well, the way Amazon does. Maybe it's research, maybe it's product development, maybe it's scenario planning, maybe it's personalization. Or maybe it's regulatory navigation, right? The ability to to to ensure that you're up to date with the regulatory compliance as these tools evolve. Maybe it's building better operational speed in your organization. Maybe it's storytelling. Maybe it's something very specific to your industry that outsiders simply don't understand. But if you don't know what capability you're trying to build, well the technology will happily give you a thousand things that you could automate. And you're gonna end up confusing movement with progress. Quick aside, because this idea, capability is the moat, is really at the center of what we do at ByteWise. A lot of organizations come to AI from the technology side first. They've bought the tools already, they've run some pilots, maybe everybody already has co-pilot, maybe someone wants agents, maybe leadership just knows they need an AI strategy. And our job is usually to move that conversation upstream. Right? We ask the questions, what are you actually trying to become capable of? Where would AI create leverage in your organization? What What skills need to exist within your people? What should you buy? What should you build? And what should you never become dependent on? And how do you make sure the technology supports the organization you're trying to become? Instead of becoming one more layer of software that nobody quite knows what to do with. That's the strategy work that we do at ByteWise. If those conversations are happening inside your organization right now, come find us, follow the work, visit the website, or reach out and start a conversation with us. All right, back to it. Here's the harder part. Capability takes time. Technology doesn't. I mean I I can buy the software today. I can announce a partnership today. I can hire a new vendor today. I can put AI powered on the website this afternoon. But Building capability that's slower. People need time to learn. Managers need to change how they work. Processes within the organization have to change. Sometimes the incentives have to change too. Data has to get cleaned up. People have to make mistakes and learn from those mistakes. And organizations have to build muscle memory. And that's why companies keep trying to skip it. We want transformation without apprenticeship, right? We we want the result without becoming the kind of organization capable of producing the result. And technology companies are very happy to sell us that dream. Because buy this product and become transformed is much easier, is a much easier sales pitch than you buy this product and then spend eighteen months changing how your organization thinks to make it actually work. Right. One fits nicely on a billboard. The other is a lot messier. But the companies that really benefit from these technologies, you see, I suspect they're going to be the ones that treat adoption more like education than installation. Because installation is binary. You either have it or you don't. You have the software or you don't. Education is different, right? People gradually become more capable. They develop judgment, they experiment, they learn what works and they learn what doesn't work. They discover weird little use cases that nobody at headquarters would have predicted or thought of. And eventually the technology stops being something the AI team does and becomes part of how the organization operates. And that's when things get interesting. And that brings us actually to leadership. Because leaders don't necessarily need to become AI engineers, right? I don't I don't think every CEO needs to understand the Transformer architecture, for example. Frankly, most probably shouldn't spend their time there. That's not what they're good at. Leaders do need enough fluency to make good decisions around technology, but there's a difference between expertise and literacy, right? You don't need to be an accountant to understand your financial statements. You don't need to be a lawyer to understand when you need legal advice. You don't need to be an engineer to understand that a bridge shouldn't collapse, right? The same applies here. Leaders need to understand enough to ask better questions. What problems are we solving? How will we know if this actually works? What data will this require? Who's accountable if it fails? When it fails? What human capability are we augmenting? What human capability might we accidentally be removing? Can we leave this vendor or that vendor? What does this change about our customer experience? Right? What do we call the second-order effects? Right? Those are questions for leadership. They're not technical questions disguised as business questions, they're business questions created by technology. And I think the organizations that build that fluency throughout leadership will have a huge advantage in the coming years. There's another dimension to capability that I don't think gets enough attention. Culture. You see, because you can train people on AI all day, but if the culture punishes experimentation, they just simply won't experiment. Right? You could tell employees to innovate, but if every failed pilot becomes a career problem for someone, well guess what? Nobody in your organization is gonna take that risk. You can give everybody powerful tools, but if information is trapped between departments, the tool isn't gonna magically fix the organization, is what I'm saying. Right? You can automate a broken process, and congratulations, you now have a broken process that moves faster. And this is what I mean when I say technology is the easy part. The technology often reveals the organization to itself. AI doesn't just automate work, it exposes where work doesn't make sense. It exposes bad data, bad incentives, all right, duplicate processes, knowledge that's trapped in someone's head. Decisions that nobody understands. Policies that are created 12 years ago that everyone still follows today because, well, that's how we've always done it. And that can be uncomfortable. But it is useful. Maybe that's one of the most interesting opportunities here. AI could become a mirror, not just a tool, but a mirror that forces companies to ask. Why do we do this? Why does this approval require six people? Why does nobody know where this information lives? Why are there three departments paying for three different systems that all do basically the same thing? Why does this employee spend four hours every Friday copying numbers from one spreadsheet to another? Why does our customer have to tell us the same information four different times? Once the technology makes those questions obvious, well leadership now has a choice. Automate the mess or redesign the system. And those are very different outcomes. This is also why I'm skeptical of companies that are racing toward fully autonomous AI before they understand their own processes. Right? Everyone wants agents right now. Agents are exciting. An AI system that can reason through a goal, interact with tools, take actions, coordinate work, that's genuinely fascinating. There's enormous potential there. But autonomy magnifies whatever system you give it. If your process is good, automation can make it better. If your process is nonsense, Nonsense, well, you've now given nonsense a motor. That's probably not transformation, that's acceleration in a bad direction. And acceleration without direction is not strategy. So the sequence really matters, right? Before automation, understanding, before agents, process clarity, before scale, experimentation. And before dependency, capability. So before technology, ask this question. What are we actually trying to achieve? That sounds obvious, but I think it's one of the most radical questions you can ask in a technology conversation. Because the technology industry is built around possibility. Look what we can do, look what the model can generate, look what the robot can automate, look how fast this is. Look how cheap this is. And I love possibility. I live in possibility. That's one of the reasons I spend so much time around emerging technology. It's genuinely exciting. Possibility needs direction. Otherwise, you're just collecting capabilities that belong to other people. The vendor's model, the vendor's platform, the vendor's workflow, the vendor's roadmap. And somewhere along the way, you have to ask, what are we becoming good at? I think that's especially important because technology is going to keep changing. Whatever tool you're obsessed with today, there's a very good chance something better is gonna replace it tomorrow. Models are gonna improve, interfaces will change, companies will disappear sometimes, new companies will appear. The thing everyone insists is revolutionary this year may become a checkbox inside another product three years from now. That's normal. Technology moves. So if your strategy depends entirely on one tool remaining special, that's fragile. But if you've built people who know how to learn, teams that know how to experiment, leaders who can evaluate technology, systems that can adapt, an organization that understands its own problems deeply, well, then when the technology changes, you change with it. That's resilience. And maybe that's the real moat. People hear the word moat and usually think. IP, intellectual property, patents, network effects, data, brand, capital, distribution, and those things absolutely matter. But capability is different because capability compounds. Right? A team learns how to solve one problem, then the next one becomes that much easier. They build a mental model. They develop confidence. They recognize patterns. And they stop needing a consultant for every decision. They become better customers of technology, better partners, better builders, better skeptics. Right? They start asking smarter questions. And better questions create better decisions. That's hard to copy. You can copy software, you can copy features, you can copy messaging, and you can hire people away. But an organization that has spent years developing judgment, habits, relationships, a shared language, ways of solving a problem together, that is much harder to duplicate. And I think this applies far beyond companies. It actually applies to countries, schools, communities, and individuals. If AI gets cheaper and more capable every year, then access alone isn't enough. What can people actually do with that access? Can students learn with it without losing the ability to think? Can workers use it to become more valuable rather than simply faster? Can governments understand these systems well enough to regulate them intelligently? Can communities use technology without surrendering all of their data and agency? And can ordinary people like you and me distinguish machine-generated confidence from truth? Those are capabilities. And increasingly, I think capability may determine who benefits from this technological transition and who simply experiences it happening around them and to them. That's a much bigger question than software adoption. There's a phrase people use all the time: the future is coming. I I don't really like it. Because it makes the future sound like the weather, right? Like there's some storm system moving in from the West and all we can do is check the forecast. Technology doesn't just happen to us. People build it. Organizations deploy it. Governments regulate it. Consumers adopt it. Leaders decide where it belongs. Workers decide how they use it. Parents decide what they allow into their home. Teachers decide what belongs in the classroom. There is agency everywhere. Maybe not perfect agency, but power is uneven. Money matters. Infrastructure matters. But choice still exists. And capability determines how many choices you actually have. That's probably the deepest reason I care about this idea, right? Capability creates agency. If you understand the technology, you have choices. If you understand your own business, you have choices. If your people can adapt, you have choices. If you can move between platforms, you have choices. If you know what you value, you have choices. But if you don't understand the technology, if you've outsourced all the judgment, and if nobody inside the organization knows how anything actually works, well, then your menu gets smaller. You're not really deciding anymore, you're just selecting from whatever options somebody else gives you. And maybe that's the hidden relationship between capability and freedom. The more capable you are, the more futures remain available to you. So when somebody asks me what's the most important AI tool my company should invest in this year, I don't know. I need to know the company. I need to know the problem they're trying to solve. Maybe you need a frontier model, or maybe you need an open source one. Maybe you need some sort of automation. Or maybe you need a better data infrastructure. Maybe you need absolutely none of these things yet. Maybe what you need is three people who actually understand the process you're trying to automate in the first place. That's usually a less exciting answer, but it is a more useful one. And I suspect this idea is going to become more important as technology gets easier to use. Because here is the paradox. The easier the technology gets, the easier it becomes to mistake usage for competence. Anyone can now generate an image in Chat GPT. That doesn't make everyone an art director. Anyone can now generate code. That doesn't make everyone a software architect. Anyone can now generate a business plan, but that doesn't make everyone a strategist. And anyone now can ask AI a question, knowledge is free, but that doesn't mean they know whether the answer is good. Ease of use removes one barrier. It doesn't remove the need for judgment. If anything, it makes judgment that much more important. Because now everyone can produce something. So the scarce skill becomes knowing what deserves to be produced. Maybe that's what this next era of technology is really asking for us. Not that we become more machine-like, not that we race computers at the things computers do well. That's, I think, a losing game. I think the challenge is becoming more deliberate about the capabilities we want to cultivate in ourselves. Good judgment, curiosity, adaptability. Taste, empathy, systems thinking, the ability to understand another human being, whether that's your customer or your partner, right? The ability to recognize when something technically possible is also strategically stupid. That's capability too. And probably an underrated one at that. I don't want to romanticize humans here. Humans make terrible decisions all the time, right? Whether we like it or not, we're biased, we're emotional, we get tired, we misunderstand things, we we form committees. Machines can improve a lot of that, no doubt. I want them to. The point isn't to protect human work simply because humans currently do it and simply so that humans can have something to do. The point is to ask what we want humans to become capable of because these machines exist. That's a much more interesting question. AI can remove the work, okay. But what do we put in its place? AI can give us all the answers we want. Great. Do we get better at asking the right questions? AI can certainly increase speed. Wonderful. Do we get better at choosing the direction? AI can certainly give us leverage, but what are we choosing to leverage? And I think that's where the human part really begins. And also, I think that's what Beyond the Byte is really going to be about. Right? So I don't want to talk about product announcements and new models and you know, I want to get past the hype, past all the specs and and the the new technologies. I want to ask, what can the technology do? I also want to ask, what does this change? What does this make possible? What dependency does it create potentially? Who gains power? Who loses agency? And what do we need to become capable of in response? Because technology is going to keep changing. That's an easy prediction. I can give you that right now. But the harder question is whether we change with it deliberately or simply adapt to whatever arrives next. And I think that difference between reacting to technology and developing the capability to shape how we use it may be one of the defining differences between organizations, communities, and people who thrive. In this next era, and everybody else. So yeah, technology matters, obviously. But the tool, isn't the moat. The models aren't the moat. The software isn't the moat. What you become capable of doing with it, that's the moat. I'm Mo Hafez, and this has been Beyond the Byte. As always, stay curious, stay ethical, and lead human.