Most companies in 2026 think that AI is a run-of-the-mill technical problem to solve. They think if they just pick the right model, buy the right licenses, and get IT to roll it out, then the results will surely follow. So they spend 70% of their budget on tools, 20% on data plumbing, and maybe 10% on the actual humans who have to use this stuff every day. Then they're shocked when absolutely nothing changes. Here's the truth AI is not a technical problem, not even close. It's a change management problem, dressed up as a technical one. The Boston Consulting Group has been publishing the data for a while. They say that 70% of AI's core value comes from people and processes, not from the technical side or from the model in the context engineering. So the gap, it's not a tool gap, it's a people gap. And today I'm going to walk you through the five moves that the top 5% of companies are making to close that gap and to actually get measurable returns on AI. All right, let's start here. And that's what this Start Here series is. If you're new, my name's Jordan Wilson, and this is the Start Here series. But before we get into that, I want to tell you the big picture. Change management decides who wins with AI this year. Here's why. I think that at this point, it's no longer a model or the harnessing that a company puts around that model that's going to be the differentiator. Because if we're being honest, you know, maybe that might give you a couple of weeks uh head start on the others, but the gap is so small now. So, you know, there's a recent study from a writer that said 54% of C-suite executives say AI adoption is actually starting to tear their company apart because it is not in human nature to give up agency to an AI model. And that's exactly the big people problem that is causing this change management crisis that is AI. And most executives feel that they can see the personal AI wins, yet very few are seeing enterprise level ROI. And the gap between those individual wins and enterprise wins, it's not the technology. It's not, do we use GPT 5.4 or Opus 4.7? That's not what this is. It's redefining traditional management and people management. And the 5% who have already cracked it are running a deliberate five move change management playbook. So that's what we're going to be diving into on today's show. We're going to talk about that budget split that separates the 5% winners from everyone else this year. I'm going to tell you why upskilling is absolutely going to set your enterprise up for failure. So don't do the whole AI upskilling thing. I'm going to show you how Moderna, BBBA, and JP Morgan run the playbook that the 5% use every week and give you three moves that you can run this week without waiting for budget or approval. All right, let's get into it. Welcome to Everyday AI. This is the Start Here series. After literally 750 podcast episodes, I hear from new listeners all the time saying, Jordan, where do I start? And until the Start Here series, I didn't have an answer, but now I do. Well, you start here with the Start Here series. All right. Uh, now we're on volume 21 of the Start Here series. I think it's best when you go in order, but this is the essential podcast series to both learn the AI basics and to double down on your AI knowledge. So go to startheseries.com. That's going to give you free access to our exclusive inner circle community. Right now, there's no other way that you can get access, I think, aside from that. And in the Start Here Series space there that you'll have access to, you can go read, uh, write, uh, write about it, listen, right? We have an entire playlist of every single episode in order. So you don't got to go looking around. All right. If you miss our last volume, uh, a lot of people said this is one of their favorite episodes ever. So uh I guess it was a good one. So go listen to volume 20 of the Start Here series. It was episode 757. It was the seven silent sins of doing AI, right? How to spot and overcome the invisible AI work traps, which leads really well into this episode. This is the AI change management that works. So here's why this is such an issue. Because most companies assume that they could take their usual digital transformation playbook, which included the traditional change management piece, how they, you know, got teams to adapt to the internet, to adapt to the cloud, whatever it is, right? But change management, if you aren't familiar, right, is just the discipline of turning new tools into new behaviors at scale. And like I said, you know, usually there's the people, the process, and the technology. So change management really uh works with front two, the people in the processes. So, you know, um, why chain traditional change management doesn't work in an AI native workplace anymore is because all those other uh big shifts in how work worked, right? You could talk about ERP, CRM, cloud, mobile, whatever, uh, all of those things were additive, right? The core job, the core responsibility of the human stayed the same, right? The people in the process were relatively the same. It's just the technology slightly changed. But now the entire role of the people in the actual process is completely different because AI is not additive, whereas those other kind of shifts, they were additive. Uh, whereas AI, it changes the actual playbook. And I think, you know, this really goes to agency, uh, right. And human choice. Um, and I think this is probably maybe harder for people who are mid-career or have been out in the workforce for at least, you know, 10, 15 years. I think this is maybe that group that is hit hardest because they've probably been rewarded maybe multiple times with promotions, uh, new jobs, pay raises for your agency, for your ability as a human to synthesize and personalize information and to create new business value for your company, right? That's my definition of a knowledge worker. Um, and agency, that is those choices that we humans make and we leverage that domain experience. We leverage that subject matter expertise of the last five, 10, 15, 20, 25 years, right? And it's extremely difficult for successful humans to give away that agency. And that's why the people in the process side of this change management in the AI native workforce is absolutely bonkers because it's hard as a human being to say, hey, you know what? You know, someone, I don't know, in their in their 40s, in their 50s, of a VP, someone that has absolutely been through it, right? They they've been through the grinder, they've they've risen to the top. And then you're like, hey, now actually you're just gonna be orchestrating this agent now, right? You no longer have to lead this team of 50, right? Or maybe this is now a team of 20, and well, everyone's just using agents, and most of your job is gonna be feeding all of you know your team's internal IP uh into these uh reasoning models and building these agentic systems. And you're like, wait, I'm I'm having to give away all of my domain expertise to an AI, right? Because the job itself is changing. And most AI rollouts rollouts are quietly tearing companies apart because of that. So studies say that 97% of executives feel those personal AI wins, while only 29% see enterprise level ROI. And you think, why is that? Right, because internally, right, when one person sits down, they're still leveraging that domain expertise uh to work one-on-one, whether they're whatever they're using, right? Co-pilot, uh, using whatever it is. And for the most part, it's very easy for people to personally see those gains using AI. But it's usually when you are starting to work with others, because that's where you would normally uh kind of flex those domain expertise muscles, right? Or those domain expert muscles. It's when you're working with others because it's the natural human inclination uh to justify your position, to justify your title, right? And when that's gone um and it starts to uh, you know, AI starts to take over, that's when you get that gap, right? And that's when that's what leads to the gap between those individual wins and the enterprise wins because the people and the processes are completely different. Uh, and in that same writer study that I talked about earlier, and that's where we see that uh the company, 54% of C-suite executives saying that AI adoption is tearing the company apart because it completely shifts how the workplace is working in 2026 and beyond. So there are those that are getting it right, right? This is the the 5% here that I say are writing the change management playbook. And these aren't the ones that are picking the right tools, right? Again, this has nothing to do with the technical side. They're just rewriting and changing the change management playbook, right? Um, I literally think, does it does anyone remember? Uh, I don't know, maybe this was just me, maybe this is the the DARE program uh or or whatever from the from the 90s, right? So 90s babies. Do you like do you remember the you know someone coming and and ripping like a phone book? I think it was, you know, something to do with the drug education awareness program or whatever it was, right? Uh but I literally think that's what needs to happen in your organization. Think of that big fat 200-page uh manual, right, that your company has been running off of for the last five, 10, 15, 20 years. If you're still using some version of that or it's just been slightly updated, you gotta be that person that's just ripping that thing apart. So you need to do it this way. It is five steps, and we're gonna walk through them. One, fund, two, rebuild, three, unlearn, four, ritualize, and five, grade. All right. So you need to fund the 70%. That's the people in the processes. You need to rebuild AI native, you need to unlearn the old job, you need to ritualize being AI native weekly, and then you need to grade and reinforce the behavior. And each move, in order, compounds on the last. So yeah, you can't just skip somewhere in the middle or start at the end and work your way backwards. It doesn't work like that. So let's talk about move one. You need to fund the 70% that actually drives behavioral change. So again, this is from the Boston Consulting Group, right? They're uh 70, 20, 10, uh, but essentially saying that 70% of the value of AI comes from actually investing in the people and processes. Only 20% is the data, and 10% is the actual tooling or the AI algorithms that you're using, right? The model choice, you know, makes only a minimal impact on the overall value that organizations derive from AI. And most companies flip it the other way. Most companies are spending the majority of their time and attention on the tools, right? And don't get me wrong, obviously, as someone that follows this every single day for three plus plus years, I get that, right? And that's what I talk about here on this show. But I don't talk about it here on this show so that you can spend 70% of the time. No, the opposite. I talk about it here almost every single day. So you don't have to spend any of the time, right? The the point isn't for me to, you know, come on with our you know Wednesday deep dives and our Friday, you know, AI features, and then for you and your team to go talk about it for, you know, three, four, five hours. No, get your team around, listen to the podcast together, talk about it for 10 minutes, and that's it. Leave it there, right? And then start spending more of the time on the actual people in the processes. All right. Um, and then the five percent fund uh the the the five percent of organizations that fund change management like the main event are the ones that are winning, right? Those are the five percent that are spending the time, resources, uh, and money on the people in the processes, right? When was the last time? And I, if I'm being honest, I don't know if I've ever met a single company that has spent more on AI training, right? Than they are spending on the actual AI tool stack, right? Which is actually so freaking sad to think about. Because I mean, the AI tools for the most part are very cheap, right? Yeah, you might be paying uh, you know, $60 a month for uh, you know, an enterprise seat or you know, now since some recent price changing, you know, maybe there's some API usage on top of that. But let's say you're paying $100 a month for an employee, right? Which is probably less than that. You're probably more in the 30 to 50 range, but still on the high side, $100 a month for an employee to have access to a tool that is absolutely revolutionary. Let's just say for easy math, very easy math, right? Let's say that an employee makes uh $50 an hour. Easy, right? Uh we'll say mid mid-America, right? Someone middle of the uh of the corporate ladder. Um that's two hours, two hours of monthly training. No one's no one's doing that, right? It should outspend that. You should be probably investing at least a five to one ratio, whatever you're spending. Um, you know, different studies say different things, but you know, an easy rule of the thumb is five to 10x, right? So whatever you're spending on AI tooling across your organization, yeah, you should be 5xing or 10xing that um on your people in the processes. So on your people to train them, and then on also those people or maybe outside organizations to help you rebuild your processes to become AI native. I don't know if I've literally, and I've met hundreds of companies. I don't know if a single company's done that. Right? Obviously, their spend for AI is very high, right? Because they're building things on the AI side, uh, you know, um, implementing, you know, I don't know, open AI's models, Gemini's models, uh, anthropics models into their own products and services. So the AI spend goes high. So they're like, well, of course we can't spend, you know, 5x to 10x on people. Well, you absolutely can if you want to be in the 5% who are getting it right. That's how you get it done. All right. So sorry, what on a tangent there? So move one was fund the 70% that actually drives the behavioral change. Move number two is rebuild AI native from scratch. Don't sprinkle AI on top. I've talked about this so much before. A McKinsey study last year uh showed that workflow redesign had the biggest uh impact of the 25 attributes tested. So the biggest workflow redesign, right? That had the biggest impact. Notice, notice it wasn't using the best model or anything like that. When you're talking about earnings before interest and tax, when you're talking about the bottom line across the 25 different attributes that McKinsey looked at, blowing it up, right? Blowing up your SOPs is the thing that leads to success. So just bolting AI onto legacy standard operating procedures just makes broken processes run faster, not better, right? AI is not some magical band-aid that you write, you wouldn't put a band-aid on a bullet wound, right? It's it's not how you heal it. Um, right. So many people think that AI is going to fix broken processes. So they stick it on top, they stick it in the middle, they know that, oh, in this, you know, let's just say this eight-step workflow, steps three and four aren't the best. So let's see, you know, use Chat GPT for that. No, absolutely not. Because there's a good chance if you just uh break it down and rebuild it, that eight-step workflow becomes maybe a one or two-step workflow where the other good majority of it can be automated. And the five percent who get it right, rip up that SOP and rebuild it from the ground up, build it up modularly. That's the biggest thing, right? I think when we talk about building standard work processes, uh we build it with five years in mind, right? Which was the right thing to do in 1990s and early 2000s and 2015 and 2020. That was the right thing to do. You would build your processes to be as future-proof as possible. That was the goal, but that is a recipe for disaster. You have to build your AI native SOPs to be obsolete in a couple of months. You have to assume that however you're building out this process is going to feel very antiquated in a year, uh, or they're building it out for a future role that they're gonna hire and that person's gonna be in the position for five years and then it's never gonna change again, right? We're gonna talk a little bit uh more on the people side and job descriptions and things like that. And I think employees aren't gonna want to hear this, but it's the truth, all right. Uh, which leads us into well, steps moves three and four address this, but move three is unlearn the new job, not just the old tool. All right. I'm not gonna go on my normal um unlearn rant if you listen to the podcast at all. Uh, you know, I hate upskilling, reskilling is okay, right? But everyone talks about upskilling the same thing, how you can't apply AI on the top of an old process as an organization. You can't do the same thing to an employee that's been in the job for 10, 15, 20 years. You can't just expect to say, hey, let's take your current skill set and let's see in your individual skill set, in your individual work processes, where AI can make it better. No, you rebuild it for everyone uh from the ground up. Upskilling is legit dangerous. It assumes that your baseline expertise that you're building on top of won't be disrupted, but it 100% is going to be disrupted, right? If you're gonna build a new expensive roof, you'd probably want to check the foundation. Well, here's the reality in this uh alliteration and this metaphor, right? The foundation is gonna change, right, every year. Um, it's gonna be wiped out. So you have to understand that your foundational baseline skills are also going to be wiped out if you haven't come to that realization already. So upskilling on top of your current skill set is a useless endeavor, right? Uh there was a Harvard Business School study that said that saw that every employee or said that every employee needs at least 30, a 30% digital in AI mindset. In other words, what that study means is that about a third of your work should be augmented. A third of your work should be interacting and talking with AI, right? To me, that's like, oh gosh, that's a lot of wasted time, right? I'm 95% of my time um is is chatting with AI. But uh that's I think that's an easy baseline to follow, right? And I'm loosely interpreting uh what they mean there by a digital and AI mindset, but that's essentially what it means. If right now you're not spending a third of your time uh augmented by AI, right? That doesn't just mean chatting with chatbots, but if a third of your time isn't uh being augmented with or collaborating with AI, you're gonna fall behind. And then those 5% that are getting it right, uh, they let the large hangout, they let the large language models hold the domain expertise and they practice the new identity weekly. All right, move four, ritualize the weekly enablement instead of the quarterly training. All right, here's what that means. A Gallup uh study said that employees with manager support are more than nine times likely to say that AI transformed their work. Okay. I look at that study uh two different ways. Number one, I'm like, well, of course they're gonna be more successful uh with their AI enablement and AI enablement if they number one. One, aren't using shadow AI, right? Which so many companies are. So when you have your managers uh explicit um support and training and resources, of course, um they're more likely to say that AI transformed the work. But talk about more than nine times, that's a huge jump. That's not a small, like, yes, you know, I'm more likely to say, you know, twice as likely to say AI transform my work. No, when you have buy-in from management from all layers and all levels, uh, that's when you can actually get to the new AI transformation. Um, but here's here's what I want to start to differentiate here. I think a lot of organizations are still looking at AI as a marathon. And it is technically a marathon, but it's also a series of sprints. Um, so this isn't something that you can look at over the year, right? I think most organizations are past the concept of yearly pilots. My gosh, if you're still thinking yearly AI pilots in 2026, you're smoked. I don't care how big you are. Um, because yeah, that's that's that's how fast the space is moving. And if you are sitting there arguing with me, just just wait. Uh, you know, I'll say 90, 90% of companies will fall on um on the brunt side of that, so to speak, right? If you're still doing year-long AI pilots, it's gonna be bad. You know, quarterly kickoffs are are okay, you know, annual seminars, you know, those can build new habit habits, but you need weekly rituals, right? Here's what I mean by that. The best and most successful AI teams that I've talked to are those, and a lot of them pick either Mondays or Fridays, which I think is really cool, right? It's it's kind of anchoring, anchoring your week um in AI. Um, and I've even talked to some organizations that do a Monday and a Friday, right? It's opening up the week and closing the week, which I think is smart as well. Um, but that's where you come, you share what's working, what's not, challenges for next week, right? Here's our our our transparency report, our observability reports, our challenges, uh, what we're scoping next, our use cases. Here's the new, you know, in in in these systems, right? We use Gemini for this and we use ChatGPT for this. Here's what changed in those models and how it impacts our work, right? You need to be having those conversations at least once a week, right? As a team, um, as a department, as an entire organization. And if you're not, you need to start that, all right. And then here we go, move five. You need to grade the behavior change, not the license, logins, or seed counts, right? So many organizations just right, it's it's like, how are you measuring AI? Right. And it's like, okay, utilization rates, right? Which is easy, right? It's quantifiable. So I get it, I understand, right? But in those instances, sometimes I ask this, sometimes I don't, because I don't always want to ruffle feathers, right? Especially for you know, clients that hire us with AI implementation, but maybe I should start doing this, just be a little more uh, you know, brash and say, hey, it's not worth working with your organization or even trying to implement AI until you absolutely rewrite every single job description, right? And a lot of people will think, oh my gosh, that'll take years. Well, no, it doesn't. With AI, it should take not very long at all. Take your old job description, you know, have a conversation with your direct reports, you know, HR, whoever, talk over the job description, record it, say, here's what it looks like with AI, you know, put that into Claude, ChatGPT, Gemini or whatever, spit it out, get it approved, make sure that they're uh consistent across the entire organization, you know, team first, done. It's not a difficult thing, but it's the expectations and and in step five is is the hardest, right? Because behavioral, behavior change is the root of change management. And that's why this thing is a big problem. Because what employee wants to sit down and say, uh, yeah, I haven't had a new job description in eight years, right? I don't want to come, right? I don't want to sit down and rework how my job works because hey, the reality is I've talked about this a lot of times. You want to talk about the number one reason that your company is not getting ROI on generative AI. Well, it's because you don't know how to measure it. And there's a good chance that your employees are either using it or they're pocketing uh that time savings from using AI, right? I said this example before. I know many smart people that have essentially automated at least 80 to 90 percent of their job. They're working remotely, uh, they're you know hybrid, and yeah, they just pocket that time, spending a lot more time on the golf course or doing things around the house, right? Um that is the problem. That is one of the problems, but it's also not that employee's fault. It's your fault, manager, it's your fault, CEO, it's your fault, department head, because they're still working off an antiquated job description, right? So, yeah, we always talk about, you know, grading behavior changes. Well, how often are people using large language models? And that's not it. I think you have to first redefine your actual job descriptions, which means redefining what your department is probably working on and redefining the expectations and the expected outputs. That's a big one. All right, so look at the tech giants. That's what they're doing now, right? For better or worse, whether you agree with it or not. And if you're a listener at one of those uh, you know, organizations, maybe you're not enjoying this. And well, that's part of change management, right? So Microsoft, Meta, Google, and Amazon are all now grading AI usage directly in performance reviews, all right. So the 5% that are getting it right, they're not just grading prompts per employee and but they're looking at session depth and shipped workflow redesigns quarterly, right? They're looking, are you changing how you're actually working? Not just are you saving time, are you producing more? Are you creating more value, but are you constantly changing how you or your team works? All right. So here's what I want you to focus on. Uh actually, let me go over some quick stats here. I wasn't gonna go over these, but uh, you know, maybe they're important enough. So uh a couple of use cases that kind of prove the five move playbook works. So Moderna, they hit 80% internal MChat adaption for their internal tool, and they run a 2,000-person weekly AI forum. Uh BBVA trained 250 senior leaders first, then scaled to 83% weekly active AI usage bank wide in JP Morgan, held headcount flat while reshaping roles. Their operation shrank and client-facing roles grew. So essentially, they were able to operationalize with AI uh some of the uh more manual knowledge work that is done internally, kept headcount flat, and then sent their people out in the world more for more client client-facing interactions. All right, so here's what I want you to do. Now that you know the five steps, and let me just re-say them. Move one is to fund the 70% that actually drives behavior change. Move two is rebuild AI native from scratch. Don't just bolt it on the top. Move three is unloan, unlearn the old job, not just the old tool. Move four is to ritualize the weekly enablement instead of quarterly training. And move five is to grade the behavior change, not license logins or seed counts. So now that you know this, here's your three moves I want you to do, right? Whether this Friday, this Monday, doesn't matter. So pick one old SOP, the way that you used to do it, and ask what it looks like to be rebuilt AI native from scratch. Get everyone together, see who's using what tool, what success are they finding, what is an old dumb SOP? Here's the thing probably most people hate it. Rebuild it from scratch. All right. Uh step two, find someone who's already right, identifying your AI champions who's finding that ROI individually, and then find that old SOP and have them tear it down and rebuild it, right? Tear it down, rebuild. All right, so unlearn, relearn, train, repeat. That is step two. And that is really rebuilding the model of digital transformation. And then step three gets into that change management. And maybe I'll end on a different note here for step three, because giving, if you do step one and step two, that's gonna ruffle feathers, right? Maybe not so much for those employees in their 20s that are maybe a little more agile uh or maybe a little more AI native. I don't know. But for everyone else, going through this process is tough because what it might look like is you know, Bill over there in IT has hung his career on a certain skill set, a certain domain expertise that now you may give to a large language model. And Bill is going to feel entirely threatened, and Bill is probably not going to feel good about this. That's the reality. This isn't easy, right? People think it's it's all fun and games, but a lot of people, which we talked about in our last episode in these seven silent sins, a lot of people hang their identity, not just their corporate identity, but even their personal identity on what they do in work. And if you all of a sudden start handing that off to agents, which y'all, it is the right thing to do. It's not easy, right? It's like I had to go through a certain grieving process many years ago when I realized that large language models with proper prompting were better writers than me, right? I previously won ACP Story of the Year. I was a journalist for a long time, I was a Pulitzer Fellow, all those things. And when I realized that AI was a better writer than me, it's hard, right? You're like, oh frick, I've been getting paid to do this thing for, you know, a decade and a half. And well, now what? Right? That uncomfortable period is where your future strength grows in an AI native workplace. But step three is important because it involves what comes after giving away that human agency. And that's where the next phase of change management, which is people management, comes into play. In your organization, you need to employ human support systems via human resources and management for what that looks like, right? How do you make sure that Bill, after he gives away the the the true skill set, right, that he's hung his career on for 20 years? Yes, he's probably going to be managing an agent, but what do you do with his time then? What do you do with his extra time? Right. Yes, you can, you know, he can become part of your AI champion team and he can go help build, you know, new SOPs, new ways of work, create new lines of revenue, but it's easier said than done, right? When you rip up the job descriptions, when you rip up the old SOPs, and when you deploy agencai AI that works, it's not always easy. So step three is you have to employ that human support system. All right. I hope this one was helpful as we tackled AI management that works, AI change management that works, and the five moves the five percent make. All right, do me a favor. Number one, go to starthireseries.com. All right, we're gonna be recapping today's episode in our newsletter, but in the Start Here Series space in our inner circle community, you can go listen to, read about, and collaborate and network with others who are going through this Start Here Series journey together. So uh there's a playlist inside there um in our uh on a Spotify playlist with every single episode. So it's a little easier to track. All right. So number one, do that. If you're listening on the podcast, appreciate your support. Please make sure you subscribe to the Everyday AI podcast. Leave us a rating if you can. So that's a wrap for today. Thank you for tuning in. Hope to see you back tomorrow and every day for more everyday AI. Thanks, y'all.