Most enterprise AI failures are change-management failures, not technology failures. That is the answer Rob Lion, an organizational strategist, gave on The Artificial Intelligence Podcast: organizations buy a tool and skip the part where people are integrated into using it, so the rollout performs nothing like what was promised.
The pattern starts before the purchase. Hype and doom-and-gloom headlines about AI eliminating jobs created a mandate, and executives now feel they have to justify AI use against their boards, which produces a lot of posturing. Then the tool arrives and the promise of the commercials, that what you ask for is what you get, does not match reality. Lion's point is that most organizations are not large. Many are bootstrapping, and a directive to adopt AI because peers in your social circle are doing it is problematic when nobody has integrated people in a thoughtful, methodical, systematic process. That is true of any tech transfer, and it is especially true of AI.
Why do employees quietly resist using AI?
Because using the tool well can look like training your own replacement. What employees say, almost universally, is some version of: make my job easier, but not so easy that I am replaceable. Lion says this is the most common thing he hears, and that it now runs top down as well, with leaders asking not to be made replaceable either.
The news cycle feeds it. Companies announced they had replaced staff with AI, and Lion's reading is that most of those firms had overhired and were using AI as cover. It still scares everyone. If using the tool looks like a route to being cut, the rational move is to slow it down. The host's point is that when you motivate someone to make a project fail, they will make it fail. Add the other common sight: the person telling everyone else to use AI is often the one person not using it.
Why does measuring AI usage backfire?
Because counting how many documents someone generated measures use, not success. Mandate a number of tasks completed with the tool and people chase volume instead of quality, because volume is what you are rewarding.
Lion's comparison is the first hire. You would not tell a new remote employee to "manage my social media, go" without a posting cadence, a frequency, a style and a definition of what success looks like. Hand a tool to an employee with no defined process and no measure of success and it fails right out of the gate. If someone does not know what a good output looks like, they sit in a constant state of stress. The same applies to what you ask the tool to do: if you cannot say what a good presentation looks like, how many slides it needs and how it gets made, you have given the work no target.
What does a good AI rollout look like?
Start with a task force rather than a top-down bestowal. Lion recommends a mixed group drawn from different levels: early adopters, and probably one slow adopter who is not resistant, just not skilled up yet. That voice is more useful than someone fighting the change.
Then field test. Once you land on a product selection, give people three months to touch it, with a couple of hours a week of their time allocated to testing it. Ask whether outputs stay on brand and on feel given the context programmed in on the front end. Sanctioned experimenting beats secret experimenting.
Lion's three C's are clarity, context and consistency. They build better prompts and better agents, and they produce better outcomes for employees. Expect it to be slower first. There is a learning curve, an adoption curve, and then muscle memory. Go in expecting speed on day one and you will quit after two days. When people hit friction, they bypass the tool and do it themselves, which is how faked AI succeeds in the short term.
What to do before you roll out another AI tool
Define the problem before you buy. Work problem to solution, not solution to problem. If you do not have a clearly defined process for the tool to automate, accelerate or replicate, it will fail right out of the gate.
Build a task force with mixed representation from different levels, including one person who is willing but not yet skilled up.
Protect a sanctioned test period of about three months, with a couple of hours a week carved out for people to use the tool on real work.
Codify the process and the measure of success. If someone shows you an output, you need the prompt, the recipe, not just the cake. What does a good presentation look like, and how was it made?
Create a standing place to share. A monthly brown bag or lunch and learn where people bring what worked, what did not, and the hacks. Lion's fun factor applies here: if it is not fun, why spend time on it.
Invite dissent and feedback early and frequently, including feedback that pushes against the direction you want to go.
What about security and shadow AI?
People are already using AI quietly inside company systems, and garbage in equals garbage out. Lion flags the risk plainly: if you do not know which tools people are using, you do not know what direction the quality of your outputs is heading, and you cannot learn from them. The exposure can be small and accidental, such as photographing a screen with a phone to ask a question, and different versions of the same tool carry different data retention settings. There is no undo.
The fix he offers is cultural, not technical: talk about it, bring attention to it, and make people feel safe discussing their tool use instead of driving it underground. He also names a problem the episode leaves open. There is not really a definition of what counts as proprietary and where the line sits, and neither guest nor host supplied a policy template for drawing it.
What holds across all of it is that expertise still carries the work. Lion's view is that you still have to be smarter than the machine, and that for an expert in their own field the tool becomes a thought partner rather than a replacement for judgment.
Original show notes
Everyone was told to adopt AI, tried it, got burned — so why do enterprise AI rollouts keep failing? Jonathan Green talks with organizational strategist Rob Lion about the real reasons big AI initiatives miss: unrealistic expectations set by the hype, skipping change management, and the employee mindset of "make my job easier, but not so easy that I'm replaceable."
Key Takeaways: • Most enterprise AI failures are change-management failures, not tech failures • The "promise of the commercials" sets expectations the rollout can't meet • If you skip how you roll it out, you bypass the most common path to success • Employees quietly resist: "make my job easier — but not so easy I'm replaceable" • Teach people the why and the workflow, not just "what's an LLM"
Notable Quotes: "Make my job easier — but not so easy that I'm replaceable." — Rob Lion "When someone doesn't know what direction they're heading, the rollout's already in trouble." — Rob Lion
Connect with Rob Lion: LinkedIn: https://www.linkedin.com/in/robertlion Website: https://blackriverpm.com
Enjoyed this? Follow The Artificial Intelligence Podcast and share it with a leader rolling out AI.
Auto-generated transcript, 5,934 words. Timestamps link to the moment in the episode.
Why do AI rollouts keep failing at the enterprise level? We're gonna find out with today's amazing special guest, Rob Lion. Now, Rob, the biggest issue we're seeing right now is that everyone has kind of touched the stove and their fingers have been burned once. Everyone was told to adopt AI. They tried to do it and it didn't work. And now they're like, well, what are we everyone else says it's working for them. It's almost become like the Emperor's new clothes where everyone else says they
can see the clothes. And you're starting to wonder if everyone is kind of faking it till they make it and it's not working for anyone. So like, is it why is that happening? Let's just start right there. ~ let's let's yeah, at the very beginning, right? Let's think about what some of the news has been in the last couple of months about scale backs, cutbacks, ~ you know, block replacing employees with AI and you know, these house cleaning things. And and and
and during that time over the last couple of months, it was really put out there. This is what's happening. It's gonna eliminate these jobs, it's gonna hammer the and the new employee market as well as white collar market. And And I was just waiting for this to kind of catch up for people to say, no, maybe this is a reflection of overhiring, right sizing, various other factors, right? So what we're seeing, and I'm using that as an example, is that it was such doom and gloom.
It's like everyone's using AI. Why aren't we using AI? We should be using AI, right? CEOs and executives feel like there's a mandate to to justify AI use in light of their board's expectations. So there's a there's a lot of you know, posturing going on. It's nice that it's starting to unravel a little bit where people are starting to acknowledge, well, there's a bunch of different things going on, right? So the tools consistently get better week after week. They're the quality of the tools.
but also that people are saying, well, we're not in it as deep as we thought we were, or we don't have the resources resource at these levels, right? There's n there's no doubt that there's there's pe big players like KPMGM groups out there that are really trying to position this. really help large organizations. But most organizations aren't large, right? Most organizations are in many instances bootstrapping. And so these directives that we need to use AI because that's what I hear other people in my social circles
are doing or what we're reading on the headlines is is problematic because we're not integrating people in a in a thoughtful, methodical, systematic process and whenever that happens, whether it's tech transfer or anything else, even a new hire, if we don't integrate the system the process well, it's not going to perform according to what we expect. And that is especially true with AI and the tools out there and and and how might people use them, right? Like most people are using them to
generate pictures and help themselves look better in on a personal level, right? Like it's still very much a toy. Yeah ~ but at the ~ the organization or systems level there's a lot of discrepancy there. I find that there's two ways that AI can implement it. The top down is like the good idea fairy comes for a visit and says we bought this AI tool. Now you guys either have to figure out how to use it or use it in this way and we're gonna
measure how many tasks you do with it. And the measurements are always very silly to me. So like I'm even Like I don't even use time tracking for my employees. I say, tell me what you can get done this week. If it feels like enough on Monday, then on Friday, tell me if you got it all done. And then you get paid because I would rather reward someone who could do Yeah. Yeah. it twice as fast than someone who can slow roll.
And I feel like a lot of the way we handle employees has kind of come from back when time was the most valuable Like when you need someone in the room, then time makes sense, right? When they're physically there, when they're managing a store. But for task based employees, and now they're saying, you have to get this many more tasks done using AI, or so they use some type of measurement that's not a measurement of success. Like how many documents did you generate with AI?
Well like that's then they're gonna chase chase volume, not quality, because that's what you're rewarding. And I think that's one of the problems. And I also see this That the person who tells everyone else to use AI is the one person not using it. And there's a lot of this, yeah. Yeah. There's a lot of this language which is we want to replace you with AI or train your replacement, or we fired everyone for AI. We're seeing that in the news.
And most of those are companies who just overhired over the past fifty five or six years and they're just using AI as a cover, but it still scares everyone. And I think that's what causes employees to go like if I use this, I'm out of here. So the longer I can slow roll it, the longer I have my job. So they're motivated. Whenever you motivate someone to make a project fail, they're gonna make it fail. Yeah. So what you're talking about is several different things here,
but one of which is what's the element of psychological safety in this organization, right? So you have this mandate and we're buying product hex, we want to know how you use it, what's the frequency, and and it it that's a very general concept, but is there a lot of conversation going on behind the scenes about how we're gonna roll this out, how we're gonna integrate this, what does success look like, right? Who are who wants to be the guinea pigs that actually
that that are the test subjects of this tool, right? Because let's be clear, like we have people on our teams that are definitely on front end early adoption, geeked out about it. Seems like fun. I want to be more effective, efficient, have greater impact, generate higher quality output. But then there's people that are late adopters that are like, I'm not interested in this to your point, Jonathan, like it might it's gonna I feel like it threatens my job or I
I don't want to be caught off guard. I think one of the key pieces about those mandates though that we kind of glossed over is what is the expectation as it relates to that and and how mandates work as long as we have clarification and alignment throughout the system, right? So that all the leaders all the way down are really telling the same story and what this is about. But but if if if if there's not a story that's being told,
if it's not clear communication and we're just buying a tool, a product, and people are just expecting to complete things then they're left to their own devices to kind of worry, stress out, be uncertain, things like that. And so that's one of the important features of of the of what we do in organizations is help people gain alignment. So the mandate could come in absolutely. Yeah. And it's especially it's even stronger when your people in your executive offices have actually vested interest in the tools,
in their impact, at least even just an appreciation of how they can make a difference. But If that stuff's not happening, we're we're setting people up not to be successful. Yeah, I feel like the biggest problem is the measurement of success, which you brought up, which is so important because I see a lot of either I mentioned earlier, like the measurement of success is how much you use it or how much time is spent using it or how many things are made using it.
And that's not a long term that's a measurement of use, not a measurement of success. And I think that's where a lot of these programs fail. Like, we want everyone using this tool and well, why? And it's very hard. Which I think you dialed into to change human behavior to get people to switch softwares. There's a learning curve, Mm-hmm. there's an adoption curve, and then there's like building the muscle memory. So they're so used to doing it one or another using it to a different way.
And if there's not a positive reason to do it, because it's going to be slower for a while, I think this is the other big expectation people have is that it'll be faster day one. It's like, no, it takes time to learn any tool, including AI. So what is the right way to? start to launch an AI campaign or to launch an AI implementation that minimizes the chances of failure, that kind of bypasses the most common mistakes. Like what are those most common mistakes in addition,
any that I haven't brought up? And then what's the right way to approach this for a company that doesn't want to have another failed pilot? Yeah, I I I think a task force approach is is helpful. as you mentioned, the top down bestowed upon people creates stress and it stresses the system. However, if we could bring a task force together with a mixed representation from some players at different levels, probably early adopters, e people, maybe one kind of slow adopter, not that they're resistant, but they're
like, Look, I I know we need to go this way or we have to be receptive to this. But I I'm not skilled up yet. Like I think that's a great voice to have as opposed to someone that's that's pushing back and fighting that. And then through that process, we determine what the tools are, right? Even just a year ago, if we were looking at slide decks, right? If you think about the different tools up there, and over the course of two weeks,
the quality of the builds of what these different slide deck companies have made ~ were substantial. And timing has a lot to do with this. And you want that task force to touch these things before they're rolled out because early early adoption in AI, let's stick with slide decks, people purchase the subscriptions or the tokens or whatever you want to call it, and and then they gave it to the employees saying, Hey, this worked well for me. Well,
in the build process of the the AI tool itself, it you know, it peaks and valleys like in terms of its success in terms of compute. And now we're getting a lot more stability, but just think about how stressful that is when you're like, here, here's a tool, we want to run with it, and you drop it on a team and you've bought into it without having a little test period to get people on board and and verify, yeah, this is what we like about it,
or here are a couple of different options that seem to do what we're looking for here. ~ Getting people involved has always been essential for all organizations to create buy in and investment. And and that's still true. today with AI tools and initiatives. I kind of think of AI tools just like the same mistakes. We're making the same mistakes here that a lot of companies make with their first hire. Like the first time someone hires a VA or a remote employee,
they go, manage my social media. Go. And it's like, well, what's your posting cadence? What's your frequency? What's your style? Exactly. What's your measurement of success? And if someone doesn't know what success looks like or what target, they don't know if they're doing a good or bad job. They're in a constant state of stress. And the same thing for a tool, if You don't have a clearly defined process for it to automate, accelerate, replicate, then it's gonna fail. Guaranteed. Like right out the gate.
And then if you don't have a measure of success, which is, well, what does a good presentation look like? What's your measurement of success for a presentation? And what's your process? This is the most important part. How did you make that presentation? There's nothing worse than when someone shows me an output and then won't give me the prompt they used. And it's like, I look the cake, like if you give me a cake and say, mate, here's the ingredients, make it.
I'm like, there's I definitely need that recipe. Like the in-between is super important for me. Yeah, I Like if you're a master chef, that all I don't know how someone can taste something and they know how it's made. To me, that's like magic. I can't understand that. But it's the same thing to try and reverse engineer a process because, like you said, these AI tools are constantly changing and the order in which you do things makes a huge difference. And there's always this micro step that
gets lost unless you watch someone doing which is why we all love watching a video instead of a set of instructions, because the one person the person would have forgot write down is like the most important step. So, in the same way, if you don't have your process of how you use this tool codified, you don't have the measurement of success codified. It's impossible that you even consider success. Like if you gave the same set of instructions to a new employee,
would you expect to be able to succeed when you go make a bunch of slides? What should they look like? I don't know, good? Like how many should I make? The right number. Like there are not enough instructions there for them to And it's the same thing when you're giving a new tool to an employee. I also think that like it's not clear Why? Often the why we're switching to this tool or what's the benefit to the employee is ever explained.
'Cause I've seen some companies that go, Now that we have AI, we want you to do twice as much work and the people like, What? Like that's the wrong that's like the worst way to motivate someone. Exactly. Yeah. Yeah. And and then you think about it and you think about how many poor prompts you might have built over time that have led you into this rabbit hole trying to build a project or something, only to learn that look, you're expect
you're expecting me to do double time in terms of work quality and output. I'm stuck in this system talking to me in a way that I didn't you know didn't really You know, th there's a time that the the the the the feedback mechanisms from the various tools now are are are so much better. But you would get you stuck in these rabbit holes where they're rewriting the script based on more recent information that's provided as context, and then all of a sudden you're you're off
point. I I I love the example you brought up w about the onboarding piece. One of the things we have to remember is that when someone that's the expert is introducing something, they often have an expert lens on. And so it's like me going, hey Jonathan, I need you to do X, Y, and Z to your point. And you're you're new and you're like, Yeah, I want to do X, Y, and Z. I'm eager. I want to do a good job,
but you don't know what's going on in my mind in terms of like you mentioned the the the what's the what's the slide count here? Well, just enough. Well, what does that mean based on what? Right. So you're you're right off there. And what I love about this example is that this is this illuminates the power of AI tools, that the same three things are true both with your onboarding example as well as AI build and integration pieces. It's clarity, right? Context and consistency.
And those are those are three C's that are that are very true that help you build better agents, tools, prompts, but also that that lead to higher quality outcomes, but are also true to higher quality outcome experiences for employees. Yeah, one of the challenges I have when I'm trying to train someone new is that I have so much history. And if you think about the way AIs behave, it's weird and it's constantly changing because they push updates. AI is the only space I've ever run into
where software will update but not update the version number. So you're using a new version of ChatGPT and you don't know it. They only update the version number like once or twice a year, but the actual model might update five or six times a day. And I've seen dramatic updates that are non numbered, so Something that worked yesterday might not work today. And that's very stressful for a user. That's I think it's horrible. I think they should version everything.
Every other software in the world does it. And that provides a challenge. And then there's like talking to ChatGPT versus Cloud to Gemini, all the different tools, they're slightly different. Like ChatGPT calls it a GPT, Cloud calls it a skill. And even difference in Cloud Desktop and Cloud Code and Cloud in the browser all operate differently with how they store they go, it's not a it's not an MCP. Now it's called a connector. And it's like what? Why you invented the term MCP,
what you're literally coming invented, and you change the name, that's not cool. So there's all these little things that come through. And I think that the training period, which is so important, which is like with most tools, right? They give you a copy and paste this text into there, see what happens, and copy and paste this and see what happens. And you kind of learn by doing the tutorial phase is so missing from most of these tools because. The promise of the commercials,
like what you ask for is what you're gonna get is such a disparity from reality. And I think that you're you've dialed into executive and then what happens when someone has a bad experience, they go, you know what, it's faster, just do it on my own. The trying to force this tool to do what I want when I already know how to do it. And if you don't go in with the expectation that it will be slower for three months or whatever that learning curve is,
then you're gonna give up after two days. And I think that's one of the biggest problems, besides the over the expectations too early, is that someone thinks, ~ I'm not getting the result I want. Either I'm not good at this, I'm dumb, or the software doesn't meet its promise. So I'm just gonna bypass it. That's why that company that like hired 50 VA or like 5000 VAs and pretended to have an AI succeeded. Because it's easy to fake AI and it's easy to bypass it.
It's an easy system to trick. And so they go, this is too hard. Let me find a way around it. And in the short term, like happened to them, you can get away with it for a while, but not eventually they keep saying, do it faster and faster and faster. And eventually you can't, because you're still just one person doing one thing. You can't do simultaneous tasks. Certainly, absolutely. yeah, it is this all makes a lot of sense. you mentioned something there about the
training and and just that integration period, right? So once again, let's come back to a mandate. And so if we're talking about really a a really strong way to roll out an initiative, it comes starts with the mandate and then we focus group it or task group it and ~ task force it. And then instead of once we land on a final product selection, we say, look, over the next three months, I want all of you to touch this and and
play with it and test it. And and on top of that, I'll give you a couple hours each week of your time allocated to exactly field testing this to see what works, to see if it's is this remaining on brand for us? Is this remaining on on feel because of the context we programmed into this on the front end? all these things so that people start to become more comfortable using it in a sanctioned type of approach, if that makes sense, as opposed to a
secret type of approach, right? Whereas you know we have people that are using AI right now. Like what's one of our biggest risks right now is that we have people quietly using AI in our systems and garbage in equals garbage out. And we're at risk because we're not talking about these things And bringing attention to these things to to focus on, like, hey, this is a real thing happening, and we want you to use these tools. I want you to feel safe talking about them
because we want to learn from you. But the the risk of people using different programs, different like you're talking about Gemini versus Cloud versus Chad, they all have their their issues. And and by us kind of having people doing this secretly, we don't know in what direction we're heading as it relates to the quality of the outputs, right? Because you you still have to be smarter than the machine. Like you still have to have that human knowledge capacity, expertise and insights. Like I find it
very powerful for me as an expert in my field to use it because it becomes a thought partner. But I also see what it gives me in terms of content. And and we have some interesting conversations through through those kind of pulling the th pieces apart because it it exists in categories that it shouldn't exist in that time. So it's looking at the wrong type of content based off of wrong parameters. So ~ integrating this and creating that space for people to play.
feel like they're having some wins and then sharing with others what are those Like you mentioned, what's the prompt that you used that gave you this? And and so why am I not getting the same outcome on the exact same prompts? There's a little bit more background that goes into that. So I I think you're you're right on there that that there's a methodical process so that we roll this out safely and effectively. And this doesn't even get into ~ proprietary information, right?
We're we're not even talking about that yet. Like think about companies, either proprietary information or healthcare information, all these other kind of accredited types of ~ sensitive information that we want our people to be using in accordance with the expectations according to the countries we're living Yeah, I think that we have this challenge where if people like we don't really cover security very well. And because we're used it used to be it would take a long time for someone to make a mistake
and the data gets lost. Now you can just hit enter on the one thing, and it could be as simple as someone picking up their phone. taking a picture of the computer screen to ask the AI a question and now the data's out there and they don't realize that this version of ChatGPT looks the same, but this one doesn't have the no data retained or don't train on our data steps. So now the data's out there could be something that simple, they don't realize it. And
There's not really a definition of what's proprietary and what's not. Where's the line? And because we don't cover those things, because it's like, we're a small company or we don't have time or we don't have a size, so let alone a fractional size, and we don't have anyone in charge of security. And those meetings are boring anyway. So we kind of skip over that stuff. And it can be that it's that quick. And there's no undo. It's just like ~ sending an email.
Like I think sending an email should always take an hour. So you click send and there's an hour for you to go, wait, right? I don't know why we got rid of that. Exactly. Like it's kind of like, maybe you're as old as me. You used to be able to you call someone, leave a voicemail, go, you sure you want to leave or you want to delete? And you always hit delete. Delete. Yeah. Plus yeah. Yeah yeah. Let me try again. You know, it's like, Yeah.
that was horrible. That was a horrible mistake. Delete that one. Like, give me another chance. So that's missing that immediacy. There's also this, like, how can a company implement an AI policy in a way that lets the employees know the AI tools are here to enhance you, not replace you? Because even if you say that. They don't believe you. Cause like nothing says you're about to get fired, like your boss going, I'm not going to fire you, but it's like, what? Yeah, yeah, exactly. Yeah.
Yeah. It comes back to that psychological safety, right? So i if your people didn't feel supported, they didn't feel like they existed in an environment where they could make mistakes and share, right? Like as a a good leader will say, make mistakes, just make small mistakes and and let's learn from them and move on. And then if you happen to make a big mistake, let's talk about it because we need to figure out what we need to do to kind of triage it, right?
and so I think that's that's that's essentially the part of the process here is that we have to have people feel welcome to explore. And one of the ways I like to celebrate this is with these knowledge groups or whether it's a lunch and learn or something that we're adopting a new procedure, technology, what have you, want you to touch it. You have that time to touch it. We're giving you some release time to touch it. And then, you know, once a month we're
gonna have a a brown bag luncheon or something where we get to talk about what worked well, what didn't work well, things that we've learned, kind of hacks, right? Use that language that's kind of disappeared on how do we get this thing to respond better for what it is we wanted to do. And and and that's that's how we create integration, right? That's how we create alignment. And but if people don't feel trusted or valued, they feel at risk. Right. Or if they're feeling like they're
being imposed on, which violates a level of intrinsic motivation, then it will become more difficult. So there's a level of I'd say leadership sophistication here in terms of how we deliver the messaging, right? And how we follow up on it so people understand that this is this is a good space that we're in right now. Yeah, I think you've dialed into something that's really important that I see missing from most of the people I talk to, most of the campaigns I come in on.
The most common thing that I hear, like the CEO or the leader who hires me or brings me in, says something along the lines of we want to implement this AI policy. This is the tool we bought. Get everyone to use it. And then I start telling employees, the first thing they say is like, Hey, make my job easier, but not so easy that I'm replaceable. Like everyone says something along the lines of don't do so much that I lose my And
Everyone says it. It's the most common thing. Every employee says it. And I know more and more lead and the CEOs are starting to say it too. They're being listen, don't do too much. Like do enough that we're doing some AI, but not so much that I'm replaceable. And I think that it's the first time I've seen a fear this pervasive that everyone from top down experiences it. And I love your approach that's human first, because it's you know, I approach
problem solution as opposed to solution problem. Like figure out the problem you're gonna solve first before you buy the AI tool. But I think the human part is so important, which is not a hundred percent my area of focus, but I know it's so important. So can you explain a little bit about your and then where people who can find out more about what you do and where they can see what you do online, like the way I found you on LinkedIn by some
of this cool stuff you're talking about. Yeah. Yeah. So one of the things I've introduced to my own like lifestyle realm is the fun factor. What is the fun factor here, right? if it's not fun, why am I going to spend my time on it? And we don't we don't get to choose everything we want to do, right? There's there's there's bunches of things in life that we just have to do to do to get it done. But how can we
make this initiative fun for people, right? Like we have early adopters, we have mid, and then we have late. Let's just break them down to three categories. They all have different needs as it relates to this. But what they have in common is probably an interest to talk about it, an interest to teach and share, right? And so the idea here is how do we elevate the fun factor on any initiative we're trying to integrate, specifically as it relates to AI for this one,
right? And so we once again we protect their time. We give them opportunity experiment. We ask, we invite early and frequently for dissent and feedback that may or may not be pushing us in the direction we want to be, right? We want to make sure the tools fit for our applications. There's nothing worse than you think about a lot of the enterprise software programs you've probably encountered over your your years. They sell you a package, or at least they sell you a promise on a package,
but then you finally get your package and it doesn't do exactly what they said because you didn't upgrade enough packages, right? So the same thing's true with this. We gotta figure this out because if we don't cultivate this fun element and we don't let these early adopters really run with this and find enthusiasm it, how else are we gonna learn about this? You might have one or two people on a staff, or like you said, a fractional, right? That that is your content expert, but but
you still have your day job to do, not just teach people what an L L is versus, you know, a neural pathway and and these different things, right? Like We need to let people exist and explore and get dirty just like we did when we were kids and and provide feedback around that. And and how we do it is through a lot of clear communication, creating an environment where people feel safe and supported to experiment, alignment throughout the system,
and we have a lot of people touch the tool so that we can gather feedback early on to see if this is a good fit for us. Is this moving us where we want to be? Because the reality is that. These are powerful tools. Not all of them are created the same. And salespeople like to sell stuff. And in some instances, they might promise you everything, or in other instances, they might promise you a product that doesn't quite meet your
needs. And we, in my opinion, have have a responsibility to protect our organization, to protect our employees, and to give this stuff due diligence so that we're not merely blindly adopting. in im imposing these things on people that we're embracing these opportunities. Amazing. And where's the best place for people to see what you're doing online and some of the stuff you're writing about? Sure. Yeah, so a couple different places. LinkedIn, as you mentioned already. my URL in LinkedIn is Robert R-O-B-E-R-T Lion L-I-O-N,
even though the last name is pronounced Lion. Thank you. Jonathan earlier. it's spelled like Lion. you can find if you get into my profile, you'll see ~ my engagements recently have just been AI plus human blog posts. So on Substack, ~ I'm I go by friction profit and What I've committed to in 2026 is every Tuesday pushing something out that relates to the interactions between humans and AI. And right now moving into an eight-week series on leading in AI. so leadership skills, acumen, stuff like that.
every once in a while I talk about there's one out there that got a lot of a lot of popularity is my son's going into college, and college is getting hammered by AI. And how do we navigate that? I'm a professor. I you know, there's a lot of uncertainty there in terms of higher education of whether or not they're doing what they should be doing. so I talk about all sorts of things related to people in AI integration. You could also visit website Black River,
the letter P, the letterm dot com. That's ~ that's our company, Black River Performance Management, where we work with things relate to culture, ~ training, development, assist people with. l hiring strategies, stuff like that, or you could go to lionclarity.com to visit our other business, which is actually intended more along the lines of what you're talking about here is major implementations, whether that's AI, strategic planning, organizations missing their marks and they need to rehaul or retool post merger work,
stuff like that. So those are a couple of ways to find me and dig around and learn more. throughout the LinkedIn pieces, there's a bunch of AI related stuff in there. Some I'll put everything in the show notes and below the video on YouTube. As always, thank you so much for being here today, Rob, for an amazing episode of the Artificial Intelligence Podcast. Awesome. Thanks, Jonathan. Good to see you again.
