Intro: Broadcasting live from the Business RadioX Studios in Atlanta, Georgia. It’s time for High Velocity Radio.

Lee Kantor: Lee Kantor here. Another episode of High Velocity Radio and this is gonna be a good one. But before we get started, it’s important to recognize our sponsor, Business RadioX. If you’re a business coach or consultant who’s tired of being a best kept secret, set up a call with Stone Payton at book Stone Payton dot com to unlock your perpetual prospect pipeline. Today on the show, we have the CEO and founder of P3 Adaptive and author of the book “Fair Game” Customizing AI to Your Business Is Easier Than You Think, Rob Collie. Welcome.

Rob Collie: Hi. Thank you so much. I’m happy to be here. How are you doing today?

Lee Kantor: I am doing well. I am so excited to talk to you about this topic. But let’s start with the book fair Game. How did this come about?

Rob Collie: Well, I. I didn’t actually intend to write a book. I, um, about a year ago, a little over a year ago, I needed to start plotting an AI course for my company. And even though I have 30 years in technology, like I used to lead teams at Microsoft on things like Excel and Bing and Power BI, I was really struggling to grasp like what to do about AI for our company and or even knowing where to start. Everything I was reading out there was like either scary and intimidating or way too high level and fluffy. And we strategy comes from like this middle altitude, like, which was missing enough detail to start building a picture of what it really means for your organization to adopt AI, but not like down in the weeds writing code. And no one was really writing at that altitude and patiently explaining, uh, so I had to go research. Even with my technology background, I had to go research all of this for myself, like, you know, go on my own little journey with it. And what I found was really so much simpler than what I expected. Lots of, you know, there’s a fair number of new concepts for sure. I mean, it is new, but taking one at a time, business leaders can understand all of them. It’s the book I needed. So I ended up writing a book that I needed when I started, because I kind of felt like I almost like, owed it to other people in my situation. And that’s, that’s how this came about. Like I was, it was not I didn’t go into it planning to write a book. I went into it planning to like figure out what to do for our company. And it was just sort of the nature of what I found that I was like, oh, now I really kind of need to share this now.

Lee Kantor: Can you educate our listeners when it comes to a conversation about AI? Because I think that AI, it means different things to different people. And a lot of leaders, you know, they say, okay, we should have an AI strategy. And it’s like, that’s a blunt instrument. You know, that’s like saying we should have an internet strategy. It’s like, it’s too broad. Like every department and team is going to use AI in a different way. And within those teams, then there’s going to be somebody that’s like a power user that’s going to really use it in really disruptive ways. And, and other people are just going to use it as kind of like a tool. Like you would use a thesaurus, you know, like, yeah, how do you help, uh, clients kind of understand holistically that an AI strategy is what it means holistically as well as, as tactically.

Rob Collie: Well, first, I want to acknowledge that you’re exactly right. I mean, there’s a lot of pressure on business leaders to come up with an AI strategy. And it’s very often worded like that, like as if it can be this monolithic thing. And, you know, there’s a reason org charts exist, which is no one can see everything. Um, and if you think about a company, a company is really made up of thousands, maybe tens of thousands on the sides, but at least thousands of individual workflows that are going on all the time. And, um, AI is essentially like different in terms of what it can do for you with every one of those workflows. So it is really kind of a, as you were suggesting, it’s, it overall in the end ends up being something that you kind of solve case by case. You know, um, bottoms up in a way. You need to have a strategy on how to approach doing that. But the idea that you’re just going to have some sort of centralized monolithic strategy and you’re supposed to come up with it is already sort of like it has. It’s an impossible task when you when you view it that way. Um, so the, the first thing I think that people need is here’s the way I describe it, Even if you haven’t built software, you didn’t, you know, work at Microsoft like I did for a while. Uh, you understand, everyone listening to this understands way more about traditional software and computing than they give themselves credit for.

Rob Collie: Like if you everyone’s what people know about software and what it can and can’t do could easily fill a 250 page book. And you don’t have to be an engineer for that to be true. Um, but this is AI is a new kind of computing. It’s the first new kind of computing we’ve gotten as a society since like World War two. Everything, everything from World War Two on is exactly what we all grew up with. And you need that same level of understanding of this new kind of computing that you have of the first kind, uh, before you can really start to even, you know, kind of like see through the noise. And so that’s the first step, I think, is, is getting that same sort of intuitive, like very, you know, like comfort, uh, with like sort of what it can and can’t do and where to use it and how to deploy it. And, and at that point, things start to get clear. That doesn’t, that doesn’t like do all of the work for you. Like that’s just, but like, that is the foundation that you need in order to get started. And that’s, that’s the stuff that I learned on my kind of my, you know, journey down the rabbit hole. And that’s the stuff that I’m sharing in this book.

Lee Kantor: So if you were a leader, you were to advise a leader. Um, what, what’s kind of the baby steps they should be taking right now? Um, or is there something that you don’t start at the, at the top level? Maybe like you mentioned, this is a bottom up kind of plan.

Rob Collie: I think there’s two things here. There’s, there are, you know, there’s again, like multiple altitudes at which you need to approach this stuff. Um, one of them, I think the, I give you an example, sort of at both, at both heights, right? So like a high level example and then a low level. So the, the high level example is basically to understand that, um, these LLMs basically have a PhD in every topic that humanity offers PhDs in. I mean, it just, they know so much and they, they are so they’re so capable of reasoning through all of those things, but they still show up at your business every day. And in every chat, they show up as a new hire. Uh, they know all of that other stuff, but just, just really, you can just picture them as this, you know, this 28 year old that has a PhD in everything, but it doesn’t know where the, the bathroom is. It doesn’t know your company. There’s so much about your company that it doesn’t know. Um, and the, the biggest thing that the biggest challenge and also the biggest opportunity is understanding that your job as an organization is to build systems that educate the LM. Every time it wakes up on all the relevant details of your business. And that’s just like, essentially like, almost like handing a training handbook to, to the AI every time.

Rob Collie: Um, and that’s not, it’s actually really not that hard to do. Um, and like, because when you buy an off the shelf subscription to, let’s say like ChatGPT, which is really kind of a natural first move, like if you’re under pressure to get your organization to adopt AI, you know, the, the most natural first move is to go buy subscriptions for everybody, like to copilot or ChatGPT or Claude. Um, which I completely respect and understand. We did that too. Um, and, but when you use these things, these AIS for personal use. They’re world beaters because they know everything that humanity has ever known. I mean, they, they, they know like 40, 50, 60 times the amount of information that’s in Wikipedia, but they know nothing about your business. And so when you turn to use them for business purposes, they tend to face plant a lot and need a tremendous amount of help. And the way to sidestep that is to, you know, like I said before, essentially like you got to feed them the information that they need about your business, which, by the way, is just a really traditional information and software problem. That’s not an AI problem. You know, giving that information to the AI. And when you do, suddenly the AI becomes a specialist in your business and becomes like hyper competent.

Rob Collie: And that transformation is a really, really, really big deal. And so like the whole reason why my book says customizing is that the off the shelf stuff that you get with a subscription is never going to get it done. It’s never going to deliver the promise that we’re, we’re talking about, um, as a society, like all the hype that you hear, is never going to happen with the off the shelf subscriptions. You have to you have to do this sort of like structured training thing. Um, which again is way easier than it sounds. And then for a low level, like, like really detailed example, um, one of my favorite things to, to, to get people started with is any place where you’re, when you need an, a question as a, as a business leader, uh, or even a mid-level manager, whenever you need a question answered, that’s about like your company’s business data, like how well did we perform in Q3 last year with this product line versus this year, Q3? Um, you shouldn’t in the new world, you shouldn’t have to go looking for a report. You shouldn’t have to go looking for a dashboard. You should just be able to ask a question, you know, like of a ChatGPT like interface, like maybe even ask your, you know, like dictate it to your phone and have the AI go look up the answer for you.

Rob Collie: Go research the answer for you. Um, and these things, you know, I think the industry is starting to call these things data agents. Uh, data agents are such a kind of like a lay up as a first thing to engage with. It’s, it, it solves a problem you already have, which is getting information conveniently is really, really, really difficult. No one likes looking for dashboards. No one likes realizing that the dashboard they need doesn’t exist. Um, no one likes trying to figure out how to use a dashboard like we think of dashboards as simple, but they really aren’t. And being able to ask a business question the way that it that it’s formed in your head, not the way that some dashboard builder thought you’d ask the question and have something go and do all of the, the work behind the scenes and come back to you like a, you know, a minute later or less with with an answer and also be willing to answer follow ups is a. It’s just a massive game changer. And I think that’s sort of like the most, the easiest place for people to start, uh, is that use case.

Lee Kantor: Now from let me make sure I’m understanding this correctly. So you’re saying that, okay, AI by itself off the shelf has all the, like you said, PhD and anything and everything. And it’s much smarter than your smartest people because it has this wide breadth of, of knowledge and, and it kind of has all the data it needs to know everything. Now you have to kind of layer that all that kind of macro knowledge with the micro knowledge of your specific business. So now you have to share and create a place where the AI can see all of your data and understand that as intimately as it understands everything. And then when you combine those two, that’s where the magic begins.

Rob Collie: That’s exactly right. When those when you give it that extra sort of last mile about your business and you know, you use the word data. And I think it’s we should sort of expand that and say data and knowledge. Um, you know, so access to your business data is important. You know, if it wants to analyze performance and help you, you know, plan things, etc., but like even just English, uh, English written handbooks that, that, that explain to the AI sort of like how your company operates, like how it’s differentiated in its industry, what your brand voice is, um, what, what important decisions and initiatives you’ve set for this quarter. Um, you know, things that we’ve thought about as just being like documents in the past are now also kind of in play as data. And so it’s this hybrid, it’s actually probably more of the English stuff, uh, even than the structured data stuff that that the eyes are missing. Those are those are things that when these eyes are being built, these LMS are being built. They don’t have access to that information about your company. Um, and even if they did, they wouldn’t train it in because it would make the LM way too large and expensive. So yeah, you have to, you have to fill that. I call it the, um, the knowledge cliff it, if it’s not, if it’s not something that was available on the internet, the LM doesn’t know it. Um, and so, and that basically like there’s a tremendous amount about your business, you know, again, this, the, all the information about your business would fill, you know, like the Library of Congress almost it’s so large, right? Like, so like you can’t give it all of it. But, um, you got to pick and choose what each of these specialists that you’re creating, what is it that they need to know and make sure that you’re handing them that information when they first wake up every time you interact with them.

Lee Kantor: Now when you’re working with clients, especially larger clients. And you’re telling them, okay, you know what? We got the, the off the shelf. It’s got all the stuff and now we’re going to give it that last mile stuff. We’re going to give it as much of the stuff you have here so that it, you know, the AI can. Now that to me, this is answering some of the holistic questions where now everybody on the in the organization has access to all of this great information is that creates some fear. Now in, in the in the leader’s mind of, well, if everything’s in one place and now this AI has access. Am I now vulnerable to other people having access?

Rob Collie: Well, I mean, because it’s a. Um, so there’s two parts to this question really. So one is that, you know, all this information we’re talking about is already in most places is already stored in the cloud somewhere.

Lee Kantor: Right. But the thing about AI is it’s so conversational and so easy that now it seems like now I’m just. Now it’ll be easy for anybody. Before, I couldn’t find stuff in my own house. Right. Like, how many people couldn’t, you know, they couldn’t answer those questions you said should be there. And it was all documented. It was somewhere there. But now with an AI layer, you’re making it easy. And now is that create maybe, you know, I’m not saying this is rational, I’m just saying that does this create some fear where, hey, if, if I can go in there and type, you know, give me this specific thing, what if somebody jumps in here and asks that same question? They can find this stuff easy to.

Rob Collie: Yeah, really, really good question. Okay, so first of all, you can’t create a company super being that has access to that sort of knows everything intrinsically because it can’t absorb that much. Um, it’s, you should think about this as like you’re creating, um, many specialists at your company, each one sort of knows its lane really well rather than like someone that like a thing that knows everything. Um, now, of course, you do want some sort of general purpose, you know, agent at your company that helps people look up answers to whatever it is that they want, right? I do understand that, of course. So you need that as well. But the really good news is that, um, behind the scenes when you’re building these, these, these systems, these custom systems that educate the AI and give it access to the information it needs, those systems are regular software systems that can and should respect all of the security rules that your organization, you know, already follows. So if you know, you know, person X, um, is logged into an AI and asking a question if that person didn’t have access, you know, if, if information is being given to the AI or potentially being given to the AI, like from some like, let’s say some, uh, strategic planning SharePoint site.

Rob Collie: If person X doesn’t have access to that strategic planning, SharePoint site, then the, the software behind the scenes that you that’s controlling information won’t even give it to the AI if person X shouldn’t be able to see it. Like if you’re absolutely correct that once you give information to an LLM and the and the user’s chatting with the LLM, even if you tell the LLM, whatever you do, pinky swear, don’t, don’t reveal this to the user. That’s not security. You can’t you if if there’s something that the user shouldn’t know, it can’t be given to the AI in that session. But then if, uh, the CEO logs in to that same agent, the CEO’s permissions are different. And the back end, the back end software that controls what the AI has access to. Well, then, you know, make that information available to the to the LLM. So it’s really just kind of the traditional security problem that we’ve always had. Um, and it just needs to be applied in a new place. Um, and the AI part of it doesn’t really actually make it any more complicated than it ever was before.

Lee Kantor: Now, does the AI, um, like you mentioned that you, you mentioned a few times now that, um, there’s, there’s so much information and data and I think people maybe underestimate the amount of compute or cost in accessing all of the data all the time at your fingertips. And it isn’t like that yet, right? It’s not that inexpensive yet to be able to access every, we can access a lot, probably a lot more than we ever have before, but the cost can creep pretty quickly. If you want to access everything all the time at your fingertips, you know? Right? Like this is something that is a little clunkier than maybe it’s being, uh, depicted.

Rob Collie: Yeah, yeah. So some good news there. Um, certainly, uh, the more, you know, lm time you burn, the more the, the meter runs and costs can, can stack up. Um, but the good news is that a customized AI system, like what I’m talking about is a hybrid of the, the new kind of computing AI and the old kind of computing we all grew up with. And, um, when you do it right, when you build these customized systems correctly, you’re using the old style of computing to do a lot of the work. Um, rather than like when you go to the out of the box system and you talk to the, um, you know, the AI, like it’s really kind of mostly just going to use the AI, which is the most expensive part of the system. But when you build these customized systems. So for instance, in the example of the data agents thing that I said, you know, like be able to ask this thing a question and have it go research the data. Um, the AI itself isn’t the one that’s looking through your data. The, the LLM is instead sending commands to traditional software that’s designed to answer that runs on CPUs is very, very, very cheap. And the, the traditional software is doing all of the querying all of the work in terms of sifting through the data. Um, and just, and then just returning the answer back to the LLM for it to synthesize and turn into a paragraph or a chart for you.

Rob Collie: And so you try to, and you strive to do as much of the work in these customized systems. You do as much of the work as you can in the traditional CPU style of computing that we’ve all grown up with, and only use the L S for the things that are uniquely good at. One of the examples I give in my book is you can give an LLM a complicated division problem like arithmetic, and today it will give you basically very, very close to, you know, within like, you know, like 15 decimal place precision. It’ll give you the right answer. It’ll, it’ll give you the right answer to that, to that, to that division problem. Right. Um, but it burned hundreds of trillions of calculations to do that. It did it in the most inefficient way possible. Whereas a calculator, a solar powered calculator, like from the 1980s could perform that same, um, that same calculation in an instant and cost you nothing. Uh, and so in situations like that, when you have total control over the system and you’ve, you’ve built these custom systems that my book talks about, like your system will be smart and it won’t ask the LLM to do that division. It will give it to the cheap part of the system, which does it in like one billionth of the effort that the AI would use.

Lee Kantor: But this is something that you need an advisor to help you do. This is not something that most organizations can do on their own. This this sounds like you need expert help.

Rob Collie: I think the line there is pretty blurry. Um, I mean, certainly that’s, that’s my company’s business, you know, is helping people with data and helping people with, with AI. Um, and it’s kind of, that’s also why, uh, I needed to come up with a strategy for this stuff sooner than most, I think is because of just the nature of the, the business that we’re in. Um, but, um, you know, I kind of lost my thought there.

Lee Kantor: Well, it’s one of those things where internally you might have somebody on the team or maybe a team that can do this for you. But for most organizations, it sounds like this is definitely a stumbling block, or at least the thing that keeps the leadership up at night.

Rob Collie: Yeah, absolutely. So, um, like, you know, every company is different. And I do talk in the book about the different personalities that you already have on staff that might be good at this stuff. Um, you know, some companies do employ, uh, software developers, um, and, um, and some don’t, um, and some software developers have embraced AI and understand the principles that I’m explaining in my book, but I still think a lot of them don’t. Um, and, you know, the first audience for my book was my own company. Uh, the technical consultants at my company have been, you know, doing data and business intelligence work for their, you know, the bulk of their careers. Ai was still kind of new to them as well. But so over the past year, I’ve been giving them early drafts of the book and, you know, they’ve been kind of racing past me. You know, once they, they’re like, they’re like, oh, that’s it. That’s the, that’s the story. Okay, we can go do that. So there’s another personality that I talk about in the book that I think exists everywhere. And I call them the crafters, and they’re the people that make the spreadsheets at your company. They’re the people that make the dashboards. They’re the they’re like the 1 in 16 that grow up in the business, not in it, but, but gravitate to and have an affinity for tools like Excel. And, um, when they read this book, when they get the fundamental understanding that they’ve been missing, um, a lot of them, I think are going to become a big part of the AI building workforce. Uh, for, for many companies, they’re not going to be the only workforce.

Rob Collie: Um, but I think they’re going to be a big part of the answer. We’re very, very, very, very early in the process of them, of this, this demographic waking up to what they can do with AI. Um, in particular, now that AI can help them write code, which they’ve never been able to do before. And given that these custom software, these custom AI systems I’m talking about are really just regular software, um, wrapped around an LLM, uh, when this crafter audience starts to get comfortable with, with, with building software solutions, like, you know, uh, they can do, they can do that wrapper stuff. Um, I think a lot of companies are going to discover that they have talent on staff that they didn’t, you know, again, they’re not ready yet. They’re not ready for it yet. But like, you know, fast forward a year or two. Um, and they will be. So I, it’s kind of like with our data business, right? Like, Um, there are lots of people out there who can do, um, like power BI work, which is one of the most common things that our company has done over the years. Um, and yet we still get hired by companies to come in and help with it. So I think it’s kind of a mix. Um, you know, I mean, I, I, I’m, I definitely think that we’re going to, you know, we’re going to be increasingly hired a lot as a company to help people with AI. Um, but like, you know, I don’t know that it’s necessarily that like your very first thing that you do needs us. Um, it really is case by case basis.

Lee Kantor: Right? But well, the first thing will probably be to get the book, I would imagine is a good starting point. But, um, for folks, uh, like in your business in, uh, in your kind of day job business, what is kind of the pain they’re having right before they call you? There has to be something. I mean, they must be feeling some sort of angst around some issue or some something has to be triaged in order for them to go, hey, maybe we should call Rob.

Rob Collie: Well, traditionally they’ve called us because they can’t make sense of their data. Um, but.

Lee Kantor: How does that show itself? Like what? What is a symptom or a signal that says we’re not making sense of our data.

Rob Collie: System is that whenever you’re in a meeting and you need, you’re like, oh, if only we knew how well that product line, we have a theory that our product line did better last year in this, this segment than it, you know, than it did the previous year after we made those changes. But you don’t know whether it did. It’s just a hypothesis, you know, um, and you’re in the meeting and you have to make the decision without knowing the answer to that. Um, or, you know, that that thing that we’ve been doing to try to improve, uh, improve performance, is it actually working? And if it is, is it working everywhere? Again, like if like very often meetings and decisions and all that kind of stuff have to be made faster than you know. You can’t just say, okay, we’re going to send someone off to research this for, you know, for a week. You know, the decision needs to be made now. And so very, very often decisions are made, uh, without the information. Even though you’ve collected all the information, your systems of record have all the data that you need. You still can’t, you can’t see it. And that’s been the thing that traditionally has brought clients to our door.

Lee Kantor: Um, and that’s a great point that, that to me that is well articulated in terms of that’s when, you know, you need help is if you have been doing something for a long time and now it’s time to make a call, a decision, you got to make a move and you’re still relying on. Yeah, let I’m trusting my gut. And it’s like, there’s too much information out there. I shouldn’t have to be trusting my gut on every move here. There has to be a better way to kind of see if at least I’m going in. I’m directionally correct.

Rob Collie: Yeah. And then, you know, in the, in the AI world, um, like it gets even better, like I mentioned earlier, like you don’t even need to go looking for a dashboard anymore. Like literally have a research agent that can go answer your data questions again, assuming you have access to the data, right? Like you, you’re approved and certified to have access to that data. Um, but within the, in the AI space, um, the reason people call us is either, um, that they’re, that they’re facing that question like, what should be our AI strategy? What should we do about AI? And they’re, you know, they’re in the same place that I was a little over a year ago. Like the information out there, just, it’s very hard to synthesize anything coherent, um, for yourself and for your organization. And like, you know, you’re a business leader, like you came up because you’re really, really good at business. You weren’t, you didn’t get to the position you’re in because you were an AI expert just like me. Um, so that people come to me for that, but also to our company for that. Um, but there’s also this like, okay, we bought the subscriptions, uh, the AI subscriptions. We did the thing we were supposed to do and we’re not getting the results out of it that we want. That’s kind of the other symptom that drives people to us. But like you said earlier, I think honestly, um, the first step, it shouldn’t be to call us.

Rob Collie: I really do think that you should like just, you know, get this book. Um, like I, we believe that our, our company that, um, informed clients are better clients, um, you know, most consulting organizations, they sort of deliberately hoard information, um, and keep their clients in the dark and keep them a little bit confused so that they can sort of like, kind of like maintain power over the relationship. And we’re not like that. Um, we would much rather have our clients, uh, possess the clarity and confidence to green light a project because they know that it’s going to work, um, than being suspicious that it might not work. Uh, or like, you know, we’re just, ah, I just, I started this company, you know, um, 13 years ago because I’d seen how the consulting industry worked around tech in general. And it just, it was just so distasteful to me. Um, and I wanted to go and see if we could build a different kind of company that was much more transparent, um, and much more like benevolent. And, um, and we’ve proven that it works. Um, and so we’re trying to bring that and we will bring that, that same approach to AI. So, you know, here I am, I’m a owner of a company that does, you know, um, five and six figure AI and data implementations. Um, and I’m telling you, you should instead, you know, start with a $30 book.

Lee Kantor: Well, if somebody wants to get a hold of the book or get a hold of you at P3adaptive, what is the best coordinates for each?

Rob Collie: Uh, so the book is, uh, you know, is this live or are we recorded?

Lee Kantor: So it’s live to tape. So we’ll, this will come out in a week or so.

Rob Collie: All right. So the book is out this week then because the book is coming out. And so, um, so the book is everywhere. Uh, it’s Amazon, Barnes and Noble, wherever, wherever books are sold. Um, there’s also a bundle that we’re selling through the publisher’s site at fair game book.ai. Um, and that bundles the hardcover, which is what you can get on Amazon, the hardcover and also includes the e-book and the audiobook formats. Uh, and you’ll receive those before those hit Amazon. Um, you know, if you order through that site, but either way, um, just get a hold of it. Um, and start reading it. The, the challenge I give people is, is say, look, just pick it up. And if it’s not a page turner, you can put it down. Um, most people are finishing it in 1 or 2 sessions. Um, I wrote the book to be, uh, to hold your attention and to be understandable. Um, and that, you know, early reviews are indicating that that has, that is the case. So, um, you know, it’s not, it shouldn’t, it does not feel like homework. It feels more like I wrote it to feel like we’re on a road trip and I’m explaining things to you.

Lee Kantor: Good stuff. Well, thank you so much for sharing your story today, Rob. You’re doing such important work and we appreciate you. And again, the website for the book, Fair Game book.ai. Rob Collie with P3 Adaptive. Thank you again.

Rob Collie: Thank you so much Liz. It’s been a real pleasure man.

Lee Kantor: All right. This is Lee Kantor. We’ll see you all next time on High Velocity Radio.