How to Build an AI Strategy That Delivers Real Business Results | Eddie Irvin

Most AI projects do not fail because the model is weak. In this episode Chris talks with Eddie Irvin, founder of Nashville AI Advisory, about the practical work leaders need to do before AI can deliver value. Eddie explains why process clarity, prioritization, and communication between business owners and technical builders are often more important than choosing the newest tool.

Chris and Eddie break down how to identify a strong first AI project, why CEOs should stay involved in defining business rules, and how richer context improves results from ChatGPT and Claude. They also cover dictation, roundtable prompting, Codex, Claude Code, and using multiple models to pressure test important decisions. Listen if you want a grounded framework for moving from AI interest to useful systems your business can actually operate.

Chapters:

(00:00) Introduction
(02:56) Bridging the Gap Between Business and Technology
(09:13) Where AI Transformation Actually Starts
(11:24) Why Process Design Comes Before AI
(18:54) Finding and Prioritizing the Right Business Problems
(22:28) What a Good First AI Project Looks Like
(27:45) Why Clear Processes Make AI Implementation Easier
(32:49) What CEOs Should Never Delegate to AI Developers
(35:52) Better AI Results Start With Better Context
(48:06) Why Leaders Need Hands-On Experience With AI

🔎 Find Out More About Eddie Irvin

Eddie Irvin on LinkedIn:
https://www.linkedin.com/in/eddie-irvin

Nashville AI Advisory:
https://nashvilleaiadvisory.com 

🛠 AI Tools and Resources Mentioned:

ChatGPT:
https://chatgpt.com/

ChatGPT Desktop:
https://chatgpt.com/download/

Codex:
https://openai.com/codex/

Claude:
https://claude.ai/

Claude Fable 5:
https://www.anthropic.com/claude/fable

Claude Code:
https://www.anthropic.com/product/claude-code

GPT-5: 
https://openai.com/gpt-5

Transcript

Copy Transcript Copied

SPEAKER_02 0:35

I find a lot of the work that I'm doing is process-oriented system design before we even get into AI. And sometimes we don't even need AI. We just need the information that's over there to go over there and be processed a certain way.

SPEAKER_00 0:46

What do you think a good first AI project should look like for somebody who's getting started again?

SPEAKER_02 0:51

Again, we want to look at the amount of time saved and who we're saving time for. We've got to decide are we saving the C-suite time or are we saving the workers' time? What are some things that a CEO should never delegate to the AI developer? I'm coming back to business process, I'm coming back to that domain expertise of giving the boundaries to the developer for, you know, here's what we've got to do this very specific way, or here's where we've got to check what's going on. If we skip that step, then we end up with things being built, but we're working on it. Oh, wow, we got a lot of things going. And then later on we realize this isn't really helping the way we thought it would. And so welcome to Using AI at work.

SPEAKER_00 1:24

I'm your host, Chris Dag. Each week we'll be learning how today's business owners, entrepreneurs, and ambitious professionals are getting more done with smart use of tomorrow's tech. Let's get started. Right now, every business leader is asking the same question. What are we going to do about AI? If this is you, ChiefAIOfficer.com has the answer. We give you a simple path forward where we provide executive and team training so your people know exactly how to safely use generative AI in their day-to-day. We also manage the deployment and implementation to make sure tools actually get adopted and deliver results. And we'll also guide company-wide transformation so AI becomes part of your operating system, not just another shiny object. The companies that act now will increase productivity, cut costs, and grow faster than their competitors. Those that wait will get left behind. So if you want to make AI work in your business, visit ChiefAIOfficer.com and see how we're helping companies of all sizes finally get results from AI. Hey everybody, welcome to another episode of Using AI at Work. My name is Chris Daigle, and I am the host of the show. And today our guest is Eddie Irvin. And Eddie is an AI strategist and founder of Nashville AI Advisory. As you might expect, he's focused on helping established businesses just make better AI and automation decisions. And this is a very uh practical conversation. It's going to be a chance for two AI implementation leads to be able to talk about what we're seeing in the market, um, maybe patterns that we're seeing with clients where they might be getting stuck on things and that sort of thing. So if you are uh someone who already has AI in the business or you're exploring it, I think that what you're gonna hear, uh this is kind of some inside baseball, at least that's my plan for the conversation today, so that you as somebody who's looking for talent to support you, um we'll kind of have a better idea of what our side of the table is thinking about when we're working with clients. Eddie, that sound okay to you? Sounds great. So excited. Let's dig in. Good. Um, anything like in particular that maybe you want to share with the folks to kind of level set the AI expertise that's represented there in Nashville by you?

SPEAKER_02 3:31

Sure, sure. Yeah. I I mean the first thing that pops to mind is the the chasm or the canyon in between the business owner and how they understand their business and the tech lead and the way that they see it. Two distinct universes almost. And this bridge in between is so helpful. Uh, as you know, they're both taking care of what they got to take care of. They got they get to get the things done that they need to get done. It's the translation back and forth. It's the way that they can connect that can really allow us to have uh the best success. And that's where I found a lot of uh where I'm playing is understanding both sides and helping us kind of make that more seamless there. Uh, but I find that a lot of business owners that dig in and start playing with AI enough to know what they don't know or kind of seeing, they they'll get so much better success uh because they they've at least touched it.

SPEAKER_00 4:18

So are you coming at it from somebody with a technical background that can translate that technical into a non-technical conversation with a business owner? Or are you um what's what's your spot in that that communication bridge that we just talked about?

SPEAKER_02 4:33

Yeah, a little bit of background. So uh I was a developer first, so writing code, and this is before AI, so goodness gracious. Um, and then moved into building an agency. So I started hiring developers, hiring managers and playing uh kind of playing higher up in the visionary uh role. And then I got into app planning. So this is all about getting clear on what we're building before we do it. A la, creating a blueprint for a house before you build it. You want to make sure you have the blueprint set and solid before you actually take the hammer out. Same kind of thing uh with apps. So I got really comfortable translating technical to non-technical. And very, very often, a lot of the people that hired me were non-technical. They wanted to build an app, they knew what they wanted from the business side, but they didn't really know how to get that done with the tech. So I ended up being that bridge in between uh the technical and the non-technical uh sides. And what I find now is that I'm playing, depending on what the what the need is of the customer, I'm typically playing the non-tech, you know, helping translate to the non-technical business person. But every now and again I'll find somebody who's a whippersnapper with AI and they've used it as much as they possibly can, at which point we can just hop, skip, and jump through a lot of that stuff because they're already starting to implement, and I can help supercharge them and supercharge their uh their systems.

SPEAKER_00 5:45

Okay, cool. Let's talk about this a little bit. So the uh businesses that are, I mean, you're listed as Nashville AI Advisory, but obviously with uh the internet, you could work with clients from all over the place. Do you prefer to work with clients that we you can go on site and work local?

SPEAKER_02 6:02

That's what I'm aiming for now, building this kind of localized Nashville AI advisory. I want to build more in and around the city that I'm in. What's happened for the past decade is I've been working for uh across the US and sometimes into Canada. And I got really excited about being more connected to uh to the city of Nashville. So so the idea of being able to drive to you know the print shop and work with, you know, and and harness AI inside of the print shop when we're on location is another level. It's different than being on a Zoom call all day. Um so that's what I'm aiming for here. But yes, absolutely, Zoom allows us to be accessible anywhere.

SPEAKER_00 6:36

So for the listeners who are, you know, kind of vetting talent or looking for talent and that sort of thing, um, what is, I mean, there's the obvious, like, oh, I can go and sit across from the person and we can break bread, we can build rapport and that sort of thing. But is there any uh production or uh like fulfillment benefit, would you say, uh, if they're looking for somebody, like look for somebody local as compared to and to kind of piggyback on that? Well, first, what do you what do you think about that? Like, what's the yeah? Would you encourage that?

SPEAKER_02 7:06

Well, let's talk about the type of business, right? So, in the case of like a print shop where we literally have, you know, there's this here and there's that there, and kind of being able to be in the space, that is so beneficial to being able to do it, still work remotely and still work on systems and and around, you know, connecting integrations, but there's something to being, you know, boots on the ground. If the majority of the work is inside of the computer anyway, and it's spreadsheets and it's stuff that can be easily transferred online, then it's not as important to be in uh in the same place. But communication is sometimes just easier to sit in the room and be in the space. So just kind of depends. I can work with both, and I'm very used to working remotely. Uh, but I love the idea of I'm in Nashville, it's a beautiful city. I want to get more involved in this city specifically. Uh so that's what I'm aiming for. Yeah.

SPEAKER_00 7:49

Yeah, you know, I would agree. I mean, for us, uh, we always kick off our engagements with clients with an an on-site, like in the room. And a lot of times it's not local, so we'll fly in, and a lot of times we'll schedule a dinner with the the principals or whoever the night before just to get to know you. What would be a takeaway? And I can tell you that the the opportunity that we get to spend that time on site with a client is it makes a big difference in them trusting you to be like an AI resource that's not gonna screw anything up, right? They can get the job done. Um just also, again, just building a relationship with a client so it's not so transactional.

SPEAKER_02 9:19

Yeah. Yeah. This idea of building relationship and and having good communication is a huge part of the success with these projects. It's not just the starting communication or getting in the door. It's as time goes on, if something breaks or doesn't work right, we want to have that open line of communication. So making sure that that's really established well at the beginning uh with not just a single person, but multiple people inside the organization allows for that information to find its way back to you. When you're the one who can build it or at least kind of direct the build, information is gold. And so you need to make sure you have great uh conversational channels.

SPEAKER_00 9:51

So there's a number of different approaches and methodologies for AI quote unquote transformation or leading AI deployments in companies. Some start at the strategic level and get the executive team tuned up before they start talking about it more broadly. Some call, you know, and already have a hundred licenses or 50 licenses or whatever it is for their team. Hey, we got these Chat GPT or Claude licenses, but we need training. They're not using it like we thought they would. Um some say, hey, look, I've got a very we've got some specific pain points in our process and our different departments, and we have a good idea that AI can help it. Um, can you come in and build the solution? Where are you typically uh getting involved, or what are what are your clients typically reaching out to you about as out of that spectrum of uh reasons for outreach?

SPEAKER_02 10:39

Yeah, well, and you you bring in a great point that there's everybody starting from a different spot. It's not like everybody is just doing this exact thing in the exact same way, and then you go, oh great, we start always start from the same place. It is different for a lot of different people. I'm very typically at the very beginning of the process. We know we need to harness AI. We may have already started harnessing AI or the team is just starting to use AI, but we don't have a comprehensive plan around it. We don't have comprehensive systems that use AI and human in the loop design uh for these different systems that are happening. We've got manual emailing, we've got manual phone calls, we've got manual all this other stuff. And these business owners are seeing the writing on the wall that if they don't harness AI or start to see what's uh you know how they can use AI and bring AI into the company, that they are going to then be left behind. So that's very typically where I'm entering in that the business owner knows that this is very important. I've got to harness this technology, but then there's a gap there. Well, how exactly? And what exactly? Yeah, and of course, they're being inundated every way with you know, with this workflow and this agent, and everybody online just says, well, turn the agent on and then it'll all work. Yeah, and then they go ahead and they try to do that and they find out that's not the case. Yeah, exactly. Yeah, yeah.

SPEAKER_00 11:52

So the are you seeing um any types of trends over the say, I guess just really in 2026, with the people that you're talking to, their level of savvy or the the awareness of problem or the interest in AI, have you seen much uh I don't know, change or or anything noticeable in the behavior of of the clients and prospects since January?

SPEAKER_02 12:15

I don't think I've noticed necessarily a a change from January, but I the first thing that popped into my mind when you're saying that is where they're where they're coming from or where they're starting. And it seems like there's this big gap between kind of what I was saying before, this idea of like, oh, AI is just gonna go fix it all. Like just turn on AI. And what we're missing here is is this kind of basic understanding of what it even is, because I think it's I think it's misrepresented online. Typically, the people who are really harnessing it are technical people. So they already know how to do software design. They're already very process oriented, right? They're very quick to adopt new technology and dig in and sit with the terminal if they need to. And they just make it seem like, well, you just spin it up and you go. And really what's missing is all the detail inside of even getting one singular job done. What's the assembly line? What happens? It's like building SOPs. What happens first, second, third, who needs to sign off? Where does the information come from? I find a lot of the work that I'm doing is process-oriented system design before we even get into AI. And sometimes we don't even need AI, we just need the information that's over there to go over there and be processed a certain way. So they're they're coming and they're saying, hey, listen, I need AI, please help with AI. And at first, I'm just building automation systems to just bring all the information into one place, just getting that solid. Now, on top of that, we can bring AI analysis, we can, you know, we can process it along the way, whatever need be. Uh, but I find that there's this this big gap between um, you know, people thinking that AI is just gonna take care of it. And really a lot of the work that I'm doing is what exactly? Okay, let's go one level deeper. Where does that information come from? Who makes that decision? And I'm mapping out their systems, at which point now, once we've got that, now we can say, ooh, step two through seven, that actually can be automated because we can use AI and automation to take care of that part. We'll still have this manual at the front, we'll still have this manual at the back to have the human gate, but everything else we can automate. But we've really got to drill down to what is actually happening. And a lot of business owners, even owners that are have have very big companies, are still struggling with that level of clarity.

SPEAKER_00 14:15

Yeah. So your experience has been that they're saying, hey, we need AI, but you're saying, whoa, whoa, whoa, before even yes, you do need AI. Yes. Whatever that looks like, whatever that means. And before we introduce, you know, the agent or the robot or whatever it is, let's take a look at why you're having these specific problems that were were pain points for you enough to reach out to me, the AI expert.

SPEAKER_02 14:40

Yes. And I would say I would I would forward that to say that you don't always necessarily need to keep the process exactly the same. And now is also a fantastic time to just say, should we be even doing this? Why are we doing it this way? At which point then you kind of take a fresh, a fresh iron to the whole thing.

SPEAKER_00 14:56

Yeah.

SPEAKER_02 14:56

And then we can also incorporate AI too.

SPEAKER_00 14:59

Yeah. So you you have the benefit of not walking in with the bias about, oh, this is just how we've always done things. You're able to say, why are you doing it that way? Right. And have that conversation. Um, let me ask you then, on on the client work that you're doing, um, of the processes that they so I would guess that clients are coming to you and they're saying, Hey, this part of our business sucks. We think AI could fix it. So they already have some idea, or are they going in, hey, we don't know what it can do, but we just know that we need it.

SPEAKER_02 15:27

It's it's two distinct places. So yeah, first of all, it's well, we need AI. Why? Yeah. Well, because we need AI and everybody needs AI, right? That first kind of, you know, which they may not have the answer for that, but knowing that. And then the second part of it is what they're actually coming to me with is business process, C-suite type of, hey, if I just could know this in my business, I could, I could get rid of that. Or hey, I'm spending a lot of time with this sort of thing. I'm analyzing, I'm me as the C-suite. I'm now going out to grab all this data for I'm doing all these research expositions. Can it can AI help with that? Or can can AI possibly they've got questions in and around uh thinking that AI can help, but it's more from the business side. It's explaining a business problem, at which point then I'm coming from it knowing all with all of my tech knowledge, and I'm trying to then design the system for whatever it is to solve.

SPEAKER_00 16:17

So, one of the things that I kind of think makes a good chief AI officer is somebody that has a business background, not necessarily somebody that has a tech or an even an AI background, right? Like I don't know that the machine learning expert is going to be able to come into the print shop and help them with process optimization that is really kind of the foundation of the introduction of AI. So you're able to do both. You have the tech background where you're able to you can think in your head like how to architect the technical side of the solution, but you're also bringing in enough understanding about business operations, having been an entrepreneur and running your own business to where you're sitting and kind of wearing both hats or sitting on both sides of the table. Is that accurate? That's right. Yeah, exactly.

SPEAKER_02 17:02

This idea of being able to, it's two distinct worlds, right? The tech expert or the AI expert. Sometimes, if we want to think about it like software developers back in the day, some software developers are visionaries and they can also code. Some software developers just say, tell me what to do and I'll do it. I'm not gonna think about the big picture, I'm not gonna think about how we're gonna make money and do the marketing and do it. Just tell me, tell me what to do and I'll get it done. And this is really where um I find that I'm different, where I can think, I can think kind of holding those two minds uh uh together. I can think from the business point of view, they don't want to talk about syntax and code and all this stuff. I've I've got a problem, I've got to solve it. And then over here, it's it's uh I'm not thinking about what's going on in the business. I'm thinking about that this is what's going on in the page and what's going on with the code. So, so give me give me my marching orders. And I I found that I'm I'm more of a visionary creative problem solver with then the skill sets of entrepreneurship and tech, at which point that we can bridge it together.

SPEAKER_00 18:02

So that sounds to me like an ideal kind of avatar for somebody to come in to help an SMB with, you know, because I primarily you're a one-man shop, or do you have like a whole dev team?

SPEAKER_02 18:12

Okay, so yeah, I've had freelancers to help me too, but yeah, pretty much myself. Yeah.

SPEAKER_00 18:16

So for the listeners out there, that's probably gonna be the the most accessible talent environment for you. If you go to the big companies, they're gonna charge you out the wazoo. And the reality is, I think if Eddie and I were to vet the people that they would send out, we'd be like, yeah, this person shouldn't be billing at that rate, right? Now, and then if you go to look for maybe even an established agency or whatever that's in your city, the odds are they're maxed out. Not many, there's not many um AI experts that have an agency that are on the bench. Most of them are like fully engaged. So for a lot of you, your opportunity to work with an AI expert is probably gonna be somebody local in Eddie's paradigm, exactly. And somebody that doesn't have the huge thing because there just hasn't been enough time really in generative AI for there to be these big agencies who organically have grown over the new stuff. So this is particularly relevant for any of you that are you know, you're you're you're looking, but you're not quite sure you want to go with the big consultancy, the big agency. My preference would be to work with a small team, right? We're a small team, Eddie's a small team. It's as a business owner making that decision, you're gonna want to be able to talk to the principal, not like if you got BCG come in, somebody sells you, and then that person hands it off to a junior. That's not the situation you want. Your business is too important, the opportunity cost of missing the window is too important for you to have that drawn-out long experience. So I I think we're we're very much aligned here, uh Eddie. So tell me about the do the business owners come to you already with uh some idea of uh problems across the company? Are they bringing like discrete, like, hey, I'm the CEO, this sucks, I can't get this date, I can't get optics on this information, I need the dashboard, or I need the automation. Um are they and and uh second question, second part of that question, are they coming to you where we've already given our people AI, now what?

SPEAKER_02 20:19

Yeah, that's a great question. So it's it's really both and, right? Sometimes the the you know, the the CEO themselves is saying, listen, I'm stuck, I'm frustrated, I need to get all this stuff done. I'm starting to harness AI, I'm not doing, you know, am I even doing it right? I want to be able to get this stuff off my plate. And a lot of the times it's it's systems for the business, at which point then they're pulling in a team member, they're pulling in somebody who owns that piece and part of it to say, to kind of answer those questions down in detail. Again, what happens is we start wide with like, hey, I think this is a problem. Over in this area, it's a problem. And then once I start asking, once I start peeling back those layers, they say, Oh, okay, let me pull in the person who's actually doing that, at which point we go deeper. So they know where these problems are. At least they have kind of this idea of it's bottlenecked over here, it's bottlenecked over here, it's bottlenecked over here. But then once we get into the deep detail, I very typically am then talking to somebody else inside the company that more readily owns that or can kind of answer for me, you know, down the way exactly how it's done or what's going on there. So they've got they've got kind of this top level 10, 20 things they want to do, at which point then we narrow in for, you know, I very typically follow the pattern of let's go wide, let's talk about it all for five minutes. And now, okay, let's pretend you can only build one thing and that's all I can do for you. What is it? Right? It forces them to focus down. Great question. Because we we've got to go, we've got to go very, very deep here. And this is where I think to your point about, you know, kind of passing it on to a junior or something, the understanding is really where the value is. The idea of being able to understand across the entrepreneurship side and the business side, the tech side, uh, are we should we even be doing this? Uh, and be able to kind of distill that down into, okay, here's the next step, here's what we're actually doing. It's not necessarily getting it built. It's the idea of, you know, if the blueprint is solid and you know the house is gonna stand, we can pass it to a lot of different builders. Yeah, they can take it. But but what's the plan? And how solid is the plan? And we would never say, you know, oh, just you know, put together a blueprint for me. I'll I'll pay to build the house, but like, uh just put put together a plan. No, we spend a lot of time in just how exactly should it be? Will it will it be sound? Will it be solid? Will it will it stand up in the wind? Is it gonna be to code? All of that stuff really matters when we're designing these AI systems. Um, and if we sh if we skip that step, then we end up with things being built. Or we're working on it. Oh wow, we got a lot of things going. And then later on we realize this isn't really helping the way we thought it would. And a lot of people are waking up to that on the other side of it.

SPEAKER_00 22:41

I think this is all uh pretty good advice here because you know, there's the I don't know if you know the stat with MIT's research at the end of the end of last year, 95% of pilots failing, and then the then a couple weeks later, Wharton comes out and says, No, actually, our research shows 74% are successful. Um and I think a big part of it was MIT was focused on enterprise, which is a mess, just dysfunction, in my opinion, my experience. Wharton was focused on SMB, where they're doing what like there's too much at stake, and there's too much, like it's very clear who's accountable. In enterprise, oh, it was the committee. Well, who's the committee, right? An SMB, it's oh, I'm the one that told the CEO we need to hire you, or I'm the CEO. So this is how I feed my family and my employees do the same. Like it really matters. So I think that the energy that they're putting into the planning and the clarity of things is um there's more at stake. So with that being said, what do you think uh a good first AI project should look like for somebody who's getting started with this?

SPEAKER_02 23:46

Well, that's great. Uh clarify the question. A good AI project for the CEO that wants to bring it into the company isn't sure where to start. For the client. Yeah. Yeah. Um, I mean, gosh, there's some things that I see over and over again. We want to look at the amount of time saved and who we're saving time for. We've got to decide are we saving the C-suite time or are we saving the workers time? And if we can save the C-suite even 10 minutes a week, 20 minutes a week, what can they do with that 20 minutes over the span of a year, right? So if we can save even just shave off a little bit of time for the C-suite, that's magnificent. And of course, we could save a lot more with AI. Um, but just starting there, or if we're gonna take a look at, hey, this type of task, I've got to hire an army of people to do it, and I want to scale my company two, three, four, five times, what do I have to do? Just hire that many more people, or can we have the same person be able to manage more accounts? At which point now we could scale the company uh without having to uh just uh put some manpower in. So coming up with systems uh to be able to do that. And so really I'm looking for either the highest, the highest people in the company are stuck, and if we can unlock them a little bit, they can do business development, and that has a huge burst effect for the organization. Or we've got a lot of repetitive stuff that is stealing our team away, or in some cases taking the management and pulling the management team down to have to do certain things. Um, it's we always want to play as as high of a possible level as we can. So if we're doing any uh any repetitive work that's stealing away our top talent, wasting our top talent's time, uh, we really want to start start there first. And so really it's taking a good look at, hey, where where are we starting in the organization? And then within that, it's it's we're playing 80-20 the whole way. I'm asking them about, you know, what are you doing this week? What are you doing over and over again? What are you doing today that you don't like? Maybe it's something that doesn't take a ton of time, but you really hate it. And so you delay it. And that's actually causing an issue for the for the business, you know, and how can we solve this with AI? So I very, I very typically like to take a it's essentially a research session. You know, tell me the tech expert where I'm building systems in my head. Tell me just what's going on. And my brain just, I'm sure your brain's the same. Your brain just starts to light up with like, well, we could do this, we could do this, we could do this, and helping them feed back to them. Well, hey, we could go these four directions. You know, what are you most excited about? Oh, I think let's do C. That sounds great, fantastic. And then we go down deep. So it's it's working with them to take the technical decisions and then serve them up as easy questions to answer, giving them this the control, but allowing me to do my processing to find uh those things. Make sense?

SPEAKER_00 26:16

It does. And it kind of so with any guest, we always do a pre-interview to get to know each other and talk about stuff. And one of the things that came up that you mentioned, and that I think really fits in here, and it's this idea that the project may not be failing because of the AI. It's a failing because that the process owner or the executive sponsor internally never fully explained the business process. And the the example that came up was the failed voice agent builds. Yeah. And I don't know if you remember telling me about this with the the business that was trying to build the two different voice agents.

SPEAKER_02 26:50

Yeah. Do you remember this one? Uh yes, I do. Yeah, this idea of you know they're trying to direct the team, here's what to do, and they're not being clear on here's the SOP of exactly how it should function. What's again, what's the plan? Go build the house. What kind of house do you want? I don't know. Just you should know. We're the AI expert, you go build it. No, no, no. No. What are we doing here, right? What does success look like? And we want to be able to essentially what I do is I go through the scenario as if I'm the user, as if I'm the one experiencing. I want to try to use my imagination and my vision to be able to call the phone. Okay, are we saying, hello, and here's the three things you can do? Are we saying, you know, this is a voice agent, I can help with this and this and this and this and this? Are we saying, welcome to this company? How can I help you? You know, even that it's so simple. It's like, well, just answer the phone and say anything. Sure, but this is a system we're trying to build. Are we trying to answer three different questions? Are we trying to send a text message? Are we trying to hook into an uh uh uh and pull information from a database and serve it up somehow? Uh, we want to be able to see what's going on. Because really, you know, really what it is is as you can give me all the Lego pieces, I can start to stack them. But if you don't give me any Lego pieces, I'm trying to stack something that I can't, I it's harder for me to build. And so as you can then uh uh uh throw at me, you know, we got to do this and this and this and this. I wanted to sell this and this and this. Here's what's worked before, here's what hasn't worked, here's you know, if I had a magic wand, this, you know, this would be perfect if you could do this. Great, throw it at me. And now I've got this playing field of, okay, we could stack this this way, and and what about that, and what about this? And I could do what I do and then serve it back to you. Go, is this right? You go, oh yeah, yeah, yeah, that's right. Except this one thing I forgot to tell you, there's some more context, and then we go down that road. So I'm I'm finding, yeah, I'm finding this information uh from them, and I've got to have it in order to build right. Um, and if if they don't, in this case, um they weren't giving enough context because they were assuming that the person would just figure it out. You know, they're kind of maybe a few softballs, a few lobs, you know, oh, change that or that needs to be different. And it would work one day, and then the next day it wouldn't work. And now they're just going, what am I doing? They felt like they're trying to push a rope and it just doesn't high doesn't work. Um getting clear to the plan, yeah, getting detailed works.

SPEAKER_00 29:11

I I think that there's probably some listeners here that have taken that exact approach. I don't know. You're the AI expert. You tell me. It's like, well, you want to know. And I yeah, exactly. And and it'd be great if I did, but the reality for for listeners is that the the part of AI that really matters, it isn't the sexy stuff. It's not the AI avatar that's it's do I have my processes identified? Have I mapped them? Have I mined and mapped them clearly? Right? So you get that stuff done, somebody like Eddie or me comes in, it gets very easy for us to introduce an AI solution effectively and quickly if you've got that like that question has been answered, right? Yeah so yeah, very important stuff. I don't I don't want the listeners to pass over that.

SPEAKER_02 29:59

Yeah, and I want I want to add something there. This idea of we would never we would never stumble over this if we were just hiring somebody. You would never hire somebody, bring them into the organization, and then say, Goodbye, you figure it out. You train them up. Here's the manual, here's the thing, go talk to this person for that. Here's an example. We're training that person to be able to mold and shape to the organization. And AI is really the same. AI can act all these different parts and play all these and do all these different things. It's amazing you can do all these things, but unless we constrain it, no, no, do it my way. No, no, the report needs to be two pages long. No, no, I need it to be. When we as we constrain it, it becomes more and more helpful to us in the same way that we constrain a new hire to say, I don't want a 15-page report, I want two, and I want it on my desk on Monday at 3 p.m. And it's gonna have these parts and these pieces. And here's five examples that I did the past week. Now go do it. Okay, great. I've got everything I need, fantastic. And then, don't forget, then when it gets delivered back, 80% of it is right or 70% of it is right. And this is where it's so important to have this revisionist cycle with AI design is we've got to throw the clay down and then we've got to use it. And when we use it, now we get to expose what information wasn't sent through, what's wrong, what's missing, and then we make it better, feed it back, use it again, feed it back, use it again, feed it back. And eventually we could say, hands off, it works the way we want it to. AI is magic, but it's because we did that processing and that planning.

SPEAKER_00 31:33

There's, you know, uh any anybody can call themselves an AI expert or uh AI consultant or whatever. And I know that there's there's some that want to come in and only work on strategy. There's some that want to come in and they only want to be reactive. Hey, if you've got a question, reach out to me. Uh, you know, I'll be the advisor or whatever, but they're not necessarily getting engaged in the discovery side of things. There's some AI folks that want to come in and just train your teams on better prompt strategies and how to, you know, uh use the tools regularly throughout their throughout their um their day-to-day. There's some that come in and don't talk about any of that and just say, hey, what do you want me to build? Um so I think that it's important for the listener to understand that uh best case scenario, you get somebody who's capable of maybe they're not an expert or the lead on all of that, but they're able to interact with either the internal experts or the external vendors that do have the strengths where they don't. And one of those areas where I think a lot of generative AI consultants are missing is the technical side. And with your background, that um that's certainly uh a distinct advantage for you, whether or not that's what the client needs to get started, knowing that you can take it in that direction. So, with that being said, I know when we spoke originally, you had mentioned that you were involved in placing offshore and nearshore talent, uh AI development talent. Is that still something that's uh in your wheelhouse?

SPEAKER_02 32:59

Um yes, that's changing a little bit now. But yes, this idea of of being kind of the the thought and the visionary on the way to, hey, this is going to be your resource now, um, is something that I did for uh about a year. Um, this idea is as as businesses want to bring in uh this talent into their business. The the first question after that is, well, what am I building? What do you want me to do? Back to that thing we initially talked about with if we can give good instructions, they're off to the races and they're running, at which point then they're getting in, you know, entwined in the organization and moving forward. I provide that springboard, that jumping off point of, hey, what do you want to do? And then meeting with me first to really design out that blueprint, then allows that's that talent to then jump in and be able to move, as opposed to starting with, huh, what are we doing first? And having to do that discovery, at which point a lot of tech talent, to your point, tech talent is not the same. You've got people that are decision, uh uh uh uh that are uh decision takers, you know, they they'll take the the the the words and move forward. And there's some people who are visionary, and and if you don't know, if you can't parse between them, um, then you might end up with somebody who needs you to tell them what to do. And if you don't understand AI or don't know what to do, then you're both kind of blind leading the blind, and that can really uh result in in some wasted time and some some wasted effort.

SPEAKER_00 34:15

Okay, well, with that, like what would be some things based on your experience of getting these people placed or helping uh clients vet that that like more full-time talent for their development stuff? What are some things that a she a CEO should never delegate to the AI developer? Ooh, I like that question.

SPEAKER_02 34:34

Um, I'm coming back to business process, I'm coming back to that domain expertise of, you know, hey, we need to do this this way because of this reason, or we we've got to make sure, like giving the boundaries to the developer for, you know, here's what we've got to do this very specific way, or here's where we've got to check what's going on. Um, if we just kind of pass off a whole process, but it's very important for us to double check this or double check that, they're gonna have more domain expertise on I'm making this up, you know, step seven, we have to double check that with an internal committee and make sure that's all good before going forward. We want to make sure we've got kind of checks and balances inside these systems, as opposed to kind of just uh hoisting this whole thing off onto tech talent and then saying you figure it out. And I put a caveat there that maybe instead of that framework, the idea of having this back and forth that the to kind of keep expanding on the idea of, you know, hey, don't please don't say you go figure it out. Let's do the opposite, right? And then and and not just let's do the opposite, let's let's distill down what's actually happening into a framework that a non-technical CEO can understand. And so this is where I was actually just helping somebody with this this this past week, where they've got talent and the talent isn't reporting back well, isn't it, or when they are reporting back, it's just it's just tech information. It's just tons and tons of information. It's not speaking to the business owner in business speak, it's the developer saying, here's all my reports and all the stuff I'm working on. And so uh one thing that I can offer there is we could use AI to translate that by asking that developer to basically speak out kind of what they're doing, but then putting that through a prompt of now you're speaking to C-suite. Yeah. Translate this into how it actually impacts the business. Now the developer can just speak the language that they speak, but then the C-suite can receive oh. Yes. That makes sense to me, versus, I don't know what you're doing back there in the tech department, but just I guess keep doing it, right? We want to be able to help there be such a good communication flow because the CEO can spot things that the developer won't, because they've got such this fantastic knowledge of what's going on in the business. The more throughput we can have, the more translation we can have back and forth uh is really gonna help us have uh better success there.

SPEAKER_00 36:54

So for the listeners, I want you to really think about what he just said. If if you're working with these, well, I don't, I've never worked with developers before, or that's usually somebody else's job, I don't know what to ask them. Hey, ChatGPT, what should I ask the developer? So great advice. And that's something that we remind people of. Like ask the models. If you get stuck, say, hey ChatGPT, I'm stuck. Hey Claude, what should I do here?

SPEAKER_02 37:18

I want to add something very important here, and this is where people, you know, if if anybody ever walks away from from AI saying, oh, it's stupid, or it didn't really do that good of a job. I want to say that that it's all about the context. So, what I mean by this is if I gave you, if I asked you to to to to create a world-class meal for me, you know, you're a chef, make a world-class meal, and all I gave you was an apple, you're gonna have our what are where's my ingredients? That's all you gave me is one apple. I can slice it for you, right? As soon as I give you a whole grocery shop, a grocery uh shop full of ingredients, pots and pans, whatever you need. Now I'm gonna, ooh, what can we make from this? So if you're going to AI saying, tell me what to ask, please also speak for the next three to five minutes all about your business, all about where you're stuck, all about what you're aiming for, all about your dreams and your ambitions and where you what you've tried before. The more you drop on it, it can parse through and give you that world-class meal. But if you just say, uh, give me five questions to ask AI talent, it'll do it. It'll say, here you go. But it won't be, it won't be tailored to you. And this is real, we want to start thinking about like custom fit suit. The only reason why you can get a custom fit suit is because the tailor knows your measurements. Without your measurements, you're just buying it from Walmart. Yeah, it's a suit. Does it look good? Not really, right? So the idea here is as we give it more context and we start dropping, I'd say, three to five minutes of dictation. Here's everything that's going on. Now we start to get some feedback and go back and forth with it too. Give it even more context, and you'll get an even better answer.

SPEAKER_00 38:55

So I think one of the suggestions Eddie just made was uh very important. Um when we're working with clients, I'm surprised how few that say, Oh yeah, I use AI all the time, that are not using the dictate feature. And Eddie's mentioned a couple times, he's like, tell it, you know, spend three to five minutes telling it. He just mentioned, like, use dictation. If you're um uh having trouble with like well, what what context do I give AI before I talk to the developer, before I get clear on stuff? The reality is you you turn that little microphone thing on uh and the input box and ramble. Well, here's our company, and oh, but we've got this, oh, and like there's ahs and there's ums and there's pauses. That's okay. That AI is smart enough to extract all that out and get the primary context. But if you're trying to type all this out, you're going to uh limit what's possible. So great tip from Eddie there. If you're not using dictate to share context on really any bigger issue, uh it's a big miss, you're gonna love it, it'll save you a ton of time. Yeah, any other little uh usage tips, tricks, hacks, things that you like? And before we do that, what's your primary model that you're using?

SPEAKER_02 40:06

Oh, love that. It's definitely changed. I mean, I typically I'm going to Claude Fable now just because it's the newest thing on the block and I'm excited about it. Um but I use Codecs. So when I'm coding, I use Codecs. Um I played with Cloud Code for uh for a little while, and really I like the stacking feature in Codecs. Maybe Cloud Code's changed this by now, but I want to be able to speak to my code and I want to be able to stack my requests. I click on the microphone, I open up the app, change this, change that, that's got to be modified, that's got to be modified, go. And then I've got another idea. While it's working, click the microphone again. Okay, change that, change that. I forgot this, go about this other thing, send. And so it's just cueing the things for it to do. And so I can just move really, really fast versus cloud code. I felt stuck because when it was working, I just kind of had to wait for it to be done and I didn't want to wait for it to be done. In some cases, I'll stack five of them and then I'll go eat lunch and then it chews on, it chews on and goes there. But I love that. And then um, but to answer your other question about, you know, some other tips, one thing I've been loving doing lately is calling a round table in AI, right? Instead of asking for one answer, I want you to pull from all your different knowledge sources and then bring, let's I want to hear from everybody. And the fact of the matter is that if you don't ask AI to do this, it's trying its best to give you an answer. It won't typically disagree with itself. Hey, do this and don't do this. No, no. But if you ask for a round table and you also say, hey, it's okay for people to disagree, you get to see, and now we can do what we do best, which is ooh, I like that, I like this, and I like that. And then we can we can be able to pass back to the model. Here's where I'm at. Instead of having to come up with it, we say, fan out for me. Give me, just let me let me pick a card, let me pick the cards that I want. But you fan out. Oh, we got this one and this and this and this, and then I go, ooh, 247. Great. And then we're moving on forward from there. And then one more thing I'd say too is to your point about going to AI and saying, you know, I don't know what I don't know, I'm not sure what to ask, whatever it is, drop the context and then say, please give me five questions I should be answering for you where I didn't give you enough context, AI. And then it will come up with questions that then you can just answer, and now you've given it everything it needs to then be able to form a great solution for you.

SPEAKER_00 42:18

So, for those of you listening, if you're not doing this type of thing, this is this is how an expert uses the tools. What I find is that a lot of uh non-technical business professionals who are using AI, they treat it like a drive-up window. Hey, I need this answer. They get the answer and they leave. What we're talking about here, the way that that Eddie's suggesting that you work with this, there is an ongoing dialogue that's happening. And I would imagine that Eddie, if we were to look at your your chat history, it's not a ton of threads, but it's threads that go very deep.

SPEAKER_02 42:50

Very deep. Molding the clay. Yes, we're molding the clay. It's never a one-shot, and we I mean, maybe if it's something I've done before and I've already trained it, but this idea of I can't give it all the context every single time. I've got to mold it and I keep going from there. Eventually we hit a point where, yep, that's good. I'm good enough, we're done. But almost typically never one. It's it's uh, you know, I'm I'm trying my best to to give it everything possible, but very often it's you know, it's it's tweaking and changing and molding and shaping.

SPEAKER_00 43:19

And as of this recording, um, for those of you that are using Chat GPT, you're like, whoa, codex, I don't know anything about that. Now, if you download the ChatGPT desktop app, it's codecs. Like they changed the name of Codex. I think the app is now called just ChatGPT Desktop, is that right?

SPEAKER_02 43:37

Yep, they put it together, and so now it's just they're trying to make it as much as accessible to people as possible. And I would say too, even if you're non-technical, using codecs is kind of cool. You can kind of keep all your files in a file folder, you can ask it questions of files on your own hard drive as opposed to having to upload them to Chat GPT. You just say, here, this folder. Um, you know, be careful that you're not deleting anything. Sometimes AI will do things outside the bounds of what you want. Uh, but I've had I've had really good success with it lately. And and the idea of of being able to write reports or go through lots of files at once. Um, you can use codecs even if you're not a software developer. And I feel like that's a big aha.

SPEAKER_00 44:15

Yeah. Yeah. And that that's basically I had downloaded Codecs, didn't really like I the same thing that the that the new business owner is like, well, I don't know where to use it. That's how I am with Codecs, because I'm not that I'm not a developer, right? But when it turned into the ChatGPT desktop, now I'm finding myself, I'm still using Chat GPT as the thought partner and the some cases the drive-up window, yes. But I see all this other functionality with codecs, and it's kind of it's opening the door for me to experiment and try with this um with the client. Now, you also mentioned that you uh have been a big fan of Fable Five. What are you noticing different with Fable? And for those listening, Claude's most powerful version is is called Fable. Um, they released it, pulled it back, re-released it. So that's kind of where we are with that. What are you finding different about Fable Five? Anything?

SPEAKER_02 45:04

Yeah, that was a whole saga uh with them giving it and then taking it away and now it's back. Yeah. Um, this this idea of you know rating models to each other, I find it really hard, especially if we're talking about um kind of expansive concepts or like thought, thought product, thought work. Um, if it's if it's like code, you know, when I've used dumb models, yeah, it didn't get it right, it did something. Wrong. I made a mistake. It missed my context. It didn't do everything I asked. You know, stuff like that where it's like, hey, you you missed something. It's very obviously there's a hole. But with the idea of, you know, hey, I want to, I want to try to do this new thing in my business and let's let's ideate together. Um, you know, it's it's hard for me to say, you know, well, you know, five wasn't, you know, this is so much better than GPT-5 because of this, this, and this. I haven't taken the same pro thought process and put it into all the different models. Now that said, I was trying to do something the other day where um it was kind of a big task. It was something that I needed a lot of thought for. So I'm going to Claude, I'm going to Fable, I'm cranking it up and just saying, give me, you know, give me the best processing you have. I ended up actually doing both models and then I intertwined them. So I gave, I had my code base in Codex, yeah, and then I had Claude by itself. And so I kind of brought my code or the idea of my code to Claude so it could have some context and drop the the prompt. And then I put that same prompt in in uh in codex, chat GPT, and I let them both chew on it. And then I ended up just grabbing Claude's and then bringing it to Codex and saying, let's intersperse these. So, you know, again, I I I've not, I can't point to a specific thing where it's like, hey, it's bad this way and hey, it's good that way. Um, in the same way that if you hired somebody, you know, hired you hired a consultant, well, like, how much better are they than the other person? Well, I don't know. I didn't work with them for four months on this project. And you know, you know, but the idea of uh being able to to use both tools and and again, let's talk about the user of the tool. If the if it's actually a tool, you know, is the calculator dumb or smart? Uh who's typing in, right? Is is is the knife, is the knife uh uh gonna make me a world-class meal? Who's using the knife? Right. So um, yes, the model can be smarter, but really what's happening now is the models are getting so good that the main differentiation you have is play with it and get better with it and be better yourself with it, at which point you get more out of it, you harness more power out of it.

SPEAKER_00 47:28

Yep. So uh one of the tips that Eddie just gave us here about comparing the models against each other or collaborating with the different models. So for those of you that have the Chat GPT account, maybe you've got a clawed account, vice versa, whatever, um it doesn't, it's not just for code, it's also for the the thought partnership activities that you're doing or the problem solving that you're doing. Um if you're not doing that now, especially maybe on the light stuff you don't do it, but on the stuff that's that where hey, this this decision really matters for the organization or for our growth opportunities or whatever, run it in both models and start to see, and then take the best of what ChatGPT gives you and run it in the cloud thread and have them expand on it and vice versa. That's great advice. So, Eddie, um, are you producing any type of like are you are you uh sharing your thoughts or experiences anywhere on any social platforms or anything like that?

SPEAKER_02 48:19

Yeah, I've got an Instagram and I'm posting videos there. I just got into Substack, uh, so I'm doing that. Are you do you do Substack? I don't. I'm I'm thinking I've heard a lot of good stuff about it. It seems very organic. A la, you know, Facebook back in the day where you could post and people would actually see it. Uh so I'm excited there, but I've got my own blog and Instagram and uh some on TikTok as well.

SPEAKER_00 48:38

So um my uh media team is gonna add all these links into the show notes. And if you're curious, if if like you've appreciated this kind of like boots on the ground conversation and approach that Eddie's bringing to things, um I would encourage you to, it's it's obvious that like he's not holding back, he's sharing everything here. I would I would encourage you to uh keep up with his perspectives on those platforms, Substack and Instagram and the blog. Um and so I guess any uh final advice, Eddie, for um somebody who is they know they gotta look, man, we gotta do something. There's pressure. Uh I I see other competitors using this stuff. We've got some people internally doing some cool stuff, but I don't quite know how to harness it or direct it. What would be the advice that you would give them to finally like get in motion with some clarity and confidence that they were heading the right direction with their AI strategy or AI efforts?

SPEAKER_02 49:32

Yeah, yeah. A lot of different directions I could go with this. I mean, I the thing that keeps coming back to me is you got to get on the bike and ride it. You know, back in the day, before AI, it wouldn't make sense for you to become a software developer. You hire a software developer. Sure. But now with AI, this is a tool that we can all use. And I would say that in the C-suite, you're gonna use it differently than how a developer is gonna use it, right? So you've got the things that you're chewing on, that the things are occupying your brain. We could still take those to AI, even though we're not creating code or creating software. You could do just a lot of thought leadership stuff there. And so the idea of playing allows you to open your mind, it expands your mind where you're you're using it yourself and seeing what works and doesn't work. And so then you're not just taking somebody else's advice where you read a post and go, I have no idea how they did that. You get you get uh into it enough to go, okay, I get it. And then when you hire talent, you're able to, you're able to just do so much more because you can ideate in new directions. And let me also let me also clarify that uh we know this tool isn't going away, right? We're gonna wake up three years, five years, ten years, AI is still gonna be here. So just get your feet wet, get in, it's just start using it, even if you don't know what you're doing. And anytime you're stuck, just ask. I'm stuck, I don't know what to do. If in case you had a consultant right there, just ask and um you'll you'll figure it out.

SPEAKER_00 50:52

Awesome. So again, especially if you're in Nashville, I would encourage you to reach out to Eddie and uh see how his solutions can definitely kind of take your company from where you are to where you should be in the AI space. So, Eddie, thank you so much for taking the time today to share that. And I uh got a couple of points of distinction here that I'm gonna introduce to the rest of our chief AI officers from that. That's awesome. Love it, Eddie. It's such a pleasure. Awesome. And uh for those of you listening, thank you so much for um uh being uh, you know, a fan and a supporter of the podcast. If there's anybody out there that you know who is on this journey that could benefit from this episode or any episode, we'd love it if you just you know sent them the link to the podcast and let them plug in and start to uh learn how to use AI at work with us. So thanks everybody. We'll see you next week with another episode of the show. Thanks for tuning in to using AI at work. Don't forget to subscribe for more conversations about how to use AI at work. And a special thank you to our sponsor, Chief AI Officer, for empowering businesses with AI education and training. Visit their website for a free AI readiness assessment and AI strategy guide to help you get started using AI at work. That's www.chiefaiofficer.com. Follow us on Twitter at the handle UsingAI at work and visit www.usingai at work.com for free resources to help you harness AI in your role.