I’m Nate Troyer. I am an Engineer and Account Executive at Corsica Technologies. And I’m here today talking with Wes DeKoninck. You almost had without saying it. Yeah. I can see it in your eyes. I’ll let you take it. Sure. I’m Wes DeKoninck. I’m the Director of Digital Transformation here at Corsica. Yeah. And we’re here to talk more about AI. So, you know, we have Wes here today, to kinda discuss how you really get started down the road of, implementing AI in your organization because AI is a buzzword, and it’s pretty cool. Yep. Now, and we’ve had some other podcasts discussing, you know, products like Copilot or the difference between, like, a ChatGPT and Copilot and, what that can offer, like, at least at the executive level, what it can what can be offered there. But, like, if I’m an executive, I’m like, alright. Well, so it makes, you know, administrative work easier. But how do I really, like, start implementing that? I mean, do I- Do I buy new? Do I let a guy like Nate, build something from scratch? Yeah. So where does that entry point really start? Yeah. And it’s it’s a really common question that I’m getting from clients when they say cool. AI is everywhere. Microsoft has AI. You know, do I go buy the thing? Do I just start using it? Do I hire someone to tell me how to use it, you know, how do we really get started with this? And I think buying something and trying or just building something and trying are probably not the right approaches. I think trying, then buying is the right approach. So in general, I try to take a framework approach to it where you wanna look for your AI people in your business. So your AI evangelists are out there in your business. Your early adopter type of individual. Those, people exist around your organization, and they’re all over it, you need to go find them, and you need to pull them together, and then you need to have a conversation because there are free tools available, and you set them to task and say, hey, guys. Take these free tools. Here are your guardrails, which you and Brian have talked about, you know, in other areas. Please don’t put you know, personal information or things like about our clients into ChatGPT. Exactly. So but there’s there’s a structured way to go about testing out these tools without putting your secret information out there, you know, to make it part of the public domain. So you find these evangelists, you set them to task with, hey, explore take what you do every day, find the mundane task, interact with these tools. Here are some possible use cases, but explore and experiment. So find the individuals let them experiment and then analyze those findings. And then you kinda come up with a list of things that, hey, these things were actually impacted positively by AI. And again, at this point, there’s no cost because you’re using free tools. Right? Yeah. Yeah. So you refine those use cases. And once you have that, you actually have people to evangelize AI. So you’re gonna get good adoption of the tool, and people will use it and get the benefits. You have no cost virtually other than time and experimentation. And you have a plan. That’s something you can take to your primary stakeholders or your business owners and say, Hey, here is the business case for AI. And here are the costs involved. Here are the tools that we’re using. What do you guys think? Yeah. Rather than just spending the money and hoping it works out in the end. Yeah. This is where you don’t want someone like me, you know, like, like running the show because- Right. I’ve never met an engineer who could well, I guess I have. I’ve met a couple, but they’re like unicorns that can sell the business use case for something. They’re usually thinking, like, man, this makes my life easier, and we need it. But I can’t really articulate it past that. So Yeah. And it’s very specific to what they’re doing. You have to think this tool has broad use cases in every organization. Yep. Whether you’re trying to draft an email or a document, marketing material, need help with coding, like across the organization- creating a PowerShell script- Creating a PowerShell script. So there are so many things that you can do with it. The problem is because there’s so many things, because there’s so many tools, and there’s a lot of newness to this whole thing, people just don’t know where to begin. And really, you have to just begin. Yep. You try some things. You get in there. You see what works, and what doesn’t work. And you start to write that stuff down and talk about it with the people at your organization, and that’s how you’re gonna ultimately have success. And you can do all that prior to investing in any technology. So where so, you know, where does where does the what are the examples that you can give of some of the, of some of the integration with AI that that is like the low hanging fruit. The what the the the where you can just start using AI and try it. Which ones have you come across? So a lot of the use cases are mostly with content generation that we’re seeing. So, I mean, that’s really the big buzz that’s out there right now. It’s right. Oh, it’ll help you write an email and create a proposal. It’ll help you analyze documents and things like that. So, you know, getting it to understand what you need to interact with from a data standpoint is kind of difficult if you really wanna dive deep, but that’s what I would view as a later stage in AI adoption. Mhmm. The early stage is, help me make this document better. Or I’m writing an email to the CEO about a thing. I don’t want it to sound like me. I want it to sound professional. Can you help me with this? So they’re using it to modify content that they’re generating. They’re using it to do things like this podcast. Like, if they wanted to start something like this, they can lower the barrier of entry by saying, give me some topics that we can talk about because I’m having writer’s block, or I’m not super creative. So they’re helping people with creativity. So, you know, I’d say clerical tasks, document generation, you know, creative tasks, like marketing design, things like that. It’s really helping those types of individuals without any additional effort. Beyond those use cases, you’re looking to help me code better, help me solve this problem, help me understand this data. That will require a deeper level of integration because the AI needs the context of the information. It needs to be grounded in what details and data you wanna talk to. So what is the what is, like, from an engineering standpoint? What is the engineer’s involvement in this is there any, like, training that has to happen to the AI? Is there any coaching that you need to do- Yeah. Or is that like an organizational-wide thing if you’re if you’re bringing in people from different parts of the org, can they all participate in training the AI? Yeah. They all can, and they all can have their own use cases and experiments. If you’re really wanting to go beyond that, you will need engineer involvement, or again, it’s just gonna depend on your data set. So if you’re talking about all of the data that you have inside of Microsoft What’s a data set? What’s a data set, types of data? So where so, you know, where does where does the what are the examples that you can give of some of the, of some of the integration with AI that that is like the low hanging fruit. The what the the the where you can just start using AI and try it. Which ones have you come across? So a lot of the use cases are mostly with content generation that we’re seeing. So, I mean, that’s really the big buzz that’s out there right now. It’s right. Oh, it’ll help you write an email and create a proposal. It’ll help you analyze documents and things like that. So, you know, getting it to understand what you need to interact with from a data standpoint is kind of difficult if you really wanna dive deep, but that’s what I would view as a later stage in AI adoption. Mhmm. The early stages help me make this document better. Or I’m writing an email to the CEO about a thing. I don’t want it to sound like me. I want it to sound professional. Can you help me with this? So they’re they’re using it to modify content that they’re generating. They’re using it to do things like this podcast. Like, if they wanted to start something like this, they could lower the barrier of entry by saying, give me some topics that we can talk about because I’m having a writer’s blocker. I’m not super creative. So they’re helping people with creativity. So, you know, I’d say clerical tasks, document generation, you know, creative tasks, like marketing design, things like that. It’s really helping those types of individuals without any additional effort. Beyond those use cases, you’re looking to help me code better, help me solve this problem, help me understand this data, That will require will require a deeper level of integration because the AI needs the context of the information. It needs to be grounded in what details and data you wanna talk to. So what is the what is, like, the from an engineering standpoint? What is the engineer’s involvement in this is there any, like, training that has to happen to the AI? Is there any coaching that you need to do Yeah. Or is that like an organizational-wide thing if you’re if you’re bringing in people from different parts of the org, can they all participate in training the AI. Yeah. They all can, and they all can have their use cases and experiments. If you’re really wanting to go beyond that, you will need engineer involvement, or again, it’s just gonna depend on your data set. So if you’re talking about all of the data that you have inside of Microsoft What’s a data set? What’s a data set, types of data? So, you know, let’s think customer data, client, invoices, you know, agreements, systems that they use, the types of different programs that they have, they have data. So data set would be a type of of data. Okay. So you’re given the you’re given the AI context. Yeah. Exactly. You’re saying go, go use that information and find out this thing about it. Okay. You’re telling it what to do and where to go, and it does it. Now, if it doesn’t have access, and again, for this trial and error type of thing, it’s not gonna have access to the secret sauce of your company. So it may not be able to perform those in-depth tasks. With Microsoft Copilot for three sixty-five, you can turn that on for select users, and it will have your tenant-wide data. But again, guardrails, make sure it’s secure. I got my license, and it’s so awesome. It is cool. It’s great. I’ve used it for various things that I didn’t even think I would. I’ve used it mostly to ask, like, Can you please give me the HR documentation on what is allowed to be said in the office? I use that all the time. All the time. Yes. And it shows. We’ve seen a drastic improvement. Yes. We really have. Anyway, so, from, like, an advanced standpoint, when you start when you start getting, getting past the, oh, it helps me make the document better or or whatever. Make those menial tasks, you know, a little easier, to you know, execute every day. Right. Are we seeing companies adopt the, AI in a form of like, like, reporting to use it in reporting and things like that? Yeah. A lot of them are, but these are larger organizations usually because, you know if we come back to what is the engineer’s role here? It’s about taking that data and getting it into a system that the AI can interact with. So that does require integration of sorts that has to allow the AI to say, hey, you can go talk to that system now. So when I ask you a question about my customers, tell me who my most active customer is. It needs to know what’s a customer. Where do I find data about this customer? So you have to train the AI through the use of config duration and other tools, to say, when I say I wanna know something about a customer, you need to go there to look for it. Don’t just start searching the web, I ask about a customer, go to my CRM. When I ask about invoicing go to my finance system. When I ask about analysis for profit and loss, I want you to go to my other account system. So you have to tell the, yeah, when when you understand that I want this thing, and this is what this means, here’s where you go to get your data to be able to pass that to me. Interesting. So that could so, you know, AI is not really doing magic. Right? Well, the AI we’re talking about is not really doing a lot of magic. It’s it’s finding similarities between data points. Right? Like, it’s making connections. Yes. Absolutely. But those connections can be pretty powerful. Right? Oh, absolutely. They can help you plan out your company’s, you know, next quarter, two quarters, next year, you know, about what you wanna do. So what has your team been, like, been doing with AI recently? You can give me some sanitized examples. So, we have been using it a lot for coding to expedite what we’re doing from a code-based perspective. Right? So it’s like, you know, I could spend all day troubleshooting this issue, but if I say, hey, you know, Get Hub Copilot, here’s my code, here’s my application. Here’s what it’s supposed to do. Why doesn’t this work? You know? Yeah. And it could say, well, this is supposed to be doing this, and you have a problem right here, and you’re like, oh, I didn’t even think about that. Or, or I’m stuck, can you tell me about this plugin or the best thing I should be using for this? It helps expedite that. It’s not solving the problem for you. It’s not coding for you. It provides you with the path forward. It’s helping you get unstuck and move forward. So You still have to know. Like, I mean, six months ago, I had I think, somebody from our service desk, you know, ask me for a Powershell script to do x because he thought I had it. Like, and I just you know, five minutes with Chad GPT, asked it a couple of questions. I gave it some guardrails. I kept scoping in what I needed it to do. And I handed him a PowerShell script. He’s like, that’s amazing. Like, how did you do that so quickly? And you just point your head. Right? That’s it. I said it was all me. It’s all me, buddy. Yeah. No. No. I said, that actually is ChatGPT. You still have to know what you’re trying to accomplish and you try to know what, what solution it’s providing, and maybe what outcome it’s providing too. So you have to test everything that it gives you, and you can’t just blindly follow. But it was remarkably accurate. Yeah. You know, it was like eighty-five, ninety percent of the way there. Right. And all I had to do was fill in the last ten percent. So, you know, it it’s it’s it’s fun to use it in those ways. Yes. Because you’re like, man. So I just cut out two hours of time, you know, writing this PowerShell script, dealing with ice, you know, this is just from an engineering standpoint. Right? Right. Imagine what it could do for someone in the c suite who essentially needs to have, like bullet points provided to them for what their next steps are on an email chain that’s like three miles long. Right? Yeah. That’s great. Or say if you’re, like, in a law firm or something like that and you’re doing, you know, a case prep, Right? I mean, if you had all that data and a data lake somewhere like in m three sixty-five and you ask Copilot, give me, you know, give me the high points of this case, you know, that I need to be, you know, focusing on and let it do that. Yeah. That’s incredible. And then you go verify and follow on with that information. Yeah. Yeah. So expediting those tasks, that’s one of the things that we’re using it for. In code or things like that for all of our systems and internally and for clients. Another thing we’re working on is just interacting with sentiments. So trying to understand, like, clients say, I, hey, I need the thing. I wanna do a thing. Creating very specific chatbots that are backed by these. And this is one of the biggest benefits I would say of the large, large language model AI is just giving you the ability to talk to computers as a per where before AI kind of blew up, it was I needed an engineer to talk to computers and data. I told the engineer what my outcome was, Right? Here’s what I want to achieve. And engineers, like, let me think about that. Okay. Here’s how I’ll go about doing that. They’ll build an application. They’ll write a query. They’ll do a thing. And they’ll come back with the data you need. Right? Well, the big benefit of AI in the large language model era is, hey, computer. I wanna do this. Can you figure this out for me? Yeah. And the computer goes and goes, I understand what you wanna do, and here’s the data I need. I’m gonna figure this out for you. So it’s giving more individuals the ability to have that conversation with data and technology through the use of this model that just can understand humans. So that’s one of the biggest benefits of it. It’s kind of abstract, but it’s really the power that lies behind this, not the fact that I can go get a report or something that could have been done. I could have done that through an engineer, but it gives virtually anyone access, and the ability to talk to that computer. The computer just needs to know what its rules are. Yeah. It’s a really interesting concept. Really, it’s kind of nebulous because you don’t know what you can do with AI until you start trying until you do it with AI. Yeah. Until you do it with AI, and you’re like, oh my goodness. I I prepped for a a meeting the other day. And asked a couple of questions about the cultural background that one of the people I was talking to was gonna be a part was a part of. And it’s a little different than the Western American culture. And I said, like, what, how, you know, just give me some ideas about how the, how, you know, where their mindset be, and, and, you know, like, what they value, and, like, it was, it was interesting the information it spat out to me. Yeah. And You know, it it worked. I, you know, I kind of carried myself a little differently in that conversation. Given that context. Yeah. Given that context. So, Yeah. It’s good. Yeah. It’s a it’s an awesome, space to be working in. I appreciate you having this conversation with us. I am really interested to see, what a conversation with you would look like in about six months down the road. And see and check in with you to see about more. Yeah. We’re we’re just crack music’s open because again, you can do so much with it. It just becomes what can I do with it? What should I do with it, what makes the most sense? And again, just like most people out there, this is new. Right? This is new technology. They don’t know. Well, It’s new for us too in the technology industry. We’re just seeing how we can interact with it and what makes sense because you can do things with it. That doesn’t mean you should. You know, it’s not the right answer for everything, but I think that’s where the power is because we can do so many things, we can try so many things. Right? Yeah. And so we’re just getting the going. And I think the important thing is, as well, it’s great to experiment and try to understand how it benefits your organization, if you do it responsibly, it can have a ridiculous impact. Very quick You know? So having guardrails, having guidance is important in this process. And if you don’t feel like, you know, what you’re doing, get a consultant you know, we’re here. Have a conversation with people who understand technology, understand businesses, and how to marry those two things together. And we want businesses to thrive. Absolutely. Yep. So Yeah. That’s all we got. Thank you for watching Unraveling IT.
