Navigating Business & IT Evolution as an Award-Winning Company with Frank DeGeorge, CTO at Impact Networking

(Intro Narrator at 00:00:00) Today, we're talking to Frank DeGeorge, CTO at Impact Networking, about how he's evolving his team and company as a leader. You're listening to Joel Beasley, Modern CTO.

(Joel Beasley at 00:00:17) I'm curious. I was stalking you a little bit to see what's going on in Frank's life. So you guys seem to win amazing awards constantly for being a great place to work. That's a trend. Tell me about that.

(Frank DeGeorge at 00:00:31) You know, it's crazy because I was just telling the story to some of our new hires that started. I've been with the company 18 years almost, which to me is way too long if from a timeline standpoint—it seems I'm not even old enough to be that person—and then way too fast from a reality standpoint. A blink of an eye. We've gone from 35 employees when I started to nearly a thousand, and we try really hard to make it a great place to be. You know, everything from our ownership to the top down, we do little things, we do big things, we do things that are individual. There's a lot of continuing education going on right now with some of our employees.

(Frank DeGeorge at 00:01:12) I went through a class at Kellogg last year to kind of just level up my knowledge and try to stay on top of things. But yeah, it's great. We just got our Best Places to Work in Chicago. We're one of the fastest growing companies in Chicago. You can't have a great company without great people, and the bigger you get, it becomes harder to maintain that culture, but you've got to work at it.

(Frank DeGeorge at 00:01:39) And everything, like I said, from our owner, Frank, on down, we really try and be communicative and clear, transparent, and give people the state of the union as it is.

(Joel Beasley at 00:01:53) Well, that's why I enjoyed you guys. We did the first show together just off podcast, then I got invited to speak at the Navy Pier event. I think I got to meet Frank and maybe a son, or I think there were some family members around the business, and then you, of course. And it was good times. That's a pleasant memory in my mind of getting to interact with you guys.

(Joel Beasley at 00:02:09) It was that that's a pleasant memory in my mind of getting to interact with you guys.

(Frank DeGeorge at 00:02:13) That was a great event, and your presentation was well arranged. So we enjoyed that. We enjoyed you coming down. I'm glad we started that relationship, but that worked out.

(Frank DeGeorge at 00:02:24) And Frank's son, I don't know if he was in the business then, but he's been in a few different roles. One of his sons, Anthony, and so him and I are close. And he's going through right now and learning different parts of the business and bringing some experience, but also bringing some white belt mindset. Right? Coming in and trying to learn things how they are and figure out how we can make things easier and better for everybody.

(Joel Beasley at 00:02:46) Yeah. Fresh perspective is always great. Yeah. So tell me a little bit about where you're spending your time today.

(Frank DeGeorge at 00:02:53) Well, I'll give you some basic numbers. So we were—I don't know if we have our final numbers from last year. Now we're about just at $200 million. If we didn't hit it, we'll be just shy of it. And so about 900 people.

(Frank DeGeorge at 00:03:08) And when we talk about digital transformation and just transforming a company, a lot of times it's focused on the tech, and you and I have talked about this. And it's also about the business transformation that your company goes through. So not only are we growing 15 to 25% a year, not only are we embarking on trying to modernize our systems and keep up with the changes, but our core business has shifted in that same concept. And so what I mean by that, in the last two years, IT went from being about $50 million in revenue to being just over $105, $110 last year. And so you pair that with—

(Joel Beasley at 00:03:49) That's crazy.

(Frank DeGeorge at 00:03:49) Yeah. You pair that with the growth of the company, with the total revenue of the company and total employee count. So that's where my time is spent, is the things that even the decisions we made three or four years ago, which at the time were sound decisions, you're looking back now and you're questioning, was that the right decision? Should we have done some things differently? Hindsight's always 2020, but you know, there's some things like, okay, well, we should have prepped for this better or planned for that better. You know, I don't look at it as mistakes. I look at it as an opportunity to learn and figure out, okay, let's change course if we need to. Let's put resources towards it. So all those things right now is what keeps me focused and, honestly, energized for the next five or ten years.

(Joel Beasley at 00:04:36) How do you go from $110 million to $200 million in IT?

(Frank DeGeorge at 00:04:41) Well, we'll do it. I mean, for the we'll there's we'll be well on our way to having that team keep up as growth. Now there's some things that we need to work on in order to make sure that happens. So there are some, you know, what started out as a workaround or an experiment supporting that business way back in 2015 when Patrick Layton, who you met at Optimize—he had been at Navy Pier—first started with us. So like I said, the decisions then were support it, figure out, is it going to work? You know, five-year contracts, managed services. A lot of these things were not new, but for the first time, you'd apply it to that SMB market. And so we put tech in place that would support it and grow. And now those things that became workarounds became core business processes, became products in some cases, and now we're looking back at it like, okay.

(Frank DeGeorge at 00:05:36) What level of core tech can we use to allow us to be innovative and agile? Right? And I think we got caught up looking back at some of the things that made us innovative and agile, but those things became core products. And so when that happens, you have to spend a lot of time focusing on things that maybe you should be able to take for granted if you had a core system that was being maintained by somebody or by an organization like a Microsoft or ConnectWise or ServiceNow, versus trying to build some of that on your own. And so that's what we're working on right now is figuring out what core can come in and take the place and allow the team of developers and business analysts and continuous improvement people focus on being agile and delivering more, I would say, better results to the company by automating more things and not focusing on just trying to keep up.

(Joel Beasley at 00:06:27) Yeah. That's, you know, I think we talked in '22 last, so it's a year and a half, two years ago or so. Probably. Yeah. What's going on with AI and GPT inside of your org? I mean, you guys do managed services, tons of human communication.

(Frank DeGeorge at 00:06:45) Right.

(Joel Beasley at 00:06:45) Right? Is AI assisting you in any of that?

(Frank DeGeorge at 00:06:48) Yes. But as people are trying to figure that out, they know that you have to be in a good spot with how your data is organized and what you're going to feed into AI systems. And that became even more clear with generative AI. Right? Machine learning, yes, we already—you know, everyone's talked about that for years about, you know, you've got to have the data, you've got to have patterns, you've got to be able to identify it, and that can be something that's more repeatable. When you start talking about generative AI, you really got to be intentional about what you give it to frame the context around it. And there's a few things that we're working on with that. I break that into like four different categories if we want to go into all that. You know, we look at what platforms are built around AI and have that. What can we use as products like ChatGPT and Copy.ai and Otter?

(Frank DeGeorge at 00:07:34) And HeyGen is one we're using for quick video production. Then we have workflows and how they hit workflows within the business processes day to day, and then data. Right? Big data, large datasets. How do I use that to better project trends or better have a chat assistant that can access large datasets versus just a small subset of knowledge bases that we've allowed it to access?

(Frank DeGeorge at 00:08:03) So all those things are kind of in play right now, and we're experimenting a lot and in some cases, failing quickly and figuring out how do we use it better or what's preventing us and what obstacles are in our way that we can then clear and then try again.

(Joel Beasley at 00:08:19) And what are you guys doing with 6sense? Because I saw them a while back. They came up the other day in a conversation I was having off the podcast about someone who was having success using their intent data. Yeah. And then I saw it come up in the interview prep for you. So I was like, what are you doing with 6sense?

(Frank DeGeorge at 00:08:37) So I would put that more in the platform, you know, part of AI that their core platform is based on machine learning models and data coming from a few different aspects. So we just started with them this year. We're actually not even live yet. We're in our pre-live kind of—they call it the Tiger Team. Right?

(Frank DeGeorge at 00:08:57) So we have our first set of reps and managers and sales reps going in there right now and seeing what's there, but we gave it essentially two years of data from our CRM. We gave it a year or so from our HubSpot and marketing automation data, Google Analytics, and a few other things. And what it does is based on the data you have, the activities, the opportunities, traffic coming to your website, and a whole bunch of other things. They bring it together and then they try and compare that, or they build a model around that, and then they compare that to data coming in in more real time. Right?

(Frank DeGeorge at 00:09:35) So who's kind of visiting our website? Who's researching us? And that's one of the biggest things I think that they have an advantage. They curate a lot of their own data from publications. So we can actually go in and see who's reading about Impact in a write-up about Impact Optimize or whatever, or who's, you know?

(Frank DeGeorge at 00:09:54) And so they have relationships with millions of other different publications or websites that they kind of feed that data in and then give it back to us in their platform. And by reviewing the patterns that have happened in the past and comparing the things that are happening today, then we could say that this company is interested in X, Y, Z. You know, in any given market, even if we have, let's say, 300 ideal customer profiles or ICPs as we call them in that territory, 95% of them are not going to be ready for what we have to offer. And we do a lot of education as part of our process. Right?

(Frank DeGeorge at 00:10:37) It's becoming easier. Like, companies know they need the help with cybersecurity and managing IT, but we do a lot of education about what it actually takes to do that. Right? So finding that other 5%, if they're interested and are finding us, that we can raise that alert in 6sense and then have a rep or an SDR reach out to them at that point in time, continue that education process with them, that's what we're hoping for to yield some success is to just find those even a couple per team or territory every quarter could make a difference.

(Joel Beasley at 00:11:10) Yeah. When I was talking with Dominic, who has a company called 2X, and this is the conversation I was having with him off the podcast about the intent data. And he had said, like, at any given point in time, one to 3% of your market's actually ready to buy. And then he goes, what he sees a lot and what we were doing, because I was asking him for tips on how to get better privately. And what we were doing is we just blast out a ton of just generalized content emails and cast a wide net.

(Joel Beasley at 00:11:40) And we didn't have a strategy to target the—we were just—our strategy was just target, cover the entire market. Make sure everybody who's in market hears our name constantly. While that works, to know that there's another strategy you can deploy in parallel, that is create content for the one to 3% of the people who are moving right now and then identify them and focus the content towards them. It just sounds stupid to not do that after you learn that that's an option.

(Frank DeGeorge at 00:12:08) Yeah. I don't disagree. And coming from where we came from, right, copier and print world, the company started in 1999. We sold copiers and printers, did some block of time network services because we had to because they plugged into the network. And in '99, that was like a new thing.

(Frank DeGeorge at 00:12:24) You know? Go back twenty-five years and that was just—it was on the news. Like, oh, this copier, they can also scan and print. Like, it was revolutionary. So yeah.

(Frank DeGeorge at 00:12:35) That is a more of a spray approach. Right? You know? It's like, okay. We'll just throw it out there. Everybody needs it. They know they need it, and then you can go and find your opportunities. As you get more refined and as we say, okay. Well, the managed IT service or the cyber service, like, that's not for everybody. Right?

(Frank DeGeorge at 00:12:53) You have to be willing to invest in technology. In most cases, you know, you can't just be a small five-person office. Right? You, you know, because that you can do on your own now, right, with Microsoft Cloud, Google Cloud, you don't really need a lot of help. It's really that mid-sized market.

(Frank DeGeorge at 00:13:08) So what started out as a spray approach then we started to kind of refine that and we defined our ICP and said, like, this is who we're going to go after. This is how we're going to go after. And these are the people that we want to go after. These are the industries that fit the most. And then you start to, you know, then you start to define the market.

(Frank DeGeorge at 00:13:24) And like you said, then there's a very small piece of that that actually is ready, that is ready to listen, that is ready to engage in the sales process, be educated, and eventually pull the trigger on something that they know they need help with, and they don't realize what it takes to actually do it properly.

(Joel Beasley at 00:13:44) Do you think that we're going to see a rise in in-person events and in-person business as the AI becomes really good at deep faking or creating digital twins?

(Frank DeGeorge at 00:13:58) 100%. We've already seen—you know, we shoot for in-person meetings first and foremost. I think they're more effective. I think people like them better. Yes, we're doing this remotely. It's convenient. Right? But I think overall, in-person events become more productive. You know, you put an interesting spin on it.

(Frank DeGeorge at 00:14:16) We have a platform, HeyGen, which I can record a video myself for two minutes. Right? And I actually did this at the all-company meeting and didn't tell anybody. And then we played a video, and everyone's like, he looks a little off or he sounds a little bit off, but they're like, maybe it's just the camera or the delay in the stream, and not until the end of the video that I come out and say that this was completely generated by AI with an AI script and it took me about—you know, besides the recording, now I can go back in there. I can go to ChatGPT, make my script, put it in there, and I can make me say anything. Right?

(Joel Beasley at 00:14:54) Mhmm.

(Frank DeGeorge at 00:14:54) And that's real and people aren't really going to know the difference anymore. So I think, yeah, from a security standpoint, that's an interesting approach that people will lean more towards those types of in-person events and meetings, because you might not very quickly be able to determine what is real and not real. And I think that is a challenge. That's something that's got to be figured out. It's not going to slow us down in the meantime. Right? I think the whole market has to figure that out first and foremost.

(Joel Beasley at 00:15:24) 100%. Yeah. I don't like ever letting fear slow you down or distract you in that sense. I'm just curious, like, from a macro level. Right? Because I have these high-level conversations constantly. And I see that there's going to be this rise in the ability to authenticate that I'm real and this is happening in real time.

(Frank DeGeorge at 00:15:45) Right.

(Joel Beasley at 00:15:46) And that technology hasn't happened yet. The closest thing we've got is the blue checkmark, you know. Right? But to validate it and so, you know, the content's coming from me, and not only do you know it's coming from me, but you know it's actually me.

(Joel Beasley at 00:16:02) Yeah. I think there's gonna be a rapid improvement in the next two to five years with being able to fake yourself actually being there, and I think that's going to push us farther into in-person relationships.

(Frank DeGeorge at 00:16:16) Yeah. To put this in perspective, when I did my course at Kellogg, I did it, I think I was done with it last summer. So it was last June and the year before that, and it was a Chief Digital Officer program. Generative AI was talked about maybe this much as part of a year-long class on chief digital officers. I even took an AI marketing, you know, elective as part of it and design thinking and other things.

(Frank DeGeorge at 00:16:45) So that's the perspective. Like, even the best university, one of the best universities with the best professors talking about this didn't see how quickly that was gonna change, you know, reality of what's going on, because there was only a very small piece of it. In fact, as part of one of my final projects I did there, I used ChatGPT. I disclosed that I used it, and honestly, I was a little concerned that they may frown upon that because it wasn't really blessed. It was like, oh, yeah, you can do it or not. But I did it just because it's like, hey, you know what? This is stuff that we're gonna have to be working with. So I did that in November of when ChatGPT first launched, you know? So it was, I think it was November. Right? I mean, my timeline might be a little off, but I think it was November when they first kinda dropped out and said, oh, you can now talk to this generative AI thing via chat.

(Frank DeGeorge at 00:17:25) And so, yeah, I know we're getting a little off topic there, but it's just interesting how fast it changed. And one of the things I just put on LinkedIn yesterday, actually, my daughter really wants a chinchilla. Right? A pet. And so she sent me screenshots of her on her iPad, and she's eight, by the way, of her on her iPad asking ChatGPT how to convince her mom or dad on how to get a chinchilla, what the requirements are. Is it a low maintenance pet? Right?

(Frank DeGeorge at 00:18:04) And so you can see her lighted questioning, and she sent it to me, and she goes, are you thinking about it now? I'm like, I'm super proud of you for finding out how to do all this. So it's just, it's gonna be in our day to day, and things have to adapt and change. But, yeah, not to be cliche, it's not gonna wait. Even my eight-year-old is using it on a day-to-day basis.

(Joel Beasley at 00:18:26) Yeah. My daughter is almost seven, and she hasn't gotten into that yet. But they have learned to do something similar to that behavior with talking to Alexa.

(Frank DeGeorge at 00:18:38) Oh, yeah.

(Joel Beasley at 00:18:38) Yeah. Just because I haven't enabled her to do the ChatGPT thing yet. She just started reading about six months ago, like, fluently on her own. And so, yeah, I'm curious to see. I've never even thought about that. Now you've got a, my wife and I tonight at dinner, Frank, we're gonna be talking about when it's okay to introduce our kids to GPT.

(Frank DeGeorge at 00:18:59) Right.

(Joel Beasley at 00:19:00) It'll probably be like YouTube where, like, we have certain apps where we let the kids watch whatever they want whenever they want because they're, like, apps that we believe in and they're moderated content. But then there's, like, YouTube or there's, like, a show they wanna watch, like Ryan or something like that, and they're only allowed to watch YouTube with us. So it might be that type of situation where you could do ChatGPT with us. Yeah. For now.

(Frank DeGeorge at 00:19:24) Supervised ChatGPT.

(Joel Beasley at 00:19:26) Supervised GPT for kids.

(Frank DeGeorge at 00:19:28) I don't know.

(Joel Beasley at 00:19:28) If they have one yet, but they should. Yeah.

(Frank DeGeorge at 00:19:31) Not a bad idea, actually.

(Joel Beasley at 00:19:32) I know. There we go. But speaking to your point of you detecting it early, I had seen some projects. So everything popped off in February of last year with the general MSNBC, Fox, the mainstream people, my grandma knowing what GPT was type deal popped off in, like, January, February.

(Joel Beasley at 00:19:54) And they definitely released many versions. Even multiple years ago, they have released versions to, like, senators and different people in legislative affairs so they could understand what was going on. But I had been following this and looking to monetize it because I've been following LLMs for, like, seven or eight years.

(Frank DeGeorge at 00:20:11) Yeah.

(Joel Beasley at 00:20:12) And I found, and so I saw it progressing, you know, from the bad. I mean, like, it wasn't creating good images. It was very sketchy, not good. And then just watching it have leaps and bounds. And so right around that end of the year time frame, probably two or three months before it popped off, I saw this opportunity and I messaged an investor at my company, and I was like, hey, he's one of the directors of Florida Funders. They're like a large VC firm. I said, pay attention to this because I don't know if it's a year out from everybody figuring it out or two years out or six months. But this is gonna be huge, and here's a couple example applications of how you'd be able to use it. And for me, I was specifically interested in using it for storytelling.

(Joel Beasley at 00:20:50) So I could put in a book, and then it could continue that book with subsequent versions, and then create video to illustrate that book. And now you're essentially creating movies from books through AI. I was like, that's one use, you know, a couple other things. And so I pushed that text to him, and he was just, like, thumbs up. And then, like, six months later, when it's all out in the news, we were talking. He's, like, oh my gosh. He's like, yes. I remember you saying that. He goes, that's unbelievable.

(Joel Beasley at 00:21:14) And now there are startups that are doing that, continuing the books, and so you can continue the stories even if the writers are gone. There's people turning it into video, and there's about a million other applications of this generative AI and GPT out there.

(Frank DeGeorge at 00:21:29) Yeah. You know, and for companies like us, we have a few different groups that are kind of, I don't wanna say playing, but experimenting. Right? Trying to figure out what the business value is in day-to-day workflow. And that was one of the things that I wrote down as far as, like, the platforms, are, you know, we're trying to figure out, yes, we can chat, and everyone can chat, or you can use Copilot and they can draft emails, but how do I actually put it into day-to-day business workflows?

(Frank DeGeorge at 00:21:53) And that's something that we're working on right now to figure out. Right, can I take this and bring it down and then reference it as an LLM and then bring it back and then do this with it, interpret it, and then bring it back as part of a process? And for us, you know, I think we gotta be careful with all the startups. I don't wanna discourage startups and that, but, you know, if it's going through a major business processes, we almost, like, have to choose a core, you know, one of the bigger platforms that we know is gonna be around for a little bit because some of these startups are going up and down and being bought and sold, you know, left and right. You know, for the experimental ones, like, we're okay with that. But as we look at things to, you know, put out to our customers and engage in customer experience, we're, you know, we're actually looking at something, you know, like an IBM Watson or stuff that we're doing on Azure with Microsoft's Azure AI services and stuff like that, because there's gonna be a bunch of things that bubble up.

(Frank DeGeorge at 00:22:51) And I think some of them are gonna get absorbed. Some of them are gonna take off, right, like OpenAI did, and some aren't. You know? But it's interesting to see what the creativity is happening with all these different startups to see what actually takes place, and what, you know, what ideas that all these different companies can bring to the table.

(Joel Beasley at 00:23:11) Oh, you're exactly right. So I've gotten involved in some decentralized AI projects because there's gonna be legislation around the models and the legality of them and their main intelligence. So there's a whole push of people in the crypto blockchain world that are making AI decentralized so that they're just available in public and kind of out there and you can't get rid of them. Right? But through that, I've gotten some really interesting experience in AI, and you're exactly right.

(Joel Beasley at 00:23:39) The things will boom and burst. The market will shake up. And so for you as a business, you know, that makes sense why you go with Six Sense. Right? They are one of the leading brands. They're known for being the highest quality product and, like, more expensive, but it's good. And so that would make sense for you guys to go with someone like that.

(Frank DeGeorge at 00:24:01) Yeah. You know, and we're working with a small company now that seems to be a unique approach of how they put generative AI and AI content in a workflow. And I know Zapier has, you know, the GPT plug and stuff like that as, you know, it has that as, like, a base level, but also the way that it can bring in data in your environment, keep it isolated so you're not just throwing it out there in other random platforms. It kind of keeps it centralized, and you can bring your data at different points along the way and follow it along the workflow. So I can actually build prompts over time as part of a workflow to take something that may start as an Excel file, that may be interpreted, that may go and then be analyzed, and then spit out a report in an email that says this is what you want, and it's three or four different prompts accessing different data along the way to get a final result.

(Frank DeGeorge at 00:24:54) Those are the things that I'm really looking forward to because that's how I start operationalize it besides giving everybody a new chat friend in their email, you know, with Copilot to make email generation faster or understand meetings that they forgot to do takeaways on.

(Joel Beasley at 00:25:11) Yeah. And I've played and programmed some of these models, and it seems to me, now I'm not that experienced. It's not my full-time job. I just armchair style. Right?

(Joel Beasley at 00:25:20) But it seems that they're really good at exactly what you talked about, training, like, making an instance of it and training it to do, like, one very specific thing, and then having a bunch of those or basically, like, a bunch of scripts or macros of GPTs, of things that this GPT has specialized really good at doing this thing, maybe combining CSVs and cleaning them up. Right. And if you try to keep teaching it more things, it seems to get a little wonky. So I've just not tried to fix that and instead just said, I'm just gonna move forward with having these narrow intelligence. Oddly enough, neural networks and large language models are based off of humans' work, and that's exactly how we utilize humans.

(Joel Beasley at 00:25:57) Right? They're specialists in this one or two specific things. And so I said, well, that all makes sense. So the next layer is the routing layer, the GPTs that are gonna get really, really, really good at figuring out what your intent is and where to route that to. And you can see the early stages of those routing GPTs inside of OpenAI system with the plugin store. They figure out where to route the request. So, yeah, I've also noticed that because of what you said, all this churn and all this movement in the market and all these advancements with these tools, that most of devs I'm talking to that run projects are intentionally building their systems to be as agnostic to the underlying LLM as humanly possible, and they can switch their models dynamically. Let's give this set to OpenAI, see how it responds. Let's give this one to, you know, this other model over here. And have you seen that too?

(Frank DeGeorge at 00:26:54) Yeah. More and more. In fact, the one that I was referencing that we're kind of experimenting with right now, you can choose really almost, I don't think any LLM. Right? But there's probably five or six that they can work with outside of GPT-4, 3.5, right, and so on and so forth. So, yeah, I do think that is gonna happen as these other organizations, especially larger ones like Google and Apple. I read an article at Apple today. I've released a research paper yesterday on actually figuring out how AI can actually sense the resolution of your screen. I'm not sure if you saw that, right, but it could actually start interacting with content on the screen based on what's there.

(Frank DeGeorge at 00:27:35) So that could be super interesting, especially if you consider that the target probably is the iPhone of what they're working on as they release the next set of AI tools. But on the LLM side, I think, yeah, it's too early to tell what's gonna, you know, another time about, you know, small language models and stuff like that. So I think depending on the use case, and I think you're gonna start seeing really specialized LLMs. Right? I think you already start to see that a little bit.

(Joel Beasley at 00:28:02) Yeah. It's just moving so fast. There's a couple of them that are now small enough, you know, you can like, Ollama and or Llama, and you can run it on your computer. Have you ever, I put it on my MacBook, like, I have a three or four-year-old MacBook, pretty slow. But on my Mac Studio, that thing was just as fast as GPT.

(Frank DeGeorge at 00:28:21) Yeah. We looked at some of the offline ones that we could run ourselves, but try not to do that right now, honestly.

(Joel Beasley at 00:28:28) Well, that's a whole skill set. You'll have to bring in a whole employee just to do that.

(Frank DeGeorge at 00:28:32) Yeah. Yeah. Yeah. Then I don't think it's any easier or less expensive to run once you get on your own environment because you're probably running it someplace in the cloud and, you know, unless I have a bunch of Mac Studios that I'm buying to do it. Right? So, yeah, that stuff we're still figuring out. But for the most part, we're using more or less private datasets with cloud-based, you know, tools.

(Joel Beasley at 00:28:53) What's the most effective use and application that has occurred with your business in relation to these AIs or LLMs?

(Frank DeGeorge at 00:29:03) Well, I would say is one that we're working on now, and we're looking at a few different platforms. One of them is IBM Watson. We're not really working with them directly yet. But that is something we're putting as an in-between between customers submitting tickets and problem tickets into our service desk, and then getting a knowledge-based response based on what the ticket says versus what we say the process should be to do X, Y, and Z. And then so that's what I said, you know, before that you can't just give it, and I think I mentioned this, but you can't just give it, you know, five or ten years of ticket history.

(Frank DeGeorge at 00:29:45) Right? Because you're gonna get a lot of noise there. So cleaning the data and using a knowledge base and then using and tweaking it for what AI expects it to be and defining clear problem statements with clear resolutions that things are documented properly. That is likely to yield probably the best results with that so we can give customers more access to the basic questions that they ask and then our spit on it. Right?

(Frank DeGeorge at 00:30:14) Like, how do you reset a password? Well, you know, on your computer, you do it this way. If you have Azure or Office 365, you do it this way. If you don't have any of those, you do it this way. And so pairing with what that customer is asking, who that customer is, and what the response is on the knowledge base and delivering that back to somebody for a quicker response than we can give when they just submit a ticket.

(Joel Beasley at 00:30:37) Of a human kinda monitoring the outcomes of that and saying if it's good or bad, like fine-tuning.

(Frank DeGeorge at 00:30:42) Yeah, especially as we get started, right? We've got to see how that works, how that gets handled. So those are the things that we're working on now that I think could obviously help increase customer experience and customer service and make it easier for us to support that customer.

(Joel Beasley at 00:30:58) One of the things that I like about you that I rank you as high competency—one of the things that you do is whenever I'm talking with you, you have the high-level business understanding, but you also have the way you speak. You can tell that you've had hands on the keyboard and played with the tools. How do you functionally do that? Do you just see something cool and you're like, "I need to know this," put two hours in my calendar to dive deep into it, or are you going in side-saddling with an engineer who knows how to do that and having them walk you through it? What's your actual process for getting your operator time in?

(Frank DeGeorge at 00:31:33) It's funny you mentioned that because I was just making fun of myself to a couple of my developers and one of our managers and directors, and how I feel like I'm losing some of that because there's so much going on that we're trying to plan and the roadmap and what gets attention, what gets priority, and listening on calls to make sure we stay on track and listening for roadblocks or things that people understand one way, but reality is another way. You know, I just—for many years in the field when I was in sales, I did a lot of hands-on demos, right? I didn't rely on anybody to build what I needed to do when I was demoing things to customers or customizing demos or the workflow and logic. I was a horrible coder.

(Frank DeGeorge at 00:32:18) You know, in fact, so much so in college that I—you know, I wish we had low code when I went to college because I would've loved that. I would've been able—because the logic is sound, right? I can do that. I was very good at that stuff, but I just hated the syntax. I didn't have the attention to detail for that type of thing that didn't interest me. But solving business problems with tech and logic and workflows, that's where I really did the best, and I did that for 15 years in front of customers. So, you know, I need to make sure I keep up with that so I can maintain that skill set and not just rely on what got me to where I'm at, even though I do a little bit. But it's harder for me to spend the time more and more doing that. But I read a lot. I listen to podcasts. I read. I listen to your podcast.

(Joel Beasley at 00:33:07) Yeah. Do you really?

(Frank DeGeorge at 00:33:08) Yeah. Oh, thank you, man. I try and—not all the time, so don't—

(Joel Beasley at 00:33:12) I've done a thousand episodes. I don't expect—

(Frank DeGeorge at 00:33:14) Don't start quizzing me, right? But I have—let me change it. I have listened to your podcast.

(Joel Beasley at 00:33:20) There we go. No, no, no. Frank's an active listener. He listens to one episode at least every year.

(Frank DeGeorge at 00:33:26) But I get in, and I try to understand when our engineers are stuck or developers are stuck or business analysts are stuck—why we're stuck and how do we get past it. And, ultimately, it comes down to why we're doing things and why we're stuck and why we're trying to solve it this way and not a different way. And so all those things together, I think, just—yeah. I mean, I've had the hands-on experience, and I try and maintain a level of that.

(Joel Beasley at 00:33:55) So let's give a shout-out to Dot Security. We might as well do some work here, Frank.

(Frank DeGeorge at 00:33:59) Sure.

(Joel Beasley at 00:33:59) If you have needs, what are the people's needs when they're like, "I'm having a need, Dot Security might be the answer." What is that?

(Frank DeGeorge at 00:34:06) So with the way that IT has changed—you know, not to sound cliché, but especially rounding from COVID on, you know, to the years after—it was a monumental shift of how companies operated, right? We still had companies going into the office because their VPNs weren't set up or they had a local set—things have come a long way with different cloud apps and how we use Teams and Slack and all these things to share and collaborate. You know, companies that didn't have a document management system, you don't have a choice, right? The invoices and POs and everything are coming in, and "I usually print them out to match up with this." And so that fundamentally forced people to change. And so as the pandemic was rounding, you know, or heading in the right direction, we definitely saw the challenge to our customers. And so Patrick Leighton and Jeff Leader—that was really their minds together that came together and said that we need to focus on this because we've been providing cybersecurity services for years, right?

(Frank DeGeorge at 00:35:12) And, you know, separating the operational components from the cyber components, because everybody needs operational IT if you're a midsize organization. You need to be able to have the expertise in patching and backup and network and firewalls and cloud security and databases, servers, sys—you name it, right? On-prem, Meraki, whatever tech it is, you need to have expertise, and it's really not feasible for one organization to have that expertise. So you work with a company like Impact that has, I don't know, 200, 300 engineers with expertise in every one of those fashions. And so you can really come in and be operationally well-run from an IT standpoint. Now that doesn't mean that you're safe from everything, right? You still have to have—you know, there's things that you have to do right, like backups and verify backups and do—you know, we use KnowBe4 for phish testing and seeing who's going to click on the email. Because you give the best environment in the world, if you click the email, type in your password—right? Yeah. So people are still the problem.

(Frank DeGeorge at 00:36:15) So the next phase of that with Dot was to put more advanced protections in place around threat protection or endpoint protection and then really monitoring environments for changes. And it's not just tech, right? Because in our SOC, we get—I think I forget the number. It's either a billion a day or a billion a week or multiple billions of alerts coming to the SOC, and we put hands on a couple hundred of them. So it's really refining what that is in that environment and improving the security posture over time.

(Frank DeGeorge at 00:37:01) Because then the next level of just having the tech handle and having someone review the SOC alerts is then: what are you okay as a business, and what risk are you accepting in order to conduct the business the way you are? And I'll give you just an easy example of that. We have a call every week, right? And so we have four security calls a month, right? Two of them are tactical, which means we're reviewing the risk register. What are things that we find at risk? Are we accepting it? Who's in charge of it? Is there a project associated with it? And who's really tracking it to make sure that happens? And then the other one is more of a strategy call. What are we not doing today? What can we be experimenting with? What new platforms are out there with AI for mail protection and spam protection, right? How do we keep that on the cutting edge? And so we review that.

(Frank DeGeorge at 00:37:42) And then the risk register—there's an old application that we ran that had plain text passwords in the database, right? A basic no-no. It wasn't being updated. The company that made it has acknowledged it and had no plans to fix it. As a business, we've got to say, "Are we accepting that risk or not?" Because it's going to come up every time that they scan and see plain text passwords that are being exchanged between the application. And so it takes a human interaction to say, "Yes. We accept it. We scramble the passwords every month. They have to be 16 characters, right? It's not used that often and we can isolate it." And so just things like that that you increase the posture over time of the security environment. And you can do that for lots of things within your organization. And so a risk register may have a couple hundred items on it or a few hundred items on it for a customer. That's managed by a person, right? Not by a platform, not by—so while there is platform software that helps us manage it, right, but it's about interactions with that organization and improving that over time, following basics like CIS controls and making sure that we're crossed off the boxes to figure out how far along those controls it makes sense for that organization.

(Frank DeGeorge at 00:39:08) And then, not to get completely off on a tangent here, but then you've got compliance and companies that are in the supply chain for the defense—you know, different defense agencies and whatnot—and that spawns a whole other world of conversation. And we haven't even talked about checking the box for cybersecurity insurance yet or cyber insurance yet. So that's why companies are starting to realize that they need help with it. There's a lot there.

(Joel Beasley at 00:39:39) Dot Security, dotsecurity. If you guys need help with the security managed services, that's where you go.

(Frank DeGeorge at 00:39:46) And I hope I did it justice and Patrick doesn't call me up and say that I said—

(Joel Beasley at 00:39:49) Patrick's going to call you up.

(Frank DeGeorge at 00:39:49) Yeah. He's going to say, "Well, you didn't say this right or that's, you know, this is that way." And so—

(Joel Beasley at 00:39:54) Awesome, Frank. We did it. We made a podcast. How do you feel?

(Frank DeGeorge at 00:39:58) I feel good. It's good to talk with you, Joel.

(Joel Beasley at 00:40:00) It's always good to talk with you. Whenever I see your name come up on my calendar, I smile. I'm like, "I told my wife, it's like I'm talking to Frank today. He's good people." Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn or send me an email, [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.