Not all AI being equal. If you're talking about trying to bring something into the organization that's going to be transformational, it should have enterprise capabilities.
SPEAKER_03Where are most people coming to you today and where were they, say, three, six months ago? Is it the same?
SPEAKER_01Most organizations are now coming to us with the story of we've purchased X amount of licenses or we've invested in this platform, but we're still not clear on the direction we should be taking, whether what they're doing or how they're using it is secure.
SPEAKER_03How long until they start seeing relief or they start seeing pilots successful?
SPEAKER_01I think uh, you know, it's not unrealistic to think that in within six months' time frame, you know, they can recoup their investment in whatever tools they're using. So let's switch gears a little bit.
SPEAKER_03What are you looking forward to the most, or what are you most excited about with what's happening with the tools right now? I think there's just so much.
SPEAKER_01You know, I've ri I've really kind of leaned into a lot of these things. I'm a big fan of home automation. I've really automated the heck out of my house. And so I'm really excited and a little nervous about seeing the convergence of robotics and AI.
SPEAKER_02Jim Spignardo is a cloud strategy and AI enablement leader at ProArc, helping companies cut through AI hype and build secure practical workflows that drive adoption, improve productivity, and deliver real ROI.
SPEAKER_03Welcome to Using AI at work. I'm your host, Chris Dagon. 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'm the host of the show. And today our guest is Jim Spignardo. And Jim and I have been trying to do this for a month and a half at least with some misconnections. But finally, we're on for today. And Jim is with ProArc, and they are providing technical services for enterprise and lower middle market and above-sized companies. But his role in particular is cloud strategy and AI enablement for these clients. So today we're going to be talking about in particular, I want to know like the patterns that he's seeing, the trends he's seeing, so that you as a listener can kind of hear how others are generally doing things or where they are on the AI journey and figure out where you stand as far as your preparedness and your, you know, uh like, are you playing the game at the level you should be? So, Jim, before we jump in, I always give the uh guests a chance to let the listener know what they should expect to walk away with today.
SPEAKER_01Yeah, so uh hopefully um this will be a nice back and forth conversation between yourself and uh and myself, Chris. Um you know, I love just spending time uh getting the word out to customers, clients, uh folks who are just getting started on their AI journeys to have them get a better appreciation for you know what is this technology going to do to their industry specifically, but uh the the work for the workplace in general. Uh there's a lot of fear, uncertainty, and doubt right now. Yeah. A lot of organizations are still struggling with, you know, what are we supposed to be doing? Uh are we behind? Are we ahead? Um do we even do we even have to spell AI? So um part of that is just uh hopefully uh helping organizations understand that uh it's better to do it right than to do it fast, essentially.
SPEAKER_03100%. There's a lot of people out there that feel the like uh an urgency to do it fast. And I'll tell you, um, one of the things that I've noticed a trend with talking to, because we do a lot of like presentations to YPO groups and vistage and professional organizations, and I'm hearing three common uh I guess points of contention when it comes to AI. They want to do it, but they don't understand the risk. So it's like can't go all crazy if we don't understand the risk. They want to do it, but they're having a hard time figuring out where do we start. So use case identification, pilot project identification. And the third one is we'd love to get started, but we don't have anybody that can lead this for us. Like we need help. We don't have anybody, right? And I would imagine that that ProArc is addressing probably all three of those to some degree for their clients.
SPEAKER_01Yeah, absolutely. And so, you know, we have kind of a playbook that can meet customers where they're at in that journey. Okay. Uh from the very beginning states, we uh engage with either prospective customers or existing customers to um deliver an executive briefing on AI just to so that they can kind of get a sense of where um where the technology sits today. To your point, you know, what things do they need to be aware of from a risk perspective? What types of security tools exist to control this? Uh, but also too, what does a successful journey look like? What are the things that we've seen through our own uh attempts to do this, right? As well as working with other customers that uh lends itself to a uh more successful outcomes. And um yeah, we the to your point, we can meet customers anywhere they are on that journey, whether it's just beginning to you know talk to the the C-suite or the board to, you know, we've been doing this a while, we get it, we're very technical, we just need some help implementing a data platform and connecting to our live business systems, right? So we can you know we meet you anywhere on that journey.
SPEAKER_03Well, let me ask you then on that spectrum that you just defined, where are most people coming to you today and where were they say three, six months ago? Sure. Is it the same?
SPEAKER_01I would say most yeah, most organizations are now coming to us with the you know the story of we've purchased X amount of licenses or we've invested in this platform. Some team or department within our group has convinced us to spend some money, but we're still not clear on the direction we should be taking, whether what they're doing or how they're using it is secure. I would say that's the vast majority of customers right now. They're kind of at that very beginning phase. They're they're experimenting, they're ideating. We do have a decent set of, you know, I would say maybe 15-20% of our customers that already have AGI governance councils, they already have training programs, they're building agents, and what they want now is some governance. They want us to come in and go, hey, tell us how to make sure we keep uh keep this thing under control and and don't lose control of it. So that's something that ProArc does.
SPEAKER_03Yeah, absolutely. Yeah. So these um, and that's a that's a pretty good uh uh I guess percentage of I would say literate AI literate or AI enabled clients that you've got working with you. So that's uh more than what I'm seeing um normally, but we're catching people earlier in the journey for sure. I was speaking at an event yesterday in Salt Lake City called Aquacon. It was about uh those who acquire businesses, so private equity, MA, family office, and we had a panel on AI, and one of the speakers said something that I thought I just took for granted, but I I didn't really look through the eyes of your potential customer. And he said, just because you've gotten some licenses for some people doesn't mean that you're ready for AI, right? So for those that are listening, I think they probably fall into that that client avatar that you mentioned where they've they've got something, so people are doing something with it, they don't really have uh optics on what's happening or or really a strategy put together. So if a client comes to you like that, you'd you'd suggest the first step for them, their first exposure to ProArc's abilities would be that executive briefing.
SPEAKER_01Um typically I would say yeah, because honestly, it kind of does a level set. Uh it we don't want to assume any knowledge or any uh understanding that may not already exist. It also gives them an opportunity to in a very uh very uh non-risky way, because we typically do these for free, um, to build trust with that individual that company or that individual so that they can see you know what we're capable of doing for them. Um and during that conversation, like you know, towards the tail end of that, we talk about you know this the roadmap to adoption and helping them understand what they need to be able to put in place uh within that roadmap to ensure success. So that kind of makes them pull back and go, okay, where are we in that? Where do we need to jump in and what's the next step for us? Right. And that's typically then when we'll engage for some sort of paid engagement with them.
SPEAKER_03Yeah. So what kind of information do you think that an executive needs to see? Because I ex we do an executive briefings, plenty do them, right? But in order to make them a valuable uh use of time or because here's here's what the listeners that are getting started don't want to do, make the wrong choice. Right? Right. Yep. So what should they be looking for in that executive briefing if if they're you know already engaged or talking to somebody?
SPEAKER_01Yeah, well, what one of the things you talk about is you know, not not all AI being equal. And if we're if you're talking about trying to bring something into the organization that's going to be transformational, uh it should have enterprise capabilities. I think you and I talked a little bit uh before that we're a Microsoft partner. So we are uh, you know, we we lean heavily into Copilot because of the fact that it does uh live inside your data environment. Uh it has and respects your security parameters, and it has the ability to understand the context of your organization, what people's roles are in the organization, the relationships they have with one another, the type of work they're performing. Um, it just makes that transition a little bit easier and can allay them of some of the concerns and risks. But we do tell, I mean, we we we want to be very honest with them because we understand that uh there are tools out there that maybe may be more well suited depending on the type of task you're trying to perform. That's not a Microsoft solution, right? And if that's the case, what are the things you need to be aware of? What what what risks do you need to to uh uh to ferret out to understand uh when you go to do this, what guidelines do you need in place for those users? What types of tools can you put in place to make sure you have visibility into the data that's coming in and out of those systems so that you're not pretty undue risk? Part of part of what we give them to is in that is just kind of explaining what what general uh or generative AI is, how it functions, and a conversation around data governance. Understanding that, hey, if you want to start on this journey, um make sure you have your data house in order, or at least the part of which your part of your organization you're gonna apply AI to, at least that part of it is has some sort of governance and clarity around it. Because ultimately that's going to have a very significant impact on the output you'll receive and the quality of the uh the performance of the uh AI tools you're trying to deploy. Um and then you know, we also take a fairly decent amount of time to talk about agents, you know, the different types of agents. Big topic assistants versus autonomous agents versus um custom agents versus uh retrieval agents, those types of things, so that they can kind of get an understanding of okay, once we get past this personal productivity tool that we use in all throughout the M365 apps, what is it that's possible? You want to expose them to you know the term Microsoft uses the art of the possible. Uh, that part of the conversation usually gets pretty interesting. And you just talked to me, talked a little bit before about you know where our clients are at. Um we're finding that clients now are moving quicker through that early stages of general adoption to agent development, right? Uh, I would say a year ago, there was still this, let's just put this in here and let people use it as they see fit, and we'll figure out a strategy. They've heard enough now, they've listened to their peers or their competitors that they're now understanding the power of agentic AI and now trying to come up with those use cases and you know, figure out which processes would benefit the most from uh AI and automation.
SPEAKER_03Well, that happened fast, man. It went from like I need a GPT license to let's talk about agents. I know it's it's amazing. You know, and and it's interesting because we're talking to you know business owners and stuff who are at that first stage of, yeah, we're are or maybe they're like we we don't know who's using what, but we want agents. Like it's it's I find it interesting that they have very little context on what it what it means to have it be part of your day-to-day you know workflow and that sort of thing. They're already talking about like how do we get autonomous agents to run the business. It's crazy.
SPEAKER_01Yeah, and then again, sometimes we have to we have to bring them back down to reality, right? We have to say to them, okay, uh, you got to crawl before you you fly. Um and part of that is you know, we always want to emphasize defining the uh the high value, low effort use cases initially. And that might not be the sexiest thing. It may be a no-code agent built in co-pilot chat that uh ult it can bring some immediate relief to a specific role in the organization. But by doing that, you're gonna establish a pattern of success. Uh, you're gonna be able to share those stories with the rest of the organization, uh, and then you can start to dream a little bit bigger, uh understanding that during that during that uh that piloted or proof of concept phase, you need to start to turn around and look at your data and go, well, where would we go next? And are we ready based on where our data, what you know, how our data is today, um, to go to that next level? If not, let's let's spend some time doing that cleanup. Let's document what systems exist, how they uh interact with each other, what dependencies they have, and let's do that cleanup so that when we get to that stage, we'll be ready to go. Okay, this is good.
SPEAKER_03How much of this is happening simultaneously? We okay, so hey Jim, we got our licenses, but we need help. So certainly get the users trained on safe use of the tools and how to actually just navigate tactically. Are you simultaneously um in the process of that starting the evaluation of agentic possibilities for workflows and processes, and also spinning up uh like a data analysis or a like a data audit and clean team?
SPEAKER_01Yeah, so so part of our smart service, the Smart Start Services, that's the kind of the brand we use, uh we'll come in and give them some foundational understanding of the capabilities and features, uh, some a little bit of skilling on prompting, uh use case workshops that we run through. But at the same time, we're typically doing a data risk review. So we're going through, we're inventorying their information, what's what what's in the in email, what's in Teams, what's in their document repositories, and uh making sure that they're not over provision or overpermissioned, uh, that they're not sharing externally, that uh they're using good sound, zero trust principles when it relates to identity security. And we do have programs that can help them go from that initial pilot proof of concept phase where we've identified those gaps within the data environment to help them actually start to put a plan in place to take action and get some consistency and some clarity around their data, whether it's changing naming conventions for files or adding metadata or just archiving and deleting old data, yeah, you name it.
SPEAKER_03What I tell people is they want AI to come in, they want to turn the switch and like, oh, the problem solved, right? And I tell the the you know interested parties that it's a process and not an event. For those who are listening, what does that timeline look like depending on let's say it's a lower middle market size company, not an enterprise. I know that that's different timelines. What like if they're ready to get off the starter block, how long until they start seeing relief or they they start seeing pilots successful?
SPEAKER_01Yeah, I think uh, you know, it's um not unrealistic to think that in within six months time frame, you know, they can recoup their investment in whatever tools they're using. Now, they do have to be intentional, they do have to track it, they have to make sure that um uh the use cases they'd established are actually being implemented. Uh, one thing that we learned pretty early on on our own journey was if we create use cases, um, we can't uh we can't make them an option for folks. Can't say, well, this is how you could do this particular job um because we've given you prompts or whatever other tools. Um so we we had to go back and go, yeah, when we told you about those use cases, that was a mandatory thing. It wasn't a wasn't a nice to do, right? So we and it's a little bit of a challenge for some folks who uh who have been very traditional in the way they approached work, yeah. Make that flip. But but you got to keep going back to the well, you got to keep tracking the usage, you gotta reinforce it um over time. And uh usually by six months, you um there's a culture of adoption um where you've st you're starting to get traction. What I will say though is you know, we've been doing this for over two years now, um, but it doesn't stop. It's not a it's not a well, we got to some this point and now we can just take our foot off. Right. Um we went from 10 licensed users uh two years ago to now we have 150, uh, which is about 30% of our business licensed. And we also went from, you know, which we I tracked the return on investment and produce reports to our executive team, probably somewhere in the neighborhood of um about $14,000, $15,000 a month in assisted value with you know a couple dozen licenses to now where we have 150 users. This last um reporting period, I believe what I was seeing was about $75,000 a month in assisted value, which is almost a million dollars a year assisted value from and again, it's that type of growth could not occur if we didn't track that investment and usage and prove to the C-suite that this is worth just putting more people on and getting more licenses for.
SPEAKER_03And and how do you go about tracking it? You look at the the before state and the resources required to execute the widget to the after state over a period of time?
SPEAKER_01So like nicely enough, Microsoft has quite a few tools that you can use to look at the uh adoption usage as it relates to the various apps that are being used, the tat the type of tasks that those they're being used in those apps, uh, and then you can translate that to a to a you know a blended hourly rate that then in the organization. Now, we don't stop there though, right? Because that's kind of a quant a qualitative type of um uh assessment, right? They're using it, but how effectively they're using it. We also within our use cases try to define key metrics. Uh so if we're talking about our help center and being able to um uh transcribe all their calls and meetings with customers and take better quality notes, uh, that's something we want to look at and say, okay, are we reducing the uh the time to um to uh to actually solve customers' problems? Are we finding that the quality of the information we're storing in our ticketing system is better and therefore we're leading to less issues? So we we like to track those things too. And again, depending on the role, depending on the use cases, it's important that you do that. And every business is gonna have different metrics you want to track.
SPEAKER_03Because I guess if you're looking at sales and you're like, oh, we saved the sales team X number of hours, that's great. But how many more sales did they close because of that? Yeah, you need to be tracking, yeah, that holistic.
SPEAKER_01Yeah, and again, if you're not if you're not tracking it now, stop, pause, figure out where. Your baseline is, then pick up your use case and and then have something to compare against. It doesn't have to be perfect, right? It doesn't have to be uh, you know, you don't have to be down to the nickel or dime, but you want to have a general sense. And the funny you should mention that because we we have done that when it comes to, for instance, uh responding to RFPs or requests for proposals. We had a certain success rate, we had a certain amount of time it used to take us to respond. And, you know, the we've drum we've created an agent to help us with all of that and dramatically reduced uh our effort. We went, we probably won about 50% of our RFPs in the past. We're now close to about 80%. And the time it used to take was a pro probably more than 10 days with about three pre people's labor. Now it's down to about one and a half people in about four days, right?
SPEAKER_03Yeah, nice. So those are the types of things that that any listener should expect. Like that's real. We're seeing something very similar there. Now, you mentioned that you guys started initially with those 10 licenses. Where were those licenses? Was it like you and some enthusiasts to tinker with?
SPEAKER_01It was probably more the people who were um early adopters. It was definitely some of our uh consultants or strategic consultants, some of the some of the exec team. But when we expanded the first the first team within our organization, we targeted our um our support engineer team, our help sire. Um, because we A, they had uh they have a certain level of technical aptitude. So we know they would take to it pretty well. But we also had some really easy, simple, fast um use cases to get out the door that could really show some impact pretty quickly.
SPEAKER_03For those that aren't necessarily in the in the technical field or whatever, we're we're I I like the idea of getting it in the hands of the executives for sure, because if they don't understand what's possible, they're not gonna be all in. And and yeah, like they need to be getting all in.
SPEAKER_01Yeah. And it's a little bit different for some organizations. We're a technical organization.
SPEAKER_03Yeah.
SPEAKER_01So the people who are at the top of our org are inherently technical people. For some organizations, that might not be the case and actually might not be the the the route you I would recommend because those might be the people who um lose trust pretty quickly. Yeah, and go, I don't want this. This is cut, I don't even know how to use this, right? Interesting. Um, so it's just got it's gonna you gotta gauge that when you're looking at the organization.
SPEAKER_03So how would an organization identify who that champion is gonna be, or if or if they should? Because there's always uh my experience has been at the executive level, there's gonna be at least one who's like, hey everybody, we we gotta do something, right? But if that one person, and let's say they want to be the the executive sponsor for this effort in the company, they still got a job to do though, so they can't do it full time. Yep, how would that individual know where they should start the the introduction of the conversation without blowing it too soon by giving it to the the Luddite, the non-technical executive who's like, I don't understand this.
SPEAKER_01Yeah, but part of part of our engagement is we really encourage a cross-section of the organization, um, you know, various stakeholders and business leaders. And um usually we we ask them, you know, bring the willing to the room and then see which ideas flow to the top, right? Which are the best ideas? You know, we go through this workshop of use case scenarios, and we we give send them back with some homework and a workbook to fill out, and ultimately um we actually even have a rating scale they can use. Um, find the use cases that um seem to resonate and find the people who are gonna be the ones who are willing to own it and see it through to the end. And again, don't play special, don't play preferences, right? If if uh this team over here just loves to use it, but they really don't have good use cases, well, maybe they're gonna need to wait a little while to get get a little more attention while this group you know actually has proven that they have something they can you know find some really early value out of.
SPEAKER_03So how is your organization distinguishing themselves in the marketplace? Because I'll tell you, when I talk to I meet somebody else that's in AI, oh, what do you do? Yeah, if you heard what what I how I describe what we do, and they heard what you describe what you do, and the guys that I spoke from stage yesterday, it's it all seems like we're doing at a macro level, we've kind of figured out a process. It's those it's the secret sauce kind of stuff. So how how are you guys like how do you how do you what do you think you're doing different than others?
SPEAKER_01Sure. Yeah, I think big differentiator for us is the fact that um we're we have so much capability across so many different um technology domains.
SPEAKER_03Yeah. Okay.
SPEAKER_01You know, we've been doing digital engineering and software design for uh almost our entire uh existence, which is coming up on 20 years. Uh, we have a um in-house security operations team um and security managed service. Um we you know have uh a cloud and infrastructure group that can help you know deploy whatever AI solutions you're looking for. Uh in governance, we have a compliance group. So we have a very broad uh capabilities within our organization. We also have a very deep relationship with Microsoft. So if you're in that universe, and we typically stay in that lane, we I mean our digital engineering team does take on projects that that um get go outside of that sometimes. But um trying to build up our our our capabilities and and our expertise in the eyes of Microsoft helps because we work with their sellers who then can go out and and pitch us to customers. That means a lot of training, internal training. Yeah, yeah. Uh a lot of specializations that we have to achieve and investments we have to make in ourselves. Uh, we also are very keen on telling our own uh customer zero journey so that it's very relatable to our customers. The other, the last thing I would say is we're trying to be very forward-thinking in the managed services we provide today, uh, both on support and security level. Unlike a lot of our competitors, we're starting to bake AI monitoring, you know, consulting right into those products. Right. So that when when customers come to us and say, Well, great, you support our infrastructure. What do you know by AI? Well, we we also can do that, and we can also, you know, be there to support you on that journey. So um yeah, I mean, it's it's it's it's a little bit of everything. It's it's not easy, but uh, I would say thankfully, right now, there is a lot of um interest and a lot of work. So um we're not necessarily having to um you know fight people off or or really kind of get into very competitive situations.
SPEAKER_03Yeah, that's what uh as I say, we're it's not like we're selling insurance, right? Like there's people there's high demand, there's there's they're aware of the problem and they're the the they're becoming aware of of the solution, but they look around and there's just not a lot of options out there. So good time to be in the AI space for sure. What were some of the challenges initially? I again, you guys aren't necessarily the standard company, you're tech forward. Right. Um, you know, what were some of the challenges initially, or did you have any with the the pilot teams when it came to because what what I'm looking for here is was there a like a change management uh effort that was required? Was there a cultural resistance to you know whatever?
SPEAKER_01Yeah, no, absolutely. This there had needed to be a change management process. But what I will say, and I think I kind of referenced it already, um uh not everybody wanted to adopt it the same way, even though we developed use cases and we said, okay, we're gonna bring on one team at a time, develop them, develop three high-value, low effort use cases. Um, people we found people were backsliding and saying, Well, I I use it when I can, but I'm not using it for that thing. So we had to be very intentional and we had just we had to really bake this, these tools and this, the these um cases into the workflows. We couldn't make it an option. We had to say, okay, this is how this job will be performed now. Yeah, I like that. Um the other part is uh I think we probably didn't give enough um attention to how much ongoing training and exposure people were gonna need. So, you know, probably I would say six to eight months into that journey, we decided this is where I kind of came in and actually took that role. We needed someone in a role that was going to be responsible to track the adoption, to to make sure that uh we were providing uh additional support and training that could be, you know, setting up a champion team, uh establishing our AI governance council. Yeah. I'll just give you some idea of some of the things we do uh regular on a regular basis. We do a you know a once-a-week co-pilot tip of the week, which is a new feature capability that goes out to the team. We do an AI news briefing which covers news topics in the industry, but relative to how it affects our organization. So it's very uh very tailored and contextual. Uh then we also do once a month a user group uh where it's a hands-on workshop and employees come in like a hackathon? Yeah, sort of. Yep. So this this this next week we're actually doing a um uh a co-pilot um uh uh treasure hunt, essentially. So we're gonna give them a list of you know 10 different things they can do with these tools. Yeah. And you know, whoever gets to it first is gonna win a prize. But yeah, very cool. And and so um, you know, we just gotta keep doing that. We we also introduced a uh AI fundamentals training that all new onboards have to go through now, which is about you know, three hours of just general AI, gen of gen AI concepts, uh, you know, responsible AI topics, and then very specific user training on co-pilot. And then we assess them and you know, rinse and repeat. So are you guys doing a lot of hiring? Um yeah, I mean we we hire pretty few.
SPEAKER_03Regularly.
SPEAKER_00Yeah.
SPEAKER_03So uh for at the different roles, how are people, how prepared are people coming in with with like how much AI skills are they bringing with them?
SPEAKER_01Yeah, it's it's mixed bag, it really is. And and I would say um we you you can't make um assumptions based on generational um either. Um it's really interesting, especially again, because we hire, you know, or at least I'm involved with hiring a lot of technical people. Um some are very um knowledgeable and like, oh yeah, I've been using this stuff and I'm helping helping me to do this, and um, and others go, yeah, I'm a I'm aware of it, I'm familiar with it. And um it's it's becoming, and we we've mentioned that too to our talent acquisition team, it's become a core competency or core skill that has to be on the resume now. Yeah, right. Um, it's it's no longer uh, I mean, can we overlook it? It guess it depends on the role, but you better be willing to be an exception to learning it pretty quickly when you get here. Yeah.
SPEAKER_03Yeah, the data that we're uh seeing is that um companies are much more interested in hiring you if you have like AI experience that may not even be relative to the role that you're sure but if you know they understand if you know how to use the models, you'll figure out the role or whatever, however, to apply it. Um, are you seeing this experience? So we we partnered with uh a group called Scaling Up, they're professional coaches and consultants, they help companies with hypergrowth, and their their kind of like head of thought is a guy named Vern Harnish. And Vern was recently uh talking to it was a software development firm in the UK, and the experiment that they did was they gave all of their developers cursor licenses, right? Okay, so and for those of you that don't know cursor, it's uh uh it's uh supports encoding by using natural language, it does all the code writing, right? So at the end of I don't know what the measurement was, it was maybe 12 weeks. They had 50 developers, and what they found was that the top five were producing as much code as the other 45. Wow, and what they did was they got rid of those 45, but they paid the hell out of those other five, right? They paid them like five 5x what they would make so that they could keep that talent, right? Are you seeing and I I hear this term and Twitter and all this stuff about the 100x developer because they're using, are you seeing that type of impact yet in your organization with those who are coming in with saying I've been using it for two years versus those that say I'm not using it, but I've been a developer for you know 15, 20 years?
SPEAKER_01Yeah, I I don't have nearly as much visibility into that side of the business, but what I will say is we absolutely have been able to hold the line on uh additional headcount based on the efficiency gains that we see, right? Yeah, yeah so it's not necessarily, hey, we're gonna, you know, uh excise a union or or a group. It's more of we're able to accomplish a lot more and and and actually slow the pace of hiring.
SPEAKER_03So yeah. That's I uh a couple places I've seen recently the the trend looks like it's keep hiring flat and let natural attrition kind of you know trim out the people that don't belong. But AI is allowing them to grow the businesses as long as the people are have gone through a process like what you're talking about, where there's ongoing training and there's you help in identifying use cases and there's measurement of enablement and adoption and all those types of things. So you're saying that for you guys, most of this tracking is being done in Copilot or the Microsoft platforms. Yep, yep. Okay. So for anybody, and you know, I don't know what the percentage is, but I would imagine most of the uh clients that you're looking for and that we're looking for by default, they're on Microsoft. They're a little bit older, they're a little bit typically, yeah.
SPEAKER_01Typically, we've we've definitely seen other situations though where you know we've been brought in to say, okay, give us uh a presentation on what's the difference between ChatGPT Enterprise and Kinkopa, you know, and try to be as as objective about it as possible. Um and you know, we've been able to cut, and I wouldn't say I don't like to use the word displace because I use ChatGPT myself as well, right? Um, but but we've been able to um illuminate for that customer that you know if you're looking to try and control the risk uh and can and be able to make sure that you can have more governance and and also keep your data inside your your organization, you know, um the Microsoft tool the platform is is it's hard to beat. I mean, especially when you look at how the security tools integrate. Yeah. Um a lot of organizations are going like, well, I don't like Microsoft's LLM. I'm like, well, you they don't have an LLM, they use OpenAI.
SPEAKER_03Open AI.
SPEAKER_01Yeah. ChatGPD 5.4 dropped today. I refreshed, I refreshed my Copilot chat window, and sure enough, there's oh there's ChatGPD 5.4. Same exact day. That's how that's how tight their their ability to launch these models is. But also, too, you you can use other models. Uh Microsoft's now opening it up for Claude. And if you get to the point where you're using something like Azure OpenAI uh OpenAI Foundry, you can swap in any models you want. So um really at the end of the day, it's it's you're to you're really having a conversation about the interface more than the back end technologies. And um, you know, and and there's definitely some um legitimate uh complaints about some of it. But I will tell you, if if you don't like something, wait 10 minutes, and typically it's being addressed. Um, you know, that the product as it's as it stood two years ago and what we're do what you're using today, it's unrecognizable, it's absolutely unrecognizable from what it was two years ago.
SPEAKER_03Even man, if you think about it, like before Thanksgiving, uh it was still Gemini 2.5, it was you know Chat GPT 5, it was Claw, like and just since then, and now people are asking about I was at that event yesterday, and people who don't even have you know a plan were asking about uh open claw. It's like man, y'all need to just relax a little bit. Yeah, yeah, yeah. Let's get a GPT license first, huh? Yep, yep. So when you have um clients that say, I understand there's risk associated with being maybe more on the cutting edge of uh testing out things like a especially autonomous agents and that sort of thing, because it it's I'm I mean, I've been in the space, you're in the space, like we know things move fast, but I am blown away by how quickly people have shifted from like understanding how to use the the large language models in their day-to-day to immediately like agents, agents, agents, agents. Like that's like I'm hearing that all over the place. And this morning with all the chief AI officers in our community, we had a conversation, and that was how do we address the the the shiny object chasing with an appropriate security posture for these agents, particularly something like open claw, clawed bot, moat bot, whatever you want to call it. So how are you guys not saying, you know, pouring cold water on them, but still managing the the because if you don't test it, man, I think there's a big gap there for a big miss for a company if they don't leverage agents, but there's that big safety and yeah, you know.
SPEAKER_01So so we we um you know, we we want our users to experiment. We we but we want to do we want to do that safely and in a controlled manner. So, you know, uh we again leverage the Microsoft security tool stack pretty heavily. So we have the ability using Microsoft Defender for Cloud Apps to have visibility into what people are using uh as it relates to cloud-based applications. And Microsoft actually can rank those those sites and those services based on a very of the various set of metrics, privacy controls, SOC compliance, and give it a rating score. And so what our CISO has done is if uh a particular uh AI application does not achieve an eight out of ten or better, we automatically uh block it, right? Nice if you want if you want to make a case to the to the to the CISO about why you think you need access, right? But even with that, even when we're we're allowing anything that's an eight or a higher, we're still keeping tabs on what it's being used for and making sure that it has a business purpose, right? So if we see something that's you know unrelated to our business, we might block it even if it is an eight, um, or give someone a chance to explain why they feel they need to have access to it. We're also making sure we understand what's going into those systems. So looking at the prompts, looking at the the documents, um, being able to see what's being shared.
SPEAKER_03Uh and how are you how are you seeing that on the back end on Microsoft?
SPEAKER_01You're able to like prompt samples or extension, it's an extension to the browser. And because it's an extension to the browser, that's all being captured when you go to those sites and see what's being uploaded, you know, what types of prompts are being put into those systems. Um and then finally, um, Microsoft came out with agent 365 uh during Ignite, and that is an incredible control plane which can really give you visibility into all the agents that exist in your environment, who created them, how many people are using them, what they're attached to, what what what data is going into them. And there's a kill switch. If you think there's an agent that's not, you know, um uh is is going outside the bounds of your cover, yeah. Um your uh uh usage policy, just kill it. Um you can then layer on top of that too, is um the data loss prevention uh uh stack and make sure that you know sensitive data is restricted from being put into these tools as well. So uh so how does that work?
SPEAKER_03So if if somebody wants to, you know, maybe there's no malice, but they're copying and pasting IP or something and they want to use that in uh a web-based access to copilot or whatever, how does that work? How does the model say, wait a minute?
SPEAKER_01If we have um automatic labeling turned on, right, that that document will be detected with having a certain sense of information type in it. Label gets placed on it, that label then comes along with an information protection policy, which determines how where and how it can be used. So we have uh one label that we use, which is restricted label. That restricted label literally means you can only share it with internal individuals that you've given explicit access to, no matter what. If I move that document to another document library where the permissions are different, doesn't matter. You cannot access that. If we tell uh copilot you can't use restricted data, if I try and prompt and ask to use, it's gonna say, don't have access to that. Sorry. So even though you may have permissions to it, it's not allowed to access it. Or again, in the case of using a third party application, if I go to attempt to upload it, It's going to be restricted. Or that system won't even be able to read it because it'll be encrypted.
SPEAKER_03It makes a lot of sense. And I declare ignorance when it comes to Copilot. Just a lot of our clients, even if they're in the Microsoft environment, we still work with the other models with them.
SPEAKER_01Yep.
SPEAKER_03But I'm not seeing that level of like control on what's being uploaded into the tools in the other models. Now I and right now I'm trying to think how would you even do that? But if that's baked in with Microsoft, that's very comforting for sure.
SPEAKER_01Yeah. Yeah, that's that's really, I think, you know what where the advantage lies. Um I think you're right. It's just the security stack and the um the control tools that uh and and again, like I said, that extends to third-party solutions too. It's not just controlling Microsoft stuff. It's able to actually, you know, uh take advantage of the security tool stack to to put some controls around, you know, using these other um third-party services.
SPEAKER_03So let's switch gears a little bit. What are you what are you looking forward to the most? Or what are you most excited about with what's happening with the tools right now? That's that's a great question.
SPEAKER_01I there's just so much. Um, you know, I've ri I've really kind of leaned into a lot of these things. Um I'm a big fan of home automation. I I've really automated the heck out of my house. And so um I'm really excited and a little nervous about seeing the convergence of robotics and AI.
SPEAKER_03Yes.
SPEAKER_01And what what will be possible there? Um, I I love to I like to think of myself as a little bit of a futurist. Um, as growing up as a kid, I I really wanted to believe that someday I would have a robot that did all the things that I didn't want to do around the house. Um, you know, I'm now in my 50s. Um I'm hoping that that reality comes to comes to fruition be before I uh you know take my last breath. Um so and I and it's interesting because um you know the you cannot uh uh untangle these these these tech these um these industries, right? They're they're they're they're really um uh are are advancing together. Um and it's exciting to see that, and it'll be exciting to see what that potentially could bring.
SPEAKER_03You know, we do a lot of work in the construction, commercial construction space with those clients, and yeah, my chief AI officer, he's a bit of a futurist as well. And that was the he was like the constraint for them is people, skilled labor. Let's go talk to Elon get 5,000 Optimus robots and transform certainly like an industry. So I I I totally agree with you. And we priced a few robots. We there's some Chinese models that you could get for 3,000, 5,000. Now, what do they do? I don't know, but um it's getting very accessible. Like that's what an electric bike bicycle costs, right? Like, yeah, right. An e-bike or a robot, that's gonna be a tough choice, I guess.
SPEAKER_01Right. Yeah, can it give me a pick? Can it pick it give me a piggyback ride?
SPEAKER_03Yeah, probably. Hey, so um for uh for those who have resonated with kind of the approach that you've shared and how you guys are looking at things and they want to find out more, they're in the Microsoft ecosystem. They're what's the best way for them to find out more about working with ProArc?
SPEAKER_01Sure, yeah. The best way is just uh to go to our website, uh www.proarc.com. It's arc with a like ARCH, not ARC. Um, we have lots of resources on there, um, free resources, white papers, um, we have webinars that we've listed. We actually have an AI return or a copilot return on investment calculator. So you can kind of plug in and say, well, if uh I get this many assisted uh hours with co-pilot, where's my break even? If I pay my user, my workers this much, how much do they have to use it before it actually pays for itself? Um, we do we have um a lot of different blog posts on there. Uh and um we've just recently also with our security team started some office hours where um it's been very, very uh well received with our customers and they come in and they just ask questions and to our SMEs. Um but uh yeah, and if if anyone's interested in following me, um Jim Spignardo on on LinkedIn. I'm the only Jim Spignardo down there. Uh and I write about three articles a week on AI and other business uh IT related topics. And I do also have um, at least I hope I do, uh, a uh ebook that I'm gonna be releasing for free in March uh called The AI Turning Point, which is gonna discuss kind of less of the technical aspects of how companies deal with the uh changes that AI are bringing to the the workforce, but more of the business uh process and and how the things are gonna have to change within businesses in order to support this this transformation.
SPEAKER_03I'd love to get a copy of that. And for all those links for the listeners, that's gonna be in the show notes. Um and it just, you know, for for as a listener, I want to share with you that the way that Jim has approached this is very prudent. Like there was nothing here that I heard where I was like, well, I disagree. This is all um really solid stuff. So I would suggest that if what you heard that fits with what you're looking for, um especially with the the depth of technical skill that they have to be able to do more than just the AI side of things or think about AI at a lot of different levels, not just the human level, but also the technical capability level, definitely plug in with with what Jim's up to. Jim, thank you so much for uh taking time out on a Friday to um have this conversation with us and share kind of the way that that ProArc is approaching this and that you got a fantastic role there. That's a lot of being that person and that company has got to be a lot of fun. So thanks again. My pleasure. And for everybody, we'll be uh releasing another episode as we always do on Mondays, and be prepared. We've got the uh the hundredth episode is coming up in April, so we'll be doing something very special for that. And thank you so much for listening. And if this helped you or if you thought that this kind of demystified some of the AI conversation for you, please feel free to share this with uh those on your team or anybody that you know think could benefit from this. So we'll see you on the next episode. Thanks everybody. Go use AI. 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. It'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.