AI won’t replace your team — but it will make them dramatically more effective.
In this episode of Entry & Exit, Stephen Olmon and Collin Trimble break down how they’re using AI to scale a security business — from simple tools like ChatGPT to fully deployed agents handling quoting, support, and operations.
From custom GPTs for sales and tech support to an estimation agent that turns plans into full quotes in minutes, this episode shows how AI drives efficiency, better customer experience, and margin growth — without cutting headcount.
Inside this episode:
→ Why AI is both overhyped and underused
→ The 3 levels of AI adoption
→ Real examples across sales, ops, and finance
→ How one AI tool reduced multi-day estimates to 5 minutes
→ Why focus beats trying to automate everything
🔗 More Entry & Exit — entryandexit.com
Connect:
Stephen Olmon — http://x.com/stephenolmon
Collin Trimble — https://x.com/TXAlarmGuy
Owned and Operated
New Episodes Every Wednesday!
Subscribe For More
This is you can't chew gum on the episode. You can't chew gum. No, you can't. You can't. No, you look like a child. We don't invite children on the podcast. Little Johnny with his gum. You can park it. You can park it. I am happy, actually. I have Joy Joy down in my heart. You can park it off to the side. I'm gonna park it off. Park to your cousin. Park to our park your HR manager. Yeah, I can't do that. Yeah. Editor, you can totally use the last 30 seconds for a little bit of a fun, fun bit if you want.
Feel free. Okay, we're gonna we're gonna start. Okay, I just need can we just have three bullet points? So we're we're doing map to how we got here.
What I would say dip your toe in the pond of AI, start with this, and then this is what we're doing now. That's kind of advanced level stuff. And we'll and we'll get very light preview of the advanced level stuff.
Okay, say it one more time.
How we got AI pill, like where we were what we were doing, kind of what in the beginning, where to start, and then the advanced level stuff. And I think also like I'm gonna start with our view of AI in terms of like it's not necessarily a people killer, it's a job enhancer. Enhancer, yep, I agree.
Okay. Uh, peace and blessings. Uh, what are we gonna frame this as? How to grow your business with AI? How to scale, how to scale your security how to scale your security company with AI. Yeah. Yeah. What do I hear in the background? Is that Connor talking? Someone's talking extremely loudly. I can hear it. Okay. Um, okay. Boo-boop boop boo-boo doop-boop. Um That's all I can hear. That's all I can hear, is that voice. Um, here we go.
Welcome to entry and exit. My name is Steven Ullman. My less attractive co-host and business partner, Colin Trimble, is also with me. Um we're gonna talk about how to scale your security company with AI. We're gonna start out talking today about how we got AI pilled.
What color is an AI pill? Black. I don't think I want to take a black pill.
Well, I couldn't say red or blue. That's too like matrix-y. I don't want to like have copyright infringement. Yeah. None of that.
So
I want to start with uh our sort of hot takes on um, or at least our viewpoint of AI and kind of the state of it. We are extremely deep into AI. I would say we are probably top 10 in the nation for security businesses doing things with AI and our business. It has not always been that way. Um and our philosophy on AI is this we believe that AI helps drive better customer experiences and make our employees more efficient. We are not focused right now on replacing employees. That's not our that's not on our roadmap. Um we may not need to add net new employees as we grow because our existing employees have become so efficient that we don't need new employees.
Yeah.
Um, but we have no plans to let an employee go. And and a lot of these people that are saying, oh yeah, I have an AI running my business from top to bottom, you know, I have a whole org chart of AI agents is way overblown, is an edge case. Um however, the fact that people don't think that AI can answer emails, even take phone calls, help do your books, help do scheduling, create quotes, and the fact that you that that is not bleeding edge. That is available today. Right. And so I think it's almost both simultaneously overblown and also way under the radar, and I think folks in our industry should know that. And I think that it's a great way to help drive better experiences and unlock and debottleneck your existing employees. Do you have anything to add to that? That's sort of the good framework of like our viewpoint or philosophy on it.
Yeah, I think that's perfect. We we talk about like human flourishing, like this is not about how many, it's not like born out of greed, oh, how many people can we eliminate? And yeah, you know, like we do think that using these tools um in in the right ways will lead to gross margin expansion and ultimately hit the bottom line. We do. And we're already seeing that a little bit. Um, but it's not born out of this like excitement to do away with parts of our team.
Yep, totally agree. And we mean that, and like we mean that because uh not because uh we're trying to be like honorable or whatever, but that's part of it.
But the bigger part of it's technically I I'm I'm honorable.
You may not think, but yeah, yeah, well questionable. You were making bird noises before we started, so I think people should know that. Um I uh I I'm saying this because the technology is simply not there today, and I think that it's and it's never gonna replace a field salesperson. It's also never gonna replace a technique. Why are you still laughing?
You just got me with that one.
Um sorry, continue. Um I don't think it's gonna, I mean, it's robots are not gonna replace technicians and they're not gonna replace field sales guys. In fact, I think the human experience is gonna be more important. Trade shows, cold calls, way more important. Cold drop-ins, face-to-face contacts, site visits, yeah. Coffee drops, all the things. Um, okay, so a couple things.
Where did we start with AI? Well, we started, like probably most of y'all, who if you have any experience with AI, which is we had a chat GPT subscription that we put um, how do I make this email sound more direct and professional, and then copy and paste it or you know, whatever. Like, did that, we had it, you know, rewrite some stuff for us, do some research on some things, answer some questions, um, but didn't really think much past that in terms of like uh what can it do for us in a huge way to like scale our team until really about six months ago, maybe nine months ago, um, we had hired a bunch of new sales reps and they needed, they were outside the industry and they needed like product questions, like, hey, if I take a picture of this reader, does it work with Brevo? Or um what kind of camera is this? Is this you know an ant a PTZ or what, you know, whatever? And so I built a on chat GPT, they have something called a custom GPT, and you can basically create a chat agent that references uh knowledge articles and uh information that you create. And so that's what I did. I created, I uploaded some Revo spec sheets and um and I added a bunch of truth like data sheets, and I verbally talked to it to say here's some stuff that you should know. And I created this agent that could answer basic questions for our sales team. Um, it could give basic understanding of topology of an access control system. Oh, this is a six-door system, you should use two three-door controllers. You know, oh, it's a 20-door, you should use a 6,000 panel. Like it could do basic stuff. Um, and it helped unlock my sales reps from having to ask a technician or our ops people just basic fundamental questions and also kind of saved a Google search, kind of sped up their process, helped them get up to speed a little bit more. Um anything to add to that? That was kind of the first thing that we really like released.
No, it's um so lots and lots of people haven't even touched a single AI tool. That first little wave is typically people using Chat GPT like it's Google. So they're still just doing the same type of search they would because they don't know any better. We didn't either, right? And then that next kind of iteration is what you spoke to where uh you're um trying to use AI to just improve simple things like help this uh help me, you know, write this email to a customer that's angry at me, you know, with empathy and you know, sound professional, right? And so that's like your first little taste of oh, AI's helping me be better at my job. Oh, yeah, that feels good. Then you took that step to actually build something net and new, kind of like a level two tool, right? That custom GPT that other people could interact with, and that is kind of that intermediate step where I think a lot of people are interested in trying those things right now. I agree, but then there is kind of what I'll call like level three, level four, where we have started to graduate towards, and I think that some of the things that we'll talk about in a few minutes most people don't think are realistic yet, but they are.
I think, by the way, if you want to get started, like just quick, very quickly, I uh if it if it were me starting over, I would go sign up for a Claude subscription, and there's something called Claude Projects, which is basically the same thing as a custom GPT. Yep. And you can create a project and you could call it uh tier one tech support and have another Claude task, just like a normal chat window, and say, hey, find all of the spec sheets and all the uh installation manuals for these uh manufacturers, and then I would take that and attach it to the Claude project. It's literally a place that says upload files, yeah, upload the files, and then you can give it like a like some guardgrails, meaning you can say, hey, here's what you're supposed to be. You're a tier one tech support for me, my company is this, and then you could literally deploy that to your people today. And that is a tier one tech support agent. Think like the customer calls in and wants to bypass a zone, or they want to know what this error code on their fire alarm panel is, or uh whatever. You could have them tell that to the folks answering the phones for you that maybe aren't the the technical folks, and now they're they're just a little more equipped, right? They're just a little more equipped to answer questions. Um, you could even take it a step further and get all the installation manuals and everything and create like a tier two support one for your technicians to say, hey, I'm getting this error code on site while I'm installing. That would be another example of like a more knowledge, right? And so that is a very easy thing to do that takes signing up for an account and four or five clicks. And so if you're starting dipping your toe in, that's where I would start right now.
And here's a funny little addition to this. One natural question might be, okay, well, what if I go do all that? How do I give people access to that?
Yeah.
And you know what? Just ask AI. Like, we've gotten stuck on things at times, and like, well, did you've done this to me a lot. You've been like, well, did you ask Claude? Did you ask Perplexity? Like, did you go back and process that? Because a lot of times those platforms and technologies will walk you through step by step whatever problem you're facing. And so it is kind of this funny loop, you know. And so it's like, well, I don't even know how to do that. Exactly.
Yeah.
You're about to. You just have to ask. You know, it's you know, and then all of a sudden, you know, you're touching databases and APIs, and you feel like, you know, you're you know, the valedictorian of your high school class, even though you didn't, you know, pass hardly anything. So like you you automatically kind of become capable of things that you weren't prior. So you don't have to think like, oh, I have all these limitations. I can't do that. I don't know how to do that.
Yeah.
You're about to. If you'll just try, if you'll just ask.
Yeah. I will say this. Um, there's a lot of ways you can um you should talk to AI about how to deploy it to your business at scale, but one way is giving feeding it context on your business so it knows like your org chart, what systems and tools you use. The more information about your business, the better. Um, and that way it can it can uh give you better results. So, anyway, um that's where I
would start. I do want to move into like where we are today. So we we basically have been fully AI pilled today, and we are doing, we're really we're I mean, we're basically on the cutting edge. Um you you may have heard of terms like open claw, which open claw is there's a lot of um uh what we would call agent harnesses that are basically agents that live on a computer. And all it really is is uh it's an application that sits on your computer that you can't see that's wrapped around an LLM, which is uh a model, an AI model, and it basically does stuff for you. And you can give it unfettered access, which silly people do, they don't put any guardrails on, and it's like racking up a massive API bill and sending text messages to your mom that you don't want to, right? Like it can it can do all sorts of really stupid stuff. You can also build it and grow it and have it do awesome stuff and give it guardrails to say, hey, this is what you're specifically gonna do, and it can even self-learn. So we're using a version of OpenClaw. We're using something called Hermes, which is from Noose Labs, which is a startup that's an open source. Shout out to those guys. Shout out to it. And they've released an amazing product. And this is very technical. This is, I would not even touch this with a 10-foot pole until you've spent uh if you have some technical chops or have somebody and and you really know what you're doing. I put mine on a dedicated Mac Mini that's its own container that it's isolated. And then we started slowly hooking up tools to it. So now, for example, we just released our very first production full front-end agent. We've got a website for our staff that they can log into, and it shows them uh what we call Eddie the estimation agent. That was Stephen's uh love of is that alliterations? Alliterations, yeah, you know?
Yeah, Casey the collar, Eddie the estimator, yeah, whatever.
Yeah, and so um Eddie uh what he does is in natural language, you can describe literally anything. Like I uploaded some plans, architectural drawings from a church, and I said, hey, it has full access to NFPA 72 from Fire Alarm, and it went in and looked up the scale, counted the doors, did the candela rating, did the voltage calculation, and gave me a full bill of material using my preferred manufacturer, made sure I approved it, and then pushed all of that into Salesforce as a quote, fully formatted and ready to go for me to click one button, which is generate a proposal and send a customer. Um, that whole end-to-end process would have taken us time. We'd had to get an APS involved, had to get the drawings over, there would have been back and forth, had to, you know, look at the uh look at all the different margin profiles. Now it has all of our margin profiles what we want to do. And so from end to end, something that would have taken multiple hours and a couple days to return back to the customer took five minutes. So again, that's not gonna replace our estimator, far from it. It's just gonna make him even more efficient. And what's even better is it's gonna free up him to have time now to figure out how do I make these more competitive bids and get more creative and figure out how to value engineer? Because that's ultimately where you want your estimator to be is value engineering solutions.
Yeah, I think that these types of solutions allow us to take people that we trust, that we've invested in, that are really talented, and push them into their highest and best use constantly. Right? Like it's like if I can have you know these people that we think are just so awesome, completely removed from tedious, repetitive tasks, um things that you know really are inefficient, then that means I get to use them in just the highest value activities all day long. Heck yeah. Sign me up.
Yeah, I agree. So so we're scaling those agents. We're working on a voice agent right now for our after hours phone call for routing to monitoring centers. Um we're doing some really awesome stuff, some cutting-edge stuff that I think is gonna really benefit our business. Customer, customer portal. Oh yeah, we we use uh have a fully AI-enabled and managed customer portal so they can do self-service, everything from booking appointments, looking at invoices, paying invoices, submitting new payment profiles, whatever. Um, so we're we're pushing a lot of stuff um faster than we ever would have been able to before. And this just took incremental progress. Like once you roll out that first claude project and you're like, oh, this is cool, and you're gonna start to dig around a little bit more and learn a little bit more, and you'll see some things that you like about it and some things you don't like about it, and you'll do you'll do research. And how will you do your research? You'll talk to an AI agent, and it's gonna it's gonna give you really thoughtful, well-researched answers, and you're gonna try some stuff and you're gonna fail. And the industry is gonna continue to move forward at a rapid pace. And so um, yeah, we we're doing a lot of fun stuff.
We are gonna spend a lot more time on sort of this what are we doing today and what's our roadmap um and some of the AI stuff that we're building because we get asked about it all the time. In fact, we released um, we actually put a preview of Eddie, the estimation agent, on LinkedIn and on Twitter yesterday just to show folks like, hey, this is what this thing looks like. We like to build in public, it's not perfect, but it's really, really close. And uh, it's awesome.
We're we're doing other things too, things that aren't as involved, like we gave it access to our accounting system and to kind of act as like a virtual CFO to give us, you know, um trends and highlights, concerns, flag things, you know, that are outside certain bounds. Like um even on the sales side, uh we're we're doing some creative things with public data sources to try to find you know, kind of fresh leads that might be relevant to us. Like there's all sorts of ways that you can scale your business. And I think you have to come up with some of those ideas, and maybe even this has given you some ideas, and then start with one. Like I wouldn't tr you know scatter shot, try to do you know, things across every single department all at once. Like, I would pick one, focus on it, kind of work the problem. You've done a good job at that, kind of staying focused on. I mean, you spend a lot of time on Eddie specifically, and so um that would be my encouragement to you is like yes, get familiar with the tools, but then just pick that first problem or that first opportunity, if you will, um, and build from there because as you do that, you're gonna learn so the next thing you go try to tackle, you have way more understanding um and context for how to go about that the best way.
Totally agree. Totally agree. Um, if you like this stuff, let us know what more you want to talk about, what things you think um or questions you have about AI in the security industry, life safety industry. We'd love to talk about, or even just in the service based industry. Like we love, we think it's we think it's relevant to all that. So leave us a comment while you're there. Hit the subscribe button if you would. You know, let us know. And man, we appreciate you listening. Thank you so much.
