Your technicians don't need to sell more, they need to collect better information.
In this episode of Entry & Exit, Stephen Olmon and Collin Trimble share how they're using AI to improve service operations at Alarm Masters. From automated technician workflows and smarter scheduling to uncovering new sales opportunities, they break down practical ways operators can increase efficiency and profitability.
In this episode:
→ Capturing better field data with AI
→ Automating technician notes and workflows
→ Improving scheduling density and utilization
→ Predicting job duration with historical data
→ Turning service calls into future sales opportunities
→ Why technician selling is the wrong goal
Connect:
Stephen Olmon — https://x.com/stephenolmon
Collin Trimble — https://x.com/TXAlarmGuy
Owned and Operated
New Episodes Every Wednesday!
Subscribe For More
There's no AI agent for attitude. Maybe we should create an AI agent. It'd be nice.
I was about I was literally just thinking, man.
There's people that say if you could solve the technician selling problem, you'd be the million dollar man. That's what I used to hear all the time. It's like, yeah. I don't know. To just collect information. Because if I can bring the information back here, then I can use AI to outbound those customers on, hey, we just did a technician visit. Correct. We heard that your video system may be a little old. If you had that automated, you don't need your text to sell you. You need them to be undercover agents.
One final note. I was a little self-conscious when I spelt slippage earlier. And I was like, oh man. Did you look it up? Did I did I misspell that? But no, I didn't. Because I got that dog in me.
Welcome to Entry and Exit. My name's Steven Ullman, and this is Colin Trimble, who is my fearless partner, business partner, to be clear, my co-host uh of this podcast. And uh we run Alarm Masters, which is a life safety company headquartered in Houston, Texas. Beautiful Houston, Texas, as I said. Third Coast represent. And today we're talking about how AI is or can transform your operations.
Yeah. I want to talk about some stuff we're doing and some stuff we haven't done or we're trying to solution for. Um as it relates to the service department in AI. So we are sold out on service in our business. It is only incredibly important to us. It's profitable, it's predictable, it provides leads a lot to our project department. And so we're always trying to figure out how can we make it better. And so there's sort of three areas in the service department that I think everyone is trying to solve for in some capacity. And we should like, I don't know, just talk about it. I think we're really working on two of them, and one of them we have not really spent any time on. The first one is um information from the field, so getting notes, pictures, communication from the customer back to the office to action on. The second one is scheduling efficiency and/or scheduling density, so like less drive time, less higher utilization, that type of thing. And then uh the last would be, and I just threw this in there because we didn't talk about this before, but I thought about it as you were doing your really great introduction. Getting your technician to sell for you. So getting your guy at the customer to sell the customer. Yeah. And like how do you enable that? And so we should just take those one by one because I think those are interesting topics that folks listening, if they're doing service, are probably thinking about. And nobody has done this perfectly. I think there's a lot of people that have done it really well. I I would say we we have done a, at least in the last year, a really good job improving in the first two areas. Um, but it's not perfectly
dialed yet. So uh bullet number one is probably the one that I care the most about, to be honest, which is getting information from the field. It seems like we always are having issues like, hey, send me your pictures, send me your work order notes, like you know, what did the customer say to you? Did you ask them, is there anything else we can help with? Is everything okay on your account? Like and that seems to be uh ongoing.
Yeah. I mean, some of those things um we we need our uh technicians to do, but that may not be a natural um action for them to take, or that some of that some of that kind of uh questions or leading questions, or some of it's just not maybe uh what they feel is like uh normal or comfortable. Comfortable is probably a better word, but it's stuff that we really need. One other thing that I'll say um we talk about all sorts of different topics. Some of them are really security specific, but I actually think today what we're talking about is applicable to a wide variety of businesses. I mean, any home service business, any kind of B2B sweaty commercial service. Like if you're rolling a truck, this is all relevant to you. This is really not security specific. Yep.
Yep, I agree. I think there's a lot of so you a you want information back from the customer, right? So it's like at a base level, it would be hey, uh, does the customer need a quote? You go out there, keypad's beeping, they need an entire replacement because they've got an inner logics panel and they need to upgrade. And obviously, you don't have the parts on your truck, and so the customer says, send me a quote. Well, that's information that you need to get. Maybe your technicians in the field are doing that. We do it in our office and then follow up with the customer. Another one would be you go out there and you need to replace a part, you don't have it, so you got to come back and do a reschedule. Uh, another one is the customer made a complaint or you used X amount of parts, or you noticed while you were out there that there was a different trouble signal, the customer declined a service for an additional thing, like whatever. You need those parts, though that information. We do a lot of acquisitions and we very rarely get the panel information from the prior owners. And so it's really helpful if uh the technician notates or takes a picture of the system so we know in the future what are we working with and what are we working on. Um, so all of those things are important. Another thing is just like we make our technicians have the customers sign a service report, which shows the charges and just kind of brief explanation of what they've done. Customer signs off on it. Um, so these are all like steps in the we will call the service ticket flow. And we ask a lot of our technicians. Like it's not like it's it's not like just hey, take a couple notes on a piece of paper, sign off, and walk away. They have to put in their line items. And so there's a lot of gaps, and we've had technicians in the past. I would say our current group is is really pretty good actually about putting in the notes. Um, but we've had techs in the past just not do it. And it's like that's at a baseline, nothing's gonna solve that problem. There's no AI agent for attitude. Um maybe we should create an AI agent.
It'd be nice.
I was about I was literally just thinking, man. I need I need one for my uh almost eight-year-old's attitude. I need somebody to like coach and train her. We've she's got a lot of opinions, that girl. But okay, so we've
got these tickets. So one of the things that we're working on is an AI agent that um effectively uh forces them through it's really less of an AI agent. There's a little bit of AI in it, it's more of an automation that forces them through the steps of the work order in order to close their ticket. They have to complete all these steps. So previously it's like, hey, complete these fields, but you could really close the ticket whenever you wanted. Now it's like, hey, you can't close this thing out and move on to your next job until you do these various flows. And I think that that's really important to do. And actually, uh, we aren't doing this because Salesforce kind of does it for us, but I actually think there's a way, like if you're using a separate uh field service application, I think there's actually a way that you could use like ChatGPT or some other agent where you could get have the technician give verbal notes and then it could loop into whatever field service software you're doing and populate those notes into those fields for you. Uh Salesforce field service is pretty robust, so we can kind of do that without that. But I think there's actually a way you could do that. But the other way we're using uh AI is like they put in their work order notes, which then flows to the service appointment and to the invoice. So one of the things we're doing is like they're kind of just like they're just kind of free flowing the thoughts out of their brain. And sometimes those thoughts don't aren't insanely coherent because they're just trying to get them down. And so the customer's like, what does any of this mean? So we actually use an agent to kind of overwrite their notes with um clean it up, clean it up, exactly. And so um that's one thing that I think is been super helpful for us is to kind of force them through that. And then it also looks at their notes and says, Hey, it asks some other questions in the automation based on what they put in the notes, which I think is pretty slick.
Yeah. Have we named that agent yet? Like, you know, we haven't cleaner, yeah.
No, you need Carly the cleaner. Um, I just no, we we haven't uh around Steven's office. Like, if you're watching this, like it looks so perfect, but if you look at the other side, there's a lot going on.
Okay, so well, I also like I have like a board with like all my kids draw my kids. Love giving me stuff to take to my office. Like all these like I'm like like random stuff, rainbows, and like yeah, yeah, it's there's a lot going on, you're right. But you know, we keep it nice and tidy for for our viewers.
Yeah. Um, anyway, so no, we haven't named it. It's really more of an automation, the LLM piece. We're gonna expand on that, but that's that's one thing that I think is really important. It's like, hey, get information back from your technicians because they're the tip of the spear, and so they have information that you should want. Like, also, we haven't done this. This kind of leads into bullet point three. We'll come back to this. But um, hey, if you're out there working on the alarm system, like look at those cameras, look at the access control system, snap a picture of it because it's it's probably not ours, and we would like to take that over. So um we'll get to that kind of at the end because
I've got some thoughts on that. The second thing is scheduling density and higher utilization, right? So utilization is just simply the amount of hours you're paying your technician and the amount of hours that you're charging the customer. And you want the amount of hours you're charging the customer to be at least 100% of what you're paying the customer in terms of hour for hour. Um, and not for dollar for dollar, or you'd really be in trouble, but hour for hour. And so, you know, uh people that are doing it best in class are over a hundred percent. We are not there. I think we're somewhere in the 85 to 95% range and we need to be improving. Uh, so how do you get over 100% would be like, hey, if you're rounding up, like you round up on hours. So if like you're for us, any we round up to the half hour. And so if we spent five minutes, that's a half hour, but they may not use that whole half hour and they drive to the next job. So like high volume service companies uh can round up. Uh, what's really hard about scheduling is you're working with a lot of variables, you're working with the technician skill, the customer preference, the geography, the next ticket, and also this like how long is that ticket gonna take? And that's probably the hardest one to work with because like skills and geography, drive time, those are definable and quantifiable. But if a customer calls in and says my keypad's beeping, and the person taking that service request tries to schedule it based on that without asking any additional information, that could be a one-hour call or an eight-hour call. Uh, and that's usually the hardest one to manage. Um, so we built an agent. We did, we did name this one Seth, the scheduling agent.
That's right.
And um this agent intakes information. So it as the rep on the phone is taking the customer's information, it like type you type in the fields, and then we have Google Maps API on there, and then we have the technician skills. But then what we have that we really love is it indexed like our last hundred our most common 150 uh work orders and it looked at the like notes from the person and the transcript from the customer to basically index on like how long did that actually take and how did they describe that problem. So it's not a perfect data set, but it gets us really close so that we can kind of pack in the schedule and it will make a recommended slot on best available, or you could type in preferred window, or you can even say, like, this is an emergency, I need like the next available slot from a technician. And what's really cool about that is it will actually surface hey, you can have Bill at one o'clock, but Bill's only available till 2:30, and Bill's next appointment is two hours away from there. So, like, if this is a strategic customer, you know, send Bill. Otherwise, you're gonna have some pain. So that's another that's another thing we've been doing with density.
One thing to add to that is you know, lots of different people listen to this. So the larger the market, the larger the operation, the more critical this is. We felt this day one. Um, you know, over three years ago, we we felt this some, but as we've grown and as we've you know added to our technician team, this just becomes an even more critical area to be as efficient as possible. When if you're in like a secondary or tertiary market and you've got two techs, you know, that's that's gonna be pretty simple. Um but the moment you start going three, four, five, six, seven, ten techs, and you're you know, for us in a one of the largest cities in the country, this is huge. I mean, like it we we have we can have significant uh uh margin slippage if we aren't on top of this. Huge slippage. Yeah, big time slippage. That's a great word. Is it a word? Slippage? Yeah, S L I P P A G E for my spelling bee friends. Um yeah, you're a spelling king.
Yeah, spelling king. Um yeah, I agree with that. And I also I think that it's uh the hardest thing you need to figure out, and again, this is where AI really dials in is like look at what the customer is describing as the problem and the questions that the operator rep is asking, and then what was the end product of that? Like how long did that take and what was involved? You can use that information to make the appointment more efficient. Because, like, one thing that it's been surfacing to us lately, or that we've been working on, is like finding parts. It's like, hey, by the way, on these type of calls, they typically need a bail-in. So make sure the technician, like we put it, it just surfaces it in the internal
comments. It's like, hey, before you go out, make sure you've got an extra bail-in because you may, you know, you may need it for a you know the video. Um, okay, so the last one we haven't solved for, and we're gonna spend the least amount of time on it because we haven't solved for it, is how do we get our technicians to sell more for us? Um and I don't know that there's like a way that you could get an AI agent to force your technician to sell for you, but I would like to position a reframe, Steven, if you'd be if you'd be so willing to hear me out.
Yeah, very I'm I'm open-minded.
Well, I there's people that say if you could solve the technician selling problem, you'd be the million-dollar man, is what I used to hear all the time. It's like, yeah, I don't know.
I've never loved that. Like and and for just let me add something else in here that is highly related from a marketing perspective. Hi, me, Mr. Marketer. Hi. Reviews. Like, yeah, it's it's the same thing. It's the same issue. Yeah, it's the same problem.
Yeah, that's almost a different deal to me, but I agree with you, and I don't even want to talk about that because you can get me going. Um I'll get you. But what I you do, what I think about is hey, you're and we've had technicians sell projects and we have paid them commission for those projects. Uh, and and you know, they're not usually huge, but great. Like I reward the technician. We have a commission structure, hardly anybody takes advantage of it. We track it, we celebrate it, we kind of do the things. Could we be better? Probably. There are people that have done a good job of getting their technicians, especially in other industries. Um, ours is a little bit different. I don't know. There's I just haven't heard of anybody that's like absolutely nailing this. What I've decided though is like the more important thing is less about them selling and more about them collecting information. So, what I really want to do is put in their work order ticket, like a section where it's like go take a picture of one of their other scopes of work. So, like, take a picture of the reader, of the access panel, of the camera, or like even give a grade of like, hey, we only do their fire system. What would you grade their access, their video, and like you pick? It's like A B C D. And it's like you can put some notes in there, something something interactive to just collect information. Because if I can bring the information back here, then I can use AI to outbound those customers on, hey, we just did a technician visit. We heard that your video system may be a little old. We'd love to when you're ready, we'd love to propose a new video system. If you had that automated, you don't need your text to sell. You need them to be undercover agents, you know, getting you the information about those systems.
Yeah, I do think that that's where we'll go in terms of like our next attempt to try to improve on this. Um and I I do think in those questions, like free response is a bad idea. I think you need to have like standardized responses to measure against and collect data on to try to you know build yeah, essentially outbound off of, like you said. Um and you know, it it may be that like one uh additional photo on a different scope is the default, but maybe there's um you know some sort of incentives when they start to expand beyond that, right? Or kind of overachievers, so to speak.
I like gamifying it. Like what if you made it like, hey, you get point for every uh the one of these things, and the most person with the top three points at the end gets paid.
Yes, that comes back to average age of technician, I think, in terms of how effective that specific mechanism is. But yeah, um cool.
Well, um, I think those are the three, right? Like we covered those three. I think that those are the big three that if you can solve for, this is kind of how we think about using AI in those applications. And by the way, um, if you have other ideas and you're doing stuff, please leave us a comment and tell us what you're doing because we love like we just got we got back from a trade show recently and we were meeting with folks all across the industry from vendors to integrators, hearing what they're using AI. And it's and it's super fun to hear like what people are doing and other apps they're exploring with. So we like to hear about that stuff. So uh please leave us a comment or shoot us an email because we'd love to hear about it.
One final note I was a little self-conscious when I spelt slippage earlier, and I was like, oh yeah. Did I did I misspell that? But no, I didn't because I got that dog in me. I was I was accurate, and so I in fact am the spelling be key of this podcast.
Okay, King. All right, there's enough of that. Let's go. Um, if you like this content and if if you like Steven, feel free to subscribe. Otherwise, I hope you have a great week.
