Episode Transcript
[00:00:00] Speaker A: Welcome to the Relay, the legal show for personal injury law firm owners, presented by lexameca, the number one attorney referral network. I'm your host, Gabriel Stertz. Joining me today is Chris Keller, managing partner, CEO, founder of Keller Swan Injury Attorneys, and one of the most innovative lawyers I know in the personal injury space. Chris is the powerhouse firm in West Palm beach, trying over 150 juror trials. And now he's taking on one of the biggest challenges in legal tech, building conversational AI for his law firm. As you've heard me Talk about on LinkedIn and on this podcast, AI is changing the way that law is practiced every day. I met Chris a couple weeks ago and started talking about his firm, and he's actually building AI into his law firm, conversational voice intake agents, and so wanted to have him on the show and dive in. This is a super interesting Chris. I've said I'll say it to anyone who listen. I think this 2025, two of the big trends are agentic AI and AI voice. And when you said those, you pulled the string in my back. So let's talk about it. I'm super excited to hear what you're working on.
[00:01:02] Speaker B: Yeah, Gabriel, thanks for having me. I really appreciate it. When I started a law firm, I never really thought about technology as being. Being a tech company in some ways. You know, I never thought that we would really have technology in our practice the way we have. But over the last couple years, with the introduction of AI, just kind of seeing how it could really help, I realized that, you know, you know, lawyers not going to be replaced by AI, but law firms will be replaced by other law firms who are using AI. And so we really decided that we need to be on the forefront. We wanted to be innovators in the space. At the very least, we want to be early adopters. Right? We never wanted to be kind of late to the game. And so that led to really jumping all in with AI and how we could use AI in our practice.
[00:01:43] Speaker A: So what's so crazy to me is I was in a Mastermind in February of last year of 2020 for. And everyone in the room was saying, what is AI? Help me understand what this even means. Fast forward. We're almost exactly a year later. Everyone's using AI. I mean, half of all plaintiffs attorneys have adopted AI already at this point, when they're asking, what the hell? This is a year ago. You're building it for your firm. So, I mean, look, I know you run a CRM internally. You're doing A lot of you're doing all the best practices in terms of technology. They've been around for a couple of years. What are you building in AI Voice right now? And why do you think that now is the time? Because I think like even six months ago, AI Voice really wasn't there. By the end of this year, I think it's going to be way there. So like, where had. When did you get to a point where you're like, now I can start building for AI Voice. And when do you think you're going to launch this thing?
[00:02:34] Speaker B: Well, you know, our firm really grew a lot over the last couple of years. And we went from signing, you know, 50, 60 cases a month, the last month we signed over 200 cases. And you know, so our call volume grew exponentially too. So I think like a lot of law firms, firms out there, we've always struggled with intake, getting it right, finding the right workforce. And we've had us based employees full time and directors of intake, intake attorneys, we've offshored it had virtual assistants. We've had, you know, we, we've tried everything, right? You know, just like most law firms and we were still missing phone calls. I think a couple weeks ago we missed like 17 phone calls. I'm like, how is this possible? We have, you know, eight dedicated VAs in there. We have intake supervisor, we have an intake attorney. We have, you know, fail safes, we have, you know, some other AI know, background stuff. We have, you know, after hours call service that also acts as kind of backup. All these different, you know, mechanisms in place to make sure that we never miss a phone call. And it just as a law firm owner, missing phone calls is, is, you know, especially an intake, you don't want to miss any phone calls. Your firm, but especially an intake. We're paying to acquire cases, we're paying to advertise, we're paying money to brand, we're paying money for, you know, other services in order to try to capture clients. And so every missed call is missed revenue. I mean, it's a potential, it's amazing. So every time we start looking at that way, so, well, that was a missed call. And then you look at your average fee and say, well, my average fee is, you know, 16,000 a case. You know, it's like, well, they just went $16,000 and there went $32,000 and so forth. So, you know, we started saying we got, there's got to be a better option here, right? How's this working? And we're constantly hiring new people and training new people and putting more people on the phones, and it's still not fixing the problem. Like, what's going on?
[00:04:12] Speaker A: And is that because. Is that just too many rings? And so it's a. It's a drop call or is it a totally missed call? It's like, hey, this person rang three times and they hung up. What? You said 17. What are those? 17 missed calls?
[00:04:22] Speaker B: Yeah, some of it was that. Some of it is, you know, we don't answer quick enough and they're hanging up and moving on. Some of it is sometimes can be our connectivity. We were having kind of an issue with, you know, using bas, so sometimes you're having transfer issues between just the kind of phone services. Right. Trying to make sure it's. It's done all kinds of, you know, Internet issues and that kind of thing, connectivity issues, and then also just, you know, people on the phone, you know, they're busy, so they're on phones. So we recently switched over to a Zoom contact center, which is absolutely spectacular. It helps us, you know, kind of see who's available. We get to see how long their phone calls are. There's a lot of AI in Zoom, which is really cool. So that's. That's been crazy for us. But the next step and the next implementation, which we're, I think, you know, a couple weeks away from rolling out, it's conversational AI. So we met, you know, a developer who had kind of developed the project, but wanted to work with the personal injury firms and said, you know, how about we work with you guys? We know you, we're local, you're local. We can kind of come in and out of your office and have a really, you know, good partnership. You know, they saw all the things that we are doing in our firm, how our firm has really grown the call volume, obviously, was a big deal. They have to. In order to train the AI, in order to train conversation. Oh, yeah. They need, you know, a big segment size of phone calls. So we provide a lot of that. And then we've used outside sources to train. And so we have call scripts and we've done, you know, a lot of great things in intake. We've spent a lot of time, energy, effort, money, and insight to get it right. And so we were able to kind of start training the AI and now. And now we've, you know, have it where it'll pick up after hours for us. And it truly is conversational. So someone can call and.
And, you know, say, hey, I was in an accident, and say, oh, I'm sorry to hear that. Can you tell me when the accident happened? And then they say when what? Can you tell me where? I mean, it's amazing. And the cool thing is, is it escalates so that if it case that someone wants to speak to actual person, they can just ask for an attorney and escalate it. If they say some keywords that we put into it, it will automatically transfer them to attorney. So one cool thing is like warm transfers, live transfers, right then to the attorney. So, you know, are you.
[00:06:23] Speaker A: You running it in the wild yet? Or is it totally. Is it sandbox like, or. Or is it taking calls from people that are calling in?
[00:06:29] Speaker B: Yes, but we have it sandboxed, so we have it only certain phone numbers, certain locations, so we can test it out. We don't have it for our. Our main number where the majority of the calls are coming right now because we, again, we want to make sure it's learning. We want to make sure it's getting better before we kind of roll it out fully. So right now it's just kind of in smaller markets where we're not spending a ton of money. We don't have a. We don't have a ton of call volume, make sure that it's going to work.
[00:06:51] Speaker A: And what kind of volume have you run through it so far? Dozens. Hundreds of calls?
[00:06:55] Speaker B: At least. Dozens. Yeah.
[00:06:57] Speaker A: Okay. Okay. And are you from just a pure compliance perspective, like, what do you view as best practice? Are you. When it. When it picks up the phone is saying, hey, this is an AI agent, or is it just answering the phone? Is there a level of disclosure there? How do you approach that?
[00:07:11] Speaker B: So we. We do have it saying that they're an AI agent, so they know that they're talking AI. You know, it's one of the things we want to be transparent, number one. Number two, we didn't want to lose people because they think, oh, this is just a, you know, just a recording. I mean, It's. It's like 99% you. You have a hard time telling it that it's not a person. But still, it's just better to like, hey, I'm an assistant. I'm here to get some basic information. I can live transfer you. You know, it kind of gives them a little bit of a script, and so we'll see how it goes. The cool thing is we can test it. We can see what's working or what's not working. If people are being turned off and hanging up or if people are going through it. We've used AI in the past with call services. It was terrible. It was kind of like calling Amex and it was clearly AI and we did lose calls. This is truly conversational. And so far we've seen it, seen pretty, pretty great results.
[00:07:55] Speaker A: And then I'm going to ask the question that any law you're not familiar with AI is going to ask, really the million dollar question, which how do you keep it from going off the rails and giving legal advice or saying, hey, your case is worth a billion dollars, or trying to do whatever it is that AI in the news about doing, have you found that you can put sufficient guardrails on, on this conversational bot?
[00:08:14] Speaker B: Yeah. So again, we spent a lot of time with the training, with the developer, putting in essentially all the parameters for what it's, what it can say, what it can't say, and when it's asked certain things or said certain things, what it's supposed to do. So there's a lot of work in it, and that's why it's hard to do, because if it was easy for everybody to do, they'd all be doing it. But we spent a lot of time, energy and effort to make sure that we get that right, to make sure it's not giving legal advice, to make sure it's not doing anything we don't want it to do. So we trained like empathy and empathy and all those kind of things to get those kind of words. We've made it, you know, very warm instead of very, you know, more professional. It is conversational.
[00:08:50] Speaker A: And I will say that for voice AI, one of the nice things about it is that it's harder to prompt, inject or to do squirrely things with it, because if you have direct access to chat with an agent, it's going to be easier to do workarounds versus voice. It's, it's going to transcribe that into written word. And so it's hard to put in weird characters or to format things in a certain way. So I actually think voice AI, the bigger risk initially is from the experience perspective, turning people off because they're dealing with an AI agent less from an initial security perspective. Again, not to downplay it, but I do think it's easier to put the guardrails on voice AI than it is with like a chatbot on your website, because you can inject anything into that and there's still a lot of potential issues with that. So that's, that's fantastic. What another thing that I see is like a real critical piece of voice AI is going to be when they're able to actually sign clients up. Because closing on one call, super, super important for MVA firms like yours. So right now you're having it escalate, I imagine you're not having it send out retainer links or trying to sign anyone up, right?
[00:09:51] Speaker B: That will be next. We're close to that. But right now it's just getting information in live training and then live transferring to an agent so that we get it signed up. We do have our phone calls answered by a lot of people 24 7. But at least if the AI picks up first and then, you know, goes through the script, then it can go ahead and transfer to someone who's ready and able to sign up. So it's given us a little bit more leeway in our, in our intake, give a little bit more time. But most of importantly it just makes sure we don't miss phone calls. It's answered within 15 seconds. It's, you know, it's. If we don't answer it, if our people are not answering the first ring, it's, it's always picking up. So we're not missing phone calls, which is, which is huge.
[00:10:25] Speaker A: And does. Do you know what the drop rate is like? It's picking up the phone. That's great. Are you having people that are hanging up on it or is it pretty good at getting someone in the door and then to a live person?
[00:10:35] Speaker B: We're still getting data on it. I mean, so far so good. We, I don't think we've missed any on the AI side. We've definitely missed it on the live side, but we haven't missed any on the other side yet. So. And so far the response has been good. And we're testing it in areas outside of South Florida. South Florida is our base and our home base. So we're kind of testing outside of areas and we're kind of concerned about that because sometimes in the location that we've been testing it, they want, they want to hear a voice and like a real person. So the fact that it's working there and it's working so good there makes me feel really good.
[00:11:02] Speaker A: I mean the thing that I think is, is like the near term future where AI starts to beat out your kind of, your, your other types of, of call centers is the fact that once you decide to roll us out, like you're a regional firm, you're going to get calls from all over the country. You get an area code, I'm in Arkansas 501. You can match the accent to the area code, if it's somewhere in Arizona, if it's New Mexico, like you can take a regional accent and apply that. Based on where the phone calls coming in, I think you're actually going to see performance gains by the end of this year that are better than a human picking up on the first call. People want to hear themselves when they pick the phone up. And so I think there's really interesting opportunities to enhance the client experience with this. And so the fact that you're rolling it out now is awesome. Like you're already getting the data, you're already stress testing it in real life. And the amazing thing, because it's software, it's just going to improve from where it is today. You're going to get those gains. I'll be really curious to see how long it is before you start saying this thing is good enough, that it's going to pick up some of the calls just because it's that performant and then start handing it off to humans to do the signup or if it's a catastrophic and you really want that, the human touch. But I'm really curious to know, like, it doesn't sound like it's particularly far away from the benchmark of having an overflow service or VA service picking up your overflow calls.
[00:12:22] Speaker B: It really is. And the cool thing is too is it speaks 150 languages. And so one cool thing was, you know, we had some people in our firm who were, you know, they were a little hesitant about this, like, I don't know. And you know, they heard it in English and we have some, a lot of Spanish speakers in our office. We're here in South Florida. And so when the, when the phone call switched to Spanish, all the Spanish people were like, oh my gosh, like, that was, you know, incredible. And now all of a sudden that hesitation that they had with it kind of went away because it was speaking, you know, Spanish to them. And they were like, the Spanish was better than the English. And they were like, what are we doing sign up now? So it was truly, you know, that was kind of remarkable. But the fact they can speak 150.
[00:12:56] Speaker A: Languages, I mean, come on, that's crazy. And from a cost perspective, where is it right now in relation to a human? Is it at parity? Is it below cost for hiring someone on a per, on a per minute basis or per hour basis?
[00:13:09] Speaker B: Incredibly below.
And that's what's amazing too. I mean, we, so we, we went the virtual assistant, the offshore route, which is obviously saved Us a lot of money in that sense. So with labor cost and so that was good, but this is even cheaper.
[00:13:21] Speaker A: That's incredible. That's incredible.
[00:13:23] Speaker B: Think about and realize too, which you know, I didn't, I didn't always kind of think about when I first went into business. I didn't understand and appreciate. But you know, when you talk about these call centers, these call centers typically have a high turnover. You know, they're, they're usually lower paying jobs that for people, whether you buy your higher, you us based people, they're going to be lower, lower salary jobs. And so you tend to get a lot of turnover there. People are like, I don't want to be on the phone all day, you know. And we did a lot in house to you know, do like personality type tests to find out are these people like really people who want to be on the phone? Because you have to like me on the phone if you're call service. And so we even did all that kind of stuff to try to make sure that we had the right people. But still you'd have turnover and then you have to go and find someone new and then you have to train them. And so what the best thing about the AI is is that you know, you train at one time, it's a one, it's a one time cost. I train it one time, teach it one time and that's it. And I can, it's infinite. Right. Says, you know, my call volume goes up. Now I have to go hire more people or I'm signing up more cases. I got to go hire more people. Now I got to go recruit, train new people over and over. The best thing about it is I never have to worry about that. It's never a problem.
[00:14:25] Speaker A: Yeah. And the one assertion that I will disagree with you on from this entire episode, because this is, I mean this is the direction that you're going. It's going to replace people. AI is going to replace every human being from the moment someone submits a lead form or calls in or needs an outbound call all the way through early case management. And it's just going to get even further away. 24 months it's going to be attorneys and some very skilled people managing AI that's running everything from intake through pre litigation. I think a lot of people are afraid to say that there's only so many places that you can reassign people in your law firm. And when you have technologies like what you're talking about, plus the co pilots that are running on top of your databases and writing your documents for you and doing client communication. I think that the headcount reductions are going to come to law firms very quickly. And the brutal reality of that is if you don't make that change now, which you're doing, Chris, and the people that I talk to, a lot of them are doing that. But it's either be eaten. Chris, once you have the ability to have 20% of the headcount you did before, what are you going to do with that money? You're going to reinvest that in marketing and case acquisition. You're going to reinvest that in the best trial talent that you can possibly find. And then you're going to be able to outcompete everyone else who hasn't adopted that technology. I think the kind of, the kind of happy go lucky like we can have our cake and eat it too. Of oh, we can have AI and we're not going to fire anyone. I don't believe it. I'm, I don't. That is not where we're going. This is too good and it's going to just get better and better. You said, I mean the only barrier here at this point is having AI be able to sign clients up. Like you said, you're working on that already. That's it. Once an AI can sign a client up and if it's as good as someone else, I just don't think that that's going to get any better. It's it really, then it's not going to be worse than someone in a call center and it's only going to get better. I really don't see an argument where we actually have human call centers in 12 months.
[00:16:12] Speaker B: I don't disagree. I mean we, we've already cut back some in intake just because this is rolling out and we, we can see the, the real benefit. So we're, we're, we're going to make some cuts. It's going to happen. And I'm not scared to say that. I mean it's true, but it's also, you're still going to have to have people that take those hiring calls and deal with some of the escalations. And so we' people in intake, probably like you said, lesser of a head count. The cool thing is the that that's come next for us, we don't have it quite doing it yet, but that should be within the next month is the outbound phone calls. I mean we're going to be able to actually have AI call our clients and ask our clients how are they doing? How do they feel? Did they go to their doctor, you know, to give us updates about their medical care and treatment, to make sure they went to the doctor and then to tell us what the doctor said. It's going to, you know, it's going to give us all this information, put it right in our case management software and that's going to just, that's going to change the ball game. I mean, because it allows us to better medically manage cases. It allows us, you know, we've done so much already internally with AI, but for it to be able to actually make a phone call and find up is going to be, you know, incredible. That's the thing I'm most excited about. I mean, obviously the intake part of it is really exciting, but for me, the client service aspect, once they are a client and, and we pride ourselves with communication to be able to communicate in all the ways that we already have. And then to add that as man, it's, it's phenomenal. I mean, so one thing our firm did was we actually hired an in house data manager who he loves, does a lot of data for. He builds out power, bi dashboards and all that. But he loves this stuff, he loves AI, he loves technology. And I knew that if we were going to make that step, make that leap and be that firm, we had to have someone in house who truly understood technology, who, who loved it. And it's amazing. My IT outsourced company called me the other day and they said, hey man, you know your data engineer is doing some, doing some weird things. I said yeah. And he said, yeah. He's like, you know, I have a lot of law firms that I, that I work for and none of them have anybody that's doing the stuff that your guys do not go. That's a good thing. So that's a good thing. I'm good. Whatever he wants to do, just let him do it. You know, we're good.
[00:18:09] Speaker A: Absolutely. No, look, I think you're in the right direction. Anyone listening to this, you gotta, if you're not doing what Chris is doing, then you're gonna get outcompeted by people like Chris. You're willing to make those investments because it's not a question of what you want to do, it's just a question about what is the market going to bear. I'm looking at my numbers right here from a recent podcast. Non attorney comp is generally around 20% of your law firm's expenses. That number is going to get cut way, way down. We've already seen it trending down, it's going to go down further. And that's money that's going to get reinvested into into litigation and case acquisition. And branded law firms that understand how to use technology are absolutely going to be the future of what's going on. So, Chris, really appreciate you coming on today and sharing what you're working on. Super exciting stuff and can't wait to catch up with you in a couple of months when you've seeing a few more iterations on this and how it's going.
[00:18:58] Speaker B: We'll love to share. We're excited. We're hoping to roll it out after our firm really develops and has a good place. We're hoping to roll it out and have it as a mass market.
[00:19:09] Speaker A: Fantastic. Looking forward to it. Thanks so much. All right.
[00:19:11] Speaker B: Thank you.
