The Future of AI Phone Receptionists | Eric Hoffman

In this episode, the host talks with Eric, a lawyer who started a legal tech business called AI Esquire. They talk about what it is really like to build and run AI tools for law firms. Eric explains how his company helps law offices by using AI phone callers to answer calls after regular business hours, so the firms do not miss out on potential new clients. He shares the real difficulties of growing a small business, like finding the right people to hire, dealing with the rising prices of AI technology, and competing with cheaper alternatives from other countries.

They also discuss whether law firms might start running their own AI software on local office computers to save money and protect client information. Eric addresses common worries that lawyers have about using AI, showing how phone systems can be carefully set up to sound caring and polite to callers in stressful situations. Lastly, he explains why taking the time to visit law offices in person to set up the software builds far more trust than just sending an email or doing a online video call.

Timestamps

• •00:00:01 - Introduction
• •00:01:15 - Startup Growth
• •00:05:27 - API Costs
• •00:08:48 - AI Competition
• •00:14:00 - Government Regulation
• •00:20:02 - In-House AI
• •00:24:31 - Tech Adoption
• •00:28:29 - AI Receptionists
• •00:32:07 - Voice Guardrails
• •00:36:38 - Customer Support
• •00:44:46 - Closing Remarks

Awais: All these companies are trying to subsidize this because the compute cost and the energy cost is too high for them and obviously it's investor money so they keep on putting it. It will go up eventually. They can't you know just keep it really low. But there's a political angle because in China all these models are three or five times cheaper because their government funds this, right? But in the US it's a complete different story. The government isn't funding it. They are either looking for the public money or the investor money. So do you think that in the long term the government has to think about it in a way if a lot of people you know sign up for this because the whole legal tech market which is based on AI is in danger if you look at the future?

So today on my show again I have invited a very special guest who is basically a lawyer and now he's running a legal tech company focused on AI for after hours calls and he makes sure that the firms do not lose those valuable leads which cannot be answered by a human. So, thank you again for joining my podcast, Eric.

Eric: It's my pleasure. It's great to be back and always nice to be invited back somewhere you've been. So, maybe I did something right.

Awais: No, you were wonderful in the last episode. A lot of people actually told me that they want you back, specifically when it comes to how things have changed and progress since we last met. So right Eric I want to start with the current situation of your legal tech startup. Tell me how it's going and what challenges are you facing right now and what are you planning in future? What features are you introducing now?

Eric: Oh so much there. Maybe we're going to break that. Let me break that down the best I can. Okay. So, let's state the status of AISquire. We're doing we're very very overwhelmed with the interest we've gotten. We, you know, we've have a lot of feelers out on LinkedIn. We've got all the social media going and then just phone calls, emails. We've gotten so much great feedback. I think we've gotten a lot of people that mentioned your show as well. I think that was a popular show. But anyway we're in a position now of growth and expansion. Sometimes expansions and pivots are indistinguishable from each other, but maybe they're one and the same. I don't know. But we are pivoting slash expanding more to lean on some of the software that we've designed. We've designed quite a bit of software, but for many reasons mostly business related, we decided to not really focus too much on the software because we were building our presence in the AI agent and AI intake. Our intake was really the thing that was getting us in the door. So it was very difficult for us to then say oh well we've also got this. It was kind of muddying the conversation. So, we just stayed focused on the agent and our intake system, and that did very well for us. So, we're happy. We're super busy. Our biggest problem now is honestly, I'd say related to that. We need to hire more, I think, and we need to grow. And that is something I guess when you're a founder you learn as you go because hiring I've done it before at law firms you know parallegals and legal assistants, but when it's your company it is a different feel to the hiring process. So, I've been a little bit of a what do they call that? Paralysis by analysis, I guess. You know, been overanalyzing. Well, where do I get the candidates from? Do I go to LinkedIn? Do I go to Upwork? There's so many places, so many ways you can do it. So that's kind of a challenge is we're a little overwhelmed and we are small, very small, and we need to be bigger. So we don't know how to take that next step to grow at this point.

Awais: Well that's interesting. That's interesting because a lot of founders feel like sales is their problem, but it's really good to know that you are into a new problem, but obviously you are going to overcome that challenge as well and it's very important that you've said this because I do find this extremely difficult to hire people and because it requires so much effort in the start and then if the person is not the right fit for the company it's also waste resources and time. That's a lot for you to take on.

Eric: Yeah, absolutely. I agree.

Awais: So Eric, my question is that since every day new models are coming up, obviously you are using an LLM behind which might be from a particular company, but the cost that we've seen are rising specifically when it comes to the API cost. How are you keeping up with that? Are you raising your prices as well or are you keeping it at a certain level where it stays affordable, and if it keeps on rising like this, do you think that it's going to be a challenge for you in the future?

Eric: Yeah, that's a great question, and yeah, it's gotten very expensive very quickly and especially if you want to chase the newest model, whether you're going with Claude or GPT, like especially with Claude, there's been a lot of pushback as you probably know from all the Fable and the Mythos stuff. But anyway, what we do candidly is we did have to raise prices actually. It didn't impact us. Our clients seemed honestly I was very nervous to do it, but our clients it didn't really affect anything. They kind of our clients almost agreed were so agreeable to it. I got a little pushback from some people I use as advisors, other attorneys, and they said you were too cheap anyway. Like you really underpriced yourself. So I guess maybe where we should be now, to be again very candid with you, I feel maybe we were a little underpriced, but now with the price increases coming through the LLMs, we really don't have a choice. I mean, we had to preserve. We don't charge per at this point. We're not charging per call. So we have as you can imagine a very challenging dynamic to remain. Our margins are very tricky because we really analyzed all the we have a whole list of like every call anonymized, you know, we have like the length of the call, how many calls, and we really tweak our pricing to allow us to make some money, keep the lights on, but also to keep it fair, but also that could deviate. Each client could we could have an outlier, an edge case or something where there's a very busy firm just getting calls all the time and then we could end up in the red for that, but we're open to that. So anyway, again, our pricing did go up, but we had to for the reasons you mentioned, and we don't anticipate that happening anytime soon. But I think clients, people interested in buying should buy sooner rather than later if you're interested in this. I'm not it's not a sales pitch, but I would definitely look into whether it's AISquire or it's another company. It's probably better sooner rather than later cuz prices I think are on an upswing pretty fast.

Awais: Yeah, exactly, and this is the reason why it has to be with the market shift and the competition from the Chinese AI models a lot because although we've basically seen that whenever the competition increases the price goes down, right? But in this case it's very strange that the companies need to increase the price even though the competitors are three or five times cheaper now. So, don't you think that there is a big challenge coming up and legal tech industry overall might shift towards Chinese models?

Eric: That's a great point. I actually don't know if I'd considered that with the increase in competition, you think there'd be a race to the bottom. So that's some of the pushback by the way that you know like people you read on Reddit or you go out there and you see some of the posts in different chat rooms and so on. A lot of people are saying like listen we could go somewhere else, to Anthropic mainly. I think Anthropic's been on the hot seat quite a bit for pricing. You know, I mean honestly though there's really nothing worse, I'm just going to again be candid with you, when you're programming and you're using Claude, which I do use along with some others, that to be in the middle of something, you know, and you're in a workflow, you're really focused and you're getting things done and you're in code and you're producing, you're making things and you're being productive and then all of a sudden you get hit with the five-hour that dreaded five-hour limit which drives me crazy. But you know you either have to empty your bank account to increase that for the moment or take a break and wait for the time to reset. So it's frustrating and people are starting to say well we have other options. So maybe we should look at DeepSeek or some of the other options that are coming from either China or even just ChatGPT if you're mad at Claude or vice versa. There's a lot of pushback coming from the pricing, but I'm trying to think of your point. That's a great point about it should be less expensive based on what I just said. It should also be less expensive because there's the threat of competition. I don't know, do you have any thoughts on I have no idea. I can't think of why that's I'm being honest with you. If you're asking me to opine on that, I think I would say the clear reason why, and I'm going to go with the obvious reason why. The obvious reason why is that they're providing better. I think it's strictly economics and it's strictly market stuff, market theory where they're just saying we're offering a better product that costs us more to create or it's a faster, better, you know, how they obviously all the specs are out, Mythos and all the legend behind that already. But you know they release Fable and Fable comes out and you're out of tokens, you're out of time in five minutes. But when you're using Fable, you're like, well, I see why it's kind of there are some things it does better. So I guess that's the thing is they're saying that we're providing a better model, so we're going to price it higher. So without regard to the fact that there's tons of competition, I think they're just ignoring that. That's my best answer on that.

Awais: Yeah, because you see in my own firm because we are into data we have to analyze a lot of data that law firms have. We use AI a lot and recently I'd be honest with you, we were using Claude and the prices were shooting really high and honestly it was not profitable anymore. So one of my programmers said that there's a new model Kim launched in China. It's almost at par with Mythos and Fable and you could just give it a shot. And I thought, okay, let's I'm not really a big fan of Chinese products in general, but I did give it a shot and to be very honest, it was very very good. But the thing is that we can shift now, but there are also political scenarios going on, right? The only reason that I see is already all these companies are trying to subsidize this because the compute cost and the energy cost is too high for them and obviously it's investor money so they keep on putting it. It will go up eventually. They can't just keep it really low. But there's a political angle because in China all these models are three or five times cheaper because their government funds this, right? But in the US it's a complete different story. The government isn't funding it. They are either looking for the public money or the investor money. So do you think that in the long term the government has to think about it in a way if a lot of people sign up for this because the whole legal tech market which is based on AI is in danger if you look at the future?

Eric: That's interesting. My thought on that is I feel like the second I almost got nervous you mentioned government and AI in the same sentence. I think people shiver and get nauseous because first of all they think of regulation obviously. People don't want the government involved. I think that's going to be the first problem with government. I think government got involved already with Mythos. I've heard that they've been slowing that down for security reasons. So they're involved I guess in the regulatory, somewhat ad hoc regulatory process. But I think that if you're asking about getting them involved with funding AI or subsidizing somehow AI, subsidizing funding because yeah I think it's possible. I mean they could probably subsidize cuz their energy use is astronomical obviously. So figuring they could do it that way, do tax abatements or certain tax incentives. I mean they could theoretically do that. The scale of these companies, I mean from everything I read, there's going to be these internet power centers on every corner, so servers and the power generation that they're going to need so much power to run these things. So the infrastructure is going to grow and grow. They're going to be building. So maybe tax breaks would be useful and attractive to them. So I guess the options there, I feel like that's a realistic possibility. I think that's a great idea and I think that you might start to see government and AI getting closer and I hope it's in a good way. It's going to be one of two things: either government gets involved like you said to help it and maybe subsidize it or they get involved to dig their teeth in and kind of regulate the heck out of it, try to slow it down, which would be exactly the worst thing because of what you're saying is we've got the Chinese competition and Chinese would love to see that, to see regulation slow America down a little bit and then we've got our own government slowing us down. So that would be easy.

Awais: Yeah, that's an interesting point there because just not about the government because obviously there is a political tension between the two countries, but obviously it also might not happen because of the competition rising from China and Trump being very aggressive sometimes he might just outright ban Chinese models within the US. So what will happen to the overall economy in that sense? Because if the businesses who are running right now in the US, they probably want to get more efficiency at a cheaper cost, right? But if that happens, they ban Chinese models and the competition is now just divided between three or four companies, they could keep on increasing the cost without any the free market thing. So that could also hamper the legal tech industry overall.

Eric: Right. Obviously in another way. I wonder which one of those I feel like one of those scenarios will play out. I think I don't see the government do you really see the government maybe outlawing Chinese AI? I mean, I feel like there's been so much tension, like you said, aggression and tension lately with especially internationally. I feel like that's not I feel initially that's crazy. That can't happen. But then another part of me says, well, you know, the escalation to like closing the Strait of Hormuz was something that I feel would have been unthinkable to happen a little while ago. So now I feel like, hey, why not? That could happen, I guess. But I hope it doesn't and I'm a very free market believer. I don't think the market should determine for itself what stays and goes. And if that were true, if that happens, then we have our original argument which is well then why aren't prices going down? Because our original point was or your original point was, you know, it was an interesting point. The Chinese are coming, it should be bringing prices down and that will become an issue if Chinese keep coming in with good models. I just don't think people trust them yet. Your point also which said I don't like Chinese or you were a little skeptical let's say about that country's products. Well, I think that's what's running the market right now. I think people are staying away from it for you know they know the government's involved and spying espionage there. So yeah, I think for now people are afraid, but if that sentiment changes, I think you're going to see American or the prices come down, but I think right now the American companies are safe.

Awais: Yeah, that's true for now because what my whole idea is that we need to figure out that if any or other situation happens, what is the impact on the AI legal tech industry? Now you see there is one firm that I'm working with right now. It's a California-based family law firm, mid-size firm. So I saw another competition and that was they actually hired a full-on team and built their own in-house proprietary AI. Now it's basically running on a very good machine. It's a Mac Studio and their AI agent is taking calls. It's also taking the important fields putting it in the CRM. They use MyCase by the way. So I was actually thinking while I was talking to them and I basically asked them that why did you get the whole system built out for a very big price rather than just go on something else? I could recommend three or four products to you and they were like no what we want is we want to keep the cost control because the model they put in is basically a Chinese open source model so they have just downloaded it and put it in the PC configured it and that works really well. They don't have to share the data and they plan to increase it to other functions of the law firm as well. Now think about it in this particular area let's say nothing goes wrong on the government side nothing goes wrong on the cost side, but if the firms decide that it's so easy to just download the whole model, put it there and run it for free or at least just pay the bills that it will generate for a computer running, do you think that then all the intake agents, all the agents which fill case details, documentation agents, how worthy will they remain in the long term?

Eric: Well, that's going to be as commoditization kind of increases with that market. I think that's happening now. And I think the commodity putting these AI agents, I'm telling you, I swear when I had the idea for my company a year, year and a half ago, it was a great idea. You know, it seems like everyone had that great idea. So back to one of the problems or you mentioned your question. I think your first question to me, maybe I dodged that question, but I said hiring. Well, another problem is to be very candid with you is the crowding of the market. It seems like if you go on Google and search any name, any name, you know, you'll find like Lisa AI type, you know, K AI. There's all these names coming out with AIs every day. And I think the crowding of the market is something that's interesting. But the commoditization will happen and which is happening and then I think after that then it's going to be like what you just discussed I think.

Awais: And probably I think a lot of SaaS companies what they how they survive actually is they roll out an enterprise pricing which is not disclosed and what they do is they basically because they're ahead with the whole product they can just put it there and in a shortest possible time they can keep make it run. So do you think that in the future with AISquire you would be offering something to clients who can pay you well to establish AI within their offices where they can use it and probably give you maintenance or something with updates rather than selling them monthly or yearly basis as SaaS model?

Eric: Like what again like in terms of the agent itself or the AI?

Awais: Yeah, the whole idea is that say I have a firm, I feel like that I should not be paying say $10,000 a year to somebody and use their product and my system and my documents, my information lives within their servers. Why can't I ask you who has already built this to come to my office establish that and I pay you a couple of thousands for just maintaining that or updating and upgrading the system?

Eric: That's an idea I've kicked around maybe once or twice. It's a great idea. I think I just don't think there's so much skepticism with technology. You have to know this working with law firms. I think initially there are some that are like yeah, but most are like okay fine give me a little bit. I want like the low just to put my toe in the water a little bit. I don't want to I'm not ready. Most attorneys are like that. It's just the way we're trained. And I think to then say to offer like when I offer them just the agent a lot of times like whoa wait it's going to be answering my phones after hours. I don't know voicemail I'm a little more comfortable with and I actually have to argue in favor of you know AI receptionist versus voicemail, which to me is an argument that wins itself, but not to attorneys. So when you're saying the prospect of giving them you building a full agent like on site, if they wanted that, if that was a demand, I would be happy to do I would love to do that. I just feel like that ask is a little too much for what the market could bear at this time in terms of the curve of adoption here because I think the adoption is maybe tolerant of minimal AI use. I think with documents it's been adopted a little in a larger scale like I think the processes in a law firm I think attorneys see very well how contract review and any document rudimentary kind of repetitive process could be helped by AI, but when you get to the phone they some of them feel like it's good enough the way it is, you know nothing's really broken so I don't know if they're going to be willing to take that leap, although I think that's probably where it's going to go ultimately once the adoption moves into a more comfortable level where more people are using it then some are going to say well why they'll ask the question you just asked well why am I paying AISquire or some other company to do this when I could maybe have them or someone else come out and build me one for my own? I think that's the sequence. It'll first adoption has to increase and the comfort level and then we'll see that.

Awais: And the reason I think the adoption here is not really well documented as well because number one we don't have any specific scale to measure the productivity versus the cost it gives us number one and every the market is so volatile right now that every day if a new model comes in it's very hard for people to justify if they want to pay more or less, number one. Plus there's one more issue that I want to basically discuss here which is a lot of firms now have started comparing when the costs are rising that is it even better to use an AI or just hire an offshore person who can stay on the phone maybe a receptionist which might cost me like a couple of hundred bucks every month?

Eric: Yeah. So if you can find someone overseas that you can trust with intake and I mean you got to make sure you're comparing apples to apples because if you just want someone to answer the phone take a message hang up basically a voicemail but a person doing it that yes and I'm not this is not a sales this is the factual breakdown of what you said. If you want that, yes, then it's a commodity. Find the cheapest price, the lowest price for what you need. Someone will answer the phone wherever it is, overseas, here. But the AISquire, I mean, at least what I know that you do something similar with what I do. I could speak from my experience. Ours does the full intake. You know, ours will actually qualify the lead. It'll actually determine a value of the lead. It'll actually, with the analytics we have, we could analyze everything down to the demeanor of the caller and everything's recorded and transcribed and you get the notification stuff too. You could get texts, you get emailed, you get a phone call when you get certain calls, you get escalation as well. So if the AI gets a call, it's an emergency and it's on a weekend, we could build rules where it'll escalate to the attorney if it's a high value case or if it's a trucking accident, or anyway I could go on and on about those things, but there's a lot that the AI does that I don't think the call center does. But if they do, if you find a call center that does all that will qualify the lead with proper guardrails won't say the wrong thing, I mean that's one thing people are afraid of the AI saying the wrong thing and I'm like, I think you should be more afraid of a human saying the wrong thing because an AI is literally I can show you the wording. It's like really tough programming. It's going to be difficult for it to just launch out on its own. I know about hallucinations and everything, but it's just not the same thing with voice agents. They don't just start like babbling about some other clients like private medical. It's just too hypothetical. That doesn't happen. So anyway, back to your question, I think yes, if you just want someone to answer the phones, go for it. But if you want something complete and you want that quality, like the leads you come out are qualified, they're done, they're signed up, they're ready to go, I think you pay a little more for. I wouldn't take that as a commodity yet. I don't think that's something you bargain shop yet. I don't think we're there yet.

Awais: Yeah, I completely agree with you because obviously one thing that for a high velocity firm it does make sense that somebody and obviously you can't keep a lot of people human actually after hours because in a high velocity firm you might not want phone lines to be busy. But the thing is that we're not just dealing with let's say personal injury which is obviously very high urgency and high velocity, but when it comes to say family law or estate or probate or other type of bankruptcy stuff people who are stressed at that particular moment

Eric: I was reading a recent journal it was basically from I think it was on Medium or something where they took a small survey from people who were calling firms with AI and a lot of people said we do not want to talk to an AI we want a real human listening to our problems, right? And how would they know it? Because a lot of times when you're talking to AI you put very strong guardrails obviously it cannot show that emotional side which humans can show in that situation where emotions should be the first priority. Don't you think that a system with very strong guardrails might have a very negative effect on your firm's image?

Awais: No, I don't think so. I believe that for instance what we've created is a combination of those two things. I never got the sense that adding guardrails and more and stricter system prompting and so on, getting it tied down cuz you know how it is, I system prompted the other day I was in the lab here doing for hours and hours simply about how the AI was pronouncing email addresses. I mean it was spelling them out. I won't go into the whole thing but it was giving a very unique way of saying the at sign. So we were pretty much focused on the at sign. We had to at one point write a but then it was and then that was kind of messing it up and then it thought it was a sentence and words and then it got confused because it thought it was an email address at first, but then with the at it turned it into an actual sentence so it started adding up. Anyway, I got too deep on that, but anyway we were for hours we were in there on that. I guess I don't really What was the initial Well, I totally missed what did we talk about? What was the

Eric: I asked you that sometimes in a lot of firms, it's more important to show emotions first to the person in distress rather than velocity.

Awais: Yeah.

Eric: Yeah, thanks. So, um, the I I think that the so all of these guardrails, what I think my point was that it we you can we can hammer down guardrails to such a unbelievable granular level. But it's funny cuz the more you do that sometimes, the more uh emotion uh and empathy you can get. Uh I know that dealing with emotion we had to really cuz you have to find a balance between whimsical kind of I oh hi how are you you know like they don't clients don't want to talk to someone who sounds like a you know like a teenager. Um but you also like you said you don't want to like tighten it up so much where it sounds like a robot. So, um I think that you can put in guard rails very stringently without uh sacrificing the empathy. And I feel like I've done it at least. I mean, I we've had no we have no complaints about the voice uh and the and that part about it. You know, the only the only thing is uh is always back of the house stuff like you know, we always get questions on um does it connect to my CRM or to you know to Cleo to my law office software? Uh you a lot of back of the house stuff, programming, how do I change it? What if something goes wrong? Uh but yeah, in terms of the voice, I think we've done a decent job. I think a good company could produce produce that without too much difficulty. They should.

Awais: Yeah. So, so on on this point that you mentioned, this is very important that you said that for a company giving support to its existing customers is very very important. Right? If something goes wrong or maybe there's nothing wrong but the you know the the client feels like it's not working. Uh what is the importance of support? first of all and how do you guys manage uh support right now?

Eric: Support is very important to adoption as well like I think I and sales. I mean I think when the client knows you're going to support them. I think that's one of the biggest fears they have is you're going to sell it to them and then just disappear, you know, you ghost them. Uh that happens with some I've heard stories, you know, sometimes that happens. So what we do is the exact opposite. We're at a certain scale that we're still barely able to do this, but we will go out as long as it's not. Now, some clients don't need this, but we do go to the office. We go. So, I usually go um unless and that matter. We have been obviously out of state. We travel. So, I feel like it's the value of being there, uh putting a face to the name, shaking hands with my client, and explaining it in person because inevitably something comes up, whether it's a legitimate problem or just a question. Uh I can't remember a time where something didn't come up where it was like, uh-oh, like focus like but um and I like to be there. I I I don't like the feeling of being on on a Zoom call and and then being like, I could show you this if I was right there, you know? I would just show you. Um, and also, they're more comfortable. And I think they they deserve that. At this point, we can do that. Um, I'm not sure how much longer we're going to exactly be able to do that, but I intend to hire uh if I can't go, I I intend to have uh staff that will go and be qualified to do what I do. Uh I think it's important to go in person and and onboard them that way.

Awais: Yeah. Yeah. Exactly. So, uh Eric, the thing is that um uh it's my fault. I was basically 16 minutes late today and Eric is so nice. He kept on waiting. Uh I I first I want to apologize for that but we are short on time and Eric is also busy. So what we're going to do is I just want to you know at the end uh people who are listening you see first of all you guys need to set your processes straight and you need to understand that when somebody's into the trenches and he builds something. Eric has been a lawyer himself. He understands every single corner of the software as well as your business. So before you decide that you want to go to some hot short legal tech u thing do contact Eric because he's full of knowledge and he completely understands his product and what it can do and I you know you should get connected with that. So thank you so much Eric for coming in uh today on my podcast. It's always a pleasure and I will definitely if I have more questions about you know legal tech and AI you are my go-to person. and I will invite you again and it would be very nice if you um if you you know let me do that.

Eric: Yeah, I I love being here. Can I just drop my the website and my email address real quick in case anyone wants. Okay, so yeah, if you want to jump on the website, it's brand new website. Uh it's aquire a sqir.io and I'm ericisquire.io. So, if you want to reach out, I I I got people writing to me on LinkedIn, heard the show the last time, so it's not out of the realm of possibility. I'd love to hear from you. So, please write. I always write back.

Awais: So, yeah. Thank you. Yeah. Thank you so much, Eric. Uh, this was lovely. See you again sometime.

Eric: Yep. Bye. Great.

Resources & Contact

📥 Growth Playbooks: White Papers

📊 Book a Strategic Law Firm Growth Audit

Meet the Guest

Eric is a lawyer who created a company called AI Esquire to fix everyday problems that law firms deal with. Because he worked as an attorney, he knows how busy law offices run and how important it is to answer client calls quickly. His company makes AI phone assistants that take calls after hours, ask basic questions to see if a caller has a real legal case, and save all the notes directly into the law firm's regular computer systems. Website: AIEsquire.io

Contact: eric@aiesquire.io