How a 5-Person Law Firm Does the Work of 8 With AI

Collin McKee

Welcome back to the Endeavors AI Podcast. I'm Collin McKee, founder of Endeavors AI — a show about AI becoming a force multiplier for firms willing to move on it. My guest today is Eric Hulett: 30 years a lawyer, a former tax accountant, and the COO of Robles Jr. Law, a veteran-founded personal injury firm out of Fort Worth, Texas. Eric, thanks for making the time — and thanks for the patience. We took a couple swings at this and finally got it to happen.

Eric Hulett

Yeah, thanks, Collin. First — you're not a Texan, I can tell. It's Robles Jr. Law — our Spanish speakers will know the difference. Just a quick little edit there.

Collin McKee

No, I appreciate the correction — I should have known that one better. So let's get to know you, sir. You've described yourself with kind of a unique term — as a platypus in your role — because you're an accountant, marketer, operator, and a closer. So walk me through how you wound up in this kind of job position.

Eric Hulett

Like a lot of people — by necessity. I did start as a tax lawyer, to be precise; the accounting part kind of came with it. I used to be a pension consultant. Then I went to law school because I was working for Arthur Andersen and I wanted the letters after my name, and it was going to be easier for me to be a JD than a CPA, oddly enough. So I went to law school, and I was a pension lawyer for a few years with the big guys. That's kind of what led to the tax lawyering, because that obviously was part of it — a lot of labor work as well. Then I ended up in a situation where I had to find another job, because I had a boss who lied about who was doing all the work. She sold the company, and it was me, but she didn't tell anybody — her ego wouldn't let her. And I was at a salary level where, when you get fired, it's really hard to find another job. They tend not to believe you if you say "it wasn't me, it was them," because we've all heard that and it's never been true — but this actually was true. So it does happen. I started as a family lawyer, and then I ended up in personal injury and kind of fell in love with it. It's also when I found out I actually could talk to people — I could convince them of the need for my services, and I found a little niche there where I was able to bond with these people in distress.

Then the evolution to here: I wanted to get away from being tied to the desk like most lawyers are — or tied to the courtroom, which has the same effect — so I could kind of be anywhere. I think the kids call it a digital nomad. I can work wherever I want to, as long as I've got internet and a phone, which literally is pretty much anywhere on the face of the earth now. With satellite phones as cheap as they are, it really is anywhere. I'm old enough to remember when sat phones were like five bucks a minute, and now it's nothing — it's just a slightly more expensive cell phone.

So that's what led to my description as a platypus, because my job here is — I don't do so much of the lawyering. I'm in charge of the operating of the firm as a company, and I also handle a ton of the marketing. You don't have too many COO-slash-CMOs. Obviously in smaller firms that's possible — Meta certainly wouldn't hire one person to do both jobs; it'd be too big a job. But for us it works for now, and I enjoy the heck out of it. It is definitely something I kind of created — you're not going to find this job category on Indeed, for sure.

Collin McKee

So you've said that, as far as working with Joe, you're more so the face to the industry and Joe is the face to the world.

Eric Hulett

Yes, that's my summary. I'm the one that evaluates, because we get bombarded with offers — and being a startup, we're three years old. We do need some of the products that are being pitched to us. We don't need all of them, and we don't need eight different providers. So it fell to me — and I kind of had to take this from Joe, because he wanted to do it until he realized what a time suck it was. I said, no, this is a big time commitment. It needs to be done, but we don't need you spending ten hours a week listening to pitches. Plus, at this point I've listened to so many pitches I can cut through pretty quick. For instance, when they're pitching SEO to me and it's "we're going to get you on page one" — no, you're not. It doesn't work that way. I don't have to be an expert to know these guys spending millions of dollars a month get way too many unique eyeballs — which are basically IP addresses — looking at them every month for me to be naturally, organically on page one. There are some tricks to get you there, and geo-limiting is different, but in general, those guys I can dispense with in five minutes, where Joe would be like, "that sounds great." I'm like, no, that's a waste of time. We can't do that. And I've done that, and that's worked well.

And he's done really well because of his background — being Hispanic, being a disabled vet, living in a part of Fort Worth where there are no other lawyers. It's worked really well for us in just the old-fashioned way lawyers marketed back before marketing was legal, which is you made a name for yourself in the community and attracted people that way. And that's worked really well.

Collin McKee

So as far as the firm itself — who do you serve? What's the mission there for Robles Jr. Law?

Eric Hulett

We're primarily after the Hispanic — and especially the Spanish-speaking — market in the personal injury space. We will take anybody, and our client list looks like that, but it definitely skews heavily toward Hispanics and Spanish speakers.

Collin McKee

Just because the market's there and you guys are primed to serve it well.

Eric Hulett

Yeah — we have an Hispanic leader, and we're not pretending about it. I do like to kid him that I speak a little better Spanish than he does. But the rest of our staff is all fully fluent, so it's an easy thing for us to do. We're not using Google Translate to try and talk to clients, which is pretty maddening for both parties, honestly.

Collin McKee

Yeah, I'm sure — they appreciate that. So, on the topic of AI, because that's the title of my show: you've been running AI in the firm for a good while, right?

Eric Hulett

Yeah, we're coming up on almost two years of having some form of AI in the firm, which for law firms — especially smaller ones — is fairly quick adoption. I go to the conferences and they're still saying, "hurry up and adopt AI." I'm like, well, we already did that. We're just trying to figure out which of the pieces, and how to put the Lego set together to make the best thing for us.

Collin McKee

So why do you think you were an early adopter? You were comfortable doing it when it was kind of a scary place for a lot of smaller firms. What made you dive in?

Eric Hulett

It's a very technical term: Joe and I are geeky as shit. We're geeks. I love technology — I've always loved technology. I'm old enough to remember when PCs became prevalent, and that was not when I was in high school, it was when I was in college. Weirdly enough, I went to school with Michael Dell. So I've always known — it's pretty easy to see it's the future, and I think fighting it is just futile. It was pretty obvious to see: this is here, it's coming, embrace it or get run over. Joe and I both knew that from the get-go. It was just, okay, we know we need it — but which one? Two years ago there were already probably a hundred offerings out there in various forms, and now it feels like there's ten times that.

Collin McKee

And then there'll be a little bit of consolidation, and then that'll expand and contract, expand and contract. So as far as AI adoption — you made a point that stuck with me, and I think there's a lot of truth to it. You said that a 50-lawyer, 400-staff-member firm looks at AI adoption as: this is going to be two months of moving backwards before I have any ability to go forward. Talk me through the advantage of your firm, as opposed to that, when it comes to AI adoption.

Eric Hulett

It really is the efficiency — as the military would call it, the force multiplier. One person can now do what it would take us one and a half to two people to do, and in a small firm you see those almost instant returns. For us, with the caseload we're running — we're running with five people now, and this caseload three years ago would have taken me eight people, minimum. And that's with us not really fully utilizing it yet, because there is a learning curve. There's certainly resistance within your employees, because they're convinced this is what's going to put them out of work. What most employers, like we have, have done is say: no, it's going to keep us from hiring the next person, but it's not going to make us get rid of you — because that's the only way you can do it without having a revolt. But I've heard it from giant law firms too — one with 200 employees, and he said attrition is going to take seventy people before I have to even think about hiring somebody. That's how much efficiency is being built in with this AI stuff.

Collin McKee

Yeah — and there's still a lot of fear in that, and there is truth to it, because there are going to be some jobs that become obsolete. Go back twenty years and those jobs existed, right?

Eric Hulett

From my perspective — and I've got a kid coming out of college now — this is one of the times the technology revolution is destroying white-collar jobs. It's not really affecting blue-collar. Not in the way the first adoption of computers did, where you got rid of all these file clerks and everything else — or the profession I was in, or obviously the industrial revolution, which wiped out tons of semi-skilled labor jobs. As I told my kid: if you're applying for any job that has the word "analyst" in the title, that job's going away, because the computer's just better than you.

Collin McKee

As far as technology — can you name your technology stack? What you're using for what?

Eric Hulett

Sure. I'll go from entry to exit. Initially, we used some custom-crafted AI agents on our intake that Joe actually built himself. The nice thing is you don't have to be a coder now — if you've got an idea, AI will help you code. So he's built some AI agents to help us with our intake: once we put in the base information we get from the first phone call, it automatically starts emailing the people that need to contact them, sets up the file, that kind of thing. Then, with that, we're using Foundation AI to handle our incoming emails. It reads your emails, figures out where each one needs to go, saves it where it needs to be, and emails whoever needs to know about it. It also updates our case management software with full notes — we got this piece of information, and it's here, it's been saved. They just announced they can also do click-throughs now: if someone sends you a link, it can get through. If it's a HIPAA-secured link, it probably can't, because the password has to be in a separate location — I'm not sure they're there yet. But on the easy ones, like what I get from most insurance companies — where it says "click here" so they can track which IP address pulled it — it will do that, which is nice. They rolled that out in the last two weeks, so a lot of this stuff is still evolving.

Then our main case management support and our demand writing is done with EvenUp. That's kind of a soup-to-nuts thing, and they keep rolling stuff out. Their newest module is a conversation module that makes phone calls and things for you. We haven't implemented that yet, but we probably will. One thing I don't want to implement — a bunch of them will do intake and take the phone calls. For the big guys that get hundreds of phone calls a day, it makes all the sense in the world. But for us, with our marketing, because we get so few phone calls, I think that personal touch really still has value. "Soft skills" is this buzzword they keep throwing out there. I know from doing this for decades: hiring a lawyer is not a logical decision for normal people — it is for giant companies — it's purely emotional. People are always at one of the low points in their life if they're hiring a lawyer, whether you're going through a divorce, dealing with grandma's will, or you're in a car accident. You're not calling a lawyer because it's a good day. You get used to that — no one's ever happy to talk to me. When you're younger it's a little tough at first; at this point, I've got the distance.

And I did overlook one — the other product we're using is Tavrn. We're using them to get medical records, and that's one place where there's a huge savings in time. When you're dealing with big hospitals, especially county hospitals, they're notoriously bad, because there's just no threat to them. You can read the rules — there are punishments for not getting records to people in time — but no county judge is going to fine the county hospital for not getting records out, because they know the county hospital's already broke and it's coming from the same coffer. So they're just not going to do it. They'll never say they're not going to do it; I just know, practically, they don't. They just take time after time, and it's not worth us wasting our time filing an action in court to try and get the fees — "well, they meant to," you know. We all kind of know that instinctively. But the AI just hounding them — it's the squeaky-wheel thing. Eventually you get your records, because they're just tired of us calling them every day. But it's not us — it's the computer, it's AI. That's one of the beauties. And it's also somebody I don't care if it ticks off, because basically it's screaming at City Hall — who cares, they get screamed at all the time; they don't get mad about it. Where you have to be careful, when you set these things up, is when you're calling clients or something like that — that you're not harassing them, where they're like, "why do you keep calling?" There's a sweet spot between "why do you keep calling me?" and "why do you never call me?" We've kind of figured that out, and a lot of times we have little notes about that, which is a little harder to program into an AI system that likes to do things on a set schedule. But I'm sure that will improve over time.

Collin McKee

So give me one of your best war stories. I like to hear when things don't work as well as hearing the positive aspects. Is there anything that broke while you were tinkering around with it — and what did you do to fix it?

Eric Hulett

Well, yeah — like everybody who's an early adopter of AI, especially in the legal space. They have a cute word for it: hallucinations, which is when AI makes shit up. And the problem is it makes shit up very convincingly. We had a demand, and it put a cite in it — and fortunately, in PI, the cites we use tend not to change much. It's not a real dynamic area of the law; if there's a new case, we all know about it. And it cited this case I hadn't heard of, so I clicked through, and it appears to be a case — and then I get on my actual legal software and look for it and realize it doesn't exist. So it's not just that it made one up — it even created something that looked like it.

That's been one of AI's failings: the more something is said, the more it carries a certain level of credibility, and we also know the internet can be an echo chamber, where even if it's crazy, if it's said enough it seems to gain traction. It's getting better, but initially it wasn't really good at evaluating sources — which is something they spent a lot of time teaching us in law school: you can trust this, you can kind of trust this, you don't ever trust this. It just treated everything the same. It's gotten better at that, but it still has a ways to go. So that's the fear — and we've seen all the horror stories, if you watch the papers at all, of lawyers getting yelled at for submitting briefs they clearly never proofread, where the computer went a little crazy.

And like I said, I see that with intake, too, because sometimes I'll combine some stories into one, just to build empathy. I'm not lying, I'm just condensing. But if you tell that to AI, it's basically just been told "I make stuff up." It doesn't know how to evaluate big lies and little lies — the white lies and the ones that are really dangerous. That's where you really have to watch it: you give it creativity and it goes a little nuts. You say "there's an elephant in the room," using allegory, and it goes, "there were ten thousand elephants in the room" — what's the difference? There really wasn't one, so why not ten thousand? A silly example of how extreme it can get.

Collin McKee

Yep. And you were telling me another story where you were dealing with some kind of confusion and you got hacked at the same time — it was like a double whammy. I'm trying to think back; fill in the blanks I'm having about that story.

Eric Hulett

Yeah — like everybody, evidently. Dallas was under cyber attack, and a lot of private firms were having that issue at the same time. We can't see how they got in — they got in through my email. And I'm one of the people that teaches people how not to get got, so it truly can happen to anybody. Not that I'm perfect by any means, but: I don't recognize the sender, I don't click through, I don't put in my own information unless it's someone I deal with all the time or I reached out to them. And it cost us — we lost two, three days of work, because we just had to shut everybody down. We don't know what's going on; we can't do anything right now until we know we've got our arms around it. And that's AI being used against us — there's no two ways about it; there was not a person doing that. Where we used to have passwords where we could say it's going to take two or three years to crack, they can crack pretty much anything in two or three days now, because they can make billions of guesses a second. Unless you've got a 32-character letter-symbol-number combination password — and even that's going to buy you a week. Maybe I'm a bit of a defeatist: it's not if they hack me, but when. I've personally had my identity stolen three times, but I'm always watching it, so I just shut it down when it happens. I think that's all you can do here. I don't think security exists anymore. My analogy: I view the internet like a highway. It's dangerous, but you've got to use it.

Collin McKee

Yep. But you don't have to use it all the time, you know.

Eric Hulett

No — you can choose, and sometimes the side streets are going to be a better way to go, but inevitably you're going to use it. And as someone who lives in a big city, I'm probably going to use it every day. That's one of my analogies I actually like, because it really is that way. Especially doing what I do, with car wrecks — the highway's a dangerous place. That's where the nasty wrecks happen; they don't happen on side streets.

Collin McKee

Well, to keep the analogy going — we talked about AI policy. The highway's got rules of the road that you've got to abide by; if you break them, you have consequences. And you had a funny response: "we're lawyers, we don't write policy until it's too late." Which is a funny thing to say, but at the same time...

Eric Hulett

Yeah — it's like the cobbler doesn't have shoes. We tell everybody what to do, and we assume we know it ourselves. Of course, we rolled out all the training to make sure everybody knew, because we'd actually gotten hacked two weeks before that, where someone clicked on something they weren't supposed to. The classic. And it was, of course, the lowest-paid person in the entire company. They're going to go after the weak link — well, they go after every link; they're going to find out which one is weak. And she didn't know. Before this, she had worked in fast food, so she'd never worked in a professional environment where there was anything worth stealing — hacking your local McDonald's really isn't going to change anything. But here, we do: we've got lots of protected health information and Social Security numbers and all this demographic data that everybody loves, because we know the most valuable commodity now is friggin' information. We're definitely fully in the information society.

So we've now rolled out the training, and we will revisit it every six months — or sooner, if we see something really clever. That's also the thing with AI: I've always gotten fake things, like "you need to pay this bill," and they used to be pretty obvious. They're really good now. My fake — I'll use the real one — Disney Plus bill: the only reason I know it's fake is I get Disney Plus for free with my cell phone; I don't pay the bill. But it looks exactly like when I did pay Disney Plus. Most people — I don't have my account number memorized, which would be the only way I'd catch it, because everything else about me is out there; it's been out there forever. They have my full name, they have the right address, all that kind of stuff. But they don't have the account number, because I don't even know what it is. I only got saved because it's not a bill I pay.

Collin McKee

Right. So if there's a firm leader out there — and our audience generally is small and medium-sized firms, so most likely there's one in the audience — and they're looking to get started with this technology, what kind of advice would you give them?

Eric Hulett

Definitely do it. Especially in the world of personal injury, or anywhere you have to draft demands and litigation, there's very specific stuff that will help you with drafting. And that's where we started — the very first thing we did was the demand product from EvenUp, and it was just doing demands. It's generating a demand in thirty minutes that would take me — as someone who knows what I'm doing — I think four hours to do right: to collect all the medical records and cross-reference them and write up the summary. It still needs editing, I think — but I have some pride in authorship. For our people that don't write very well, it's a game changer, because it comes out professionally written. The tone can be a little stilted, but I'd rather have it accurate and stilted versus getting something where I go, "I'm just going to rewrite this — it's not worth editing." I've had that happen with some case managers who just weren't good writers — great case managers, weren't good writers. And I'm sure people have that too; they know who can write and who can't. But AI lets everybody write. You're only going to know the difference when they get on the phone, and people go, "you didn't write that, did you?" And I would suggest: don't lie about it. Go, "yeah, I definitely had help."

Collin McKee

Yeah — "I let it draft it for me and I put in my personal touches." That's all you can really expect. And for that matter, use the tool — use it, but check it. Trust, but verify.

Eric Hulett

Yeah. It's a tool, it's not a substitute — I think I heard somebody say that, and I think that's a great way to put it. If you use it as a substitute, it will bite you.

Collin McKee

Yep, a hundred percent. So as we close it up — looking back at yourself two years ago, if you could have known anything then that you know now, implementing the systems that you did, what piece of knowledge would you want to take back?

Eric Hulett

It reminds me of a seminar session where the guy said, "hire people before you need them." I'd say: use it before you need it. The problem is that assumes unlimited money — so do it as fast as you can afford. And like I said, don't try to do the whole thing; do it in bite-sized pieces. Get comfortable with it, and figure it out — it is different. All of these overlay slightly differently. They all work with what you have, but they're overlays — usually hidden behind it, but all of a sudden you have new buttons that appear, or new drop-downs in your otherwise normal web page. But do it as soon as you can, because it's not going away. Some leaders are definitely showing up already in this stuff. There are lots of players that are good, but for me, I like going with the one I know is going to be around — because, like you alluded to earlier, there's going to be consolidation here. There are too many players, as with any new tech. Look at how many web search engines there were at the start of the internet, and now we're down to basically one. There are some others that exist, but you'd be hard-pressed — I mean, Yahoo still exists, but who uses it? And that's going to happen here too. And if you have to switch platforms, you're going to go through this kind of implementation pain more than once, and I don't like doing that. So I used a little hesitancy and waited to see who it was going to be — and I got lucky. Almost everybody we've adopted, we've stayed with; we've replaced one. Other than that, it's worked pretty well. But yeah — the sooner the better.

Collin McKee

Get after it, be disciplined, don't be scatterbrained.

Eric Hulett

Yeah — but don't break yourself. Don't financially ruin yourself to do it. That's easy to do, because all of this seems fabulous, and it's kind of like a buffet: you take one of everything, you can't eat it all. It's not going to happen — you've wasted a bunch. You have to be careful there, because it's easy to waste money on this stuff. And also know you can negotiate every price. Most lawyers know that, but it's always worth a reminder, because compared to what I was initially offered — it's kind of like buying a car. I walked away twice, and all of a sudden, wow, I got a whole different level of deal.

Collin McKee

Yeah — sound advice, a hundred percent. Well, Eric, this has been great. I really appreciate it. Where can people learn more about you and the firm?

Eric Hulett

We are on the World Wide Web at Robles Jr. Law — that's R-O-B-L-E-S, roblesjrlaw.com. That is the easiest way to find us.

Collin McKee

There you go — we'll put the link to that in the show notes. For everyone watching, thanks. This has been Endeavors AI. We will see you in the next episode.

Eric Hulett

Thanks, Collin.