AI Intake for Personal Injury Law Firms: What’s Actually Workable Today

In this episode of *Personal Injury Marketing Minute*, Lindsay Busfield talks with Jacob Adversine about where AI intake stands today for personal injury law firms. They explore the gap between the hype around AI and the practical realities of using it in one of the most important parts of a legal operation. The conversation focuses on empathy, competence, edge cases, and why intake quality directly affects signed cases and case value. Jacob explains how AI intake has improved in speed, reasoning, and voice quality, while Lindsay grounds the discussion in real-world intake and marketing challenges. Together, they discuss where firms should be cautious, what good AI intake should actually do, and how firms can evaluate vendors intelligently.

Introduction

Lindsay Busfield: Welcome to *Personal Injury Marketing Minute*.

Welcome to the *Personal Injury Marketing Minute*, where we quickly cover the hot topics in the legal marketing world. I’m your host, Lindsay Busfield, the vice president and partner over at Optimize My Firm.

Today, I’m joined by Jacob Adversine, co-founder and CEO of Reflect Legal. Jacob’s background spans engineering and highly reliable system design, and today he focuses on AI built specifically for legal intake and law firm workflows.

Today we’re talking about AI intake for personal injury law firms, and what’s actually workable today, where firms should be cautious, and whether AI is truly ready for a role that can directly affect signed cases and case value.

Thank you so much, Jacob. I’m so excited to have you here on the show. We just got back from PILMMA in San Antonio a couple of weeks ago. I met you out there and was really impressed by what you have going on.

AI Is Everywhere, but What Does AI Intake Actually Look Like?

Lindsay Busfield: Jacob, one of the big topics that we were running into at PILMMA is AI and AI in marketing, AI in operations, AI in case management, and it is getting so much attention right now just because of the versatility that AI offers law firms.

But with all of that in mind, with all of the attention that it’s getting, talk to me a little bit about what AI intake actually looks like in a legal practice.

Jacob Adversine: First of all, thank you for having me here. It’s really cool, the production that you have and everything. It’s very, very cool. Also, like you said, I enjoyed chatting with you at PILMMA by the water container, so I had to go by you guys to say hi pretty often.

Lindsay Busfield: It was a strategic water cooler placement. There was one water cooler.

Jacob Adversine: For us it was similar with the coffee. We were right by the coffee, but the only side effect is I had so much coffee because I saw people there and I was like, oh, I need to go talk to them. I have to get some coffee. I’ll talk to them. So, right effect. Too much coffee.

Lindsay Busfield: Too much. At least they had a little bit of decaf there.

Jacob Adversine: Yeah.

Lindsay Busfield: And it was so cold in that conference room. At one point I was drinking coffee out of necessity to keep from turning into ice.

Jacob Adversine: Yeah. There was a lot of people getting coffee and tea just to stay warm. But no, it was a great time, and thank you for having us.

Going back to your question, what do we see about AI in intake, particularly for PI?

I think I’ll take a quick step back. What you said is very true. Right now there’s AI talk everywhere, and it’s supposed to be. It’s great. But the side effect of it is there’s a lot of noise. You don’t really know what’s legit, what’s not legit. Companies can come and say, we do all of this, and someone else can come and say, hey, we do the same amount. A firm goes and talks to one person and they’re like, nope, this doesn’t work, and you make a blanket assumption across the board.

So it’s a very interesting point right now, and it’s an interesting part of the AI journey specifically with legal being a legacy industry where tech was not really a primary concern, and then all of a sudden AI is just like, oh, by the way, this is happening here and it’s everywhere.

Lindsay Busfield: And it’s touching every part of your business.

Why Intake Matters So Much

Lindsay Busfield: When it comes to AI in particular, I was excited to have you on because intake is clearly one of the most important parts of a legal operation. Coming from marketing, I would say that intake is neck and neck because, as a marketer, I can get clients to the door. I can get those calls coming in. But if your intake team isn’t getting those calls to convert into cases, then that’s hundreds of thousands of dollars that you just flushed away in marketing, and all because of your intake team.

So when you have AI software at its infancy, and we were truly seeing it at its infancy a couple of years ago, and you are trusting it with something as important as your intake at that pivotal moment, it’s a little scary. We definitely have lots of horror stories a couple of years ago with AI intake bots or agents or whatever it was back then. But we also saw some really cool adaptations of it as well.

So for law firms who aren’t familiar with using AI intake or haven’t heard the stories or haven’t seen it done well in practice, can you give us a little bit of an overview of how AI is being used as a part of that intake process?

How Reflect Legal Thinks About AI “Employees”

Jacob Adversine: Yeah, for sure. I can tell you a little bit about what we know. Obviously, us from Reflect Legal, we specifically use the term employees because that’s a little bit about how we see the future and the vision of the company.

We know interactions, law firms, everything is relationship-based. So even from our point of view as a company, thinking of calling them agents or bots, it’s not the vision of what it should be. It should be an employee, because employee means relationship, it means part of your firm.

In our case, we have AI employees that are conversational. They handle the phone calls, text messages, even emails for law firms, particularly on the intake side. Meaning a firm gets a phone call that, instead of an answering service or someone inside the firm or going to voicemail, the AI answers the phone call. It can determine if it’s an existing client, if it’s a new client, if it’s a third party, like other counsel or insurance adjusters, and conversationally it will help them out.

If it’s a new lead, it can do the intake, follow the qualification questions, ask the right things, hear the story, and so on. If it’s an existing client, it can route to the right person and team, give a case update, whatever it is. But everything from the point of view that it’s an extension of the firm.

I agree with you. It’s scary. We see it with law firms. It’s scary to just be like, hey, I’m just going to let AI answer my phone calls. We’re in a business of relationships. People are not going to want to talk to AI. That’s what we hear often. Or, what if AI messes up?

I think you were right in saying that over the years that thing has only adapted or improved. Previously, the capabilities might not have been there. Now, not every company doing it is there, but some of them are doing a good job.

The Real Challenge: Nuance, Reasoning, and Empathy

Jacob Adversine: In the case of us, we knew that the biggest thing for intake was the nuance of the conversations. There’s nuance in the legal part, there’s nuance in the case type, there’s nuance in scenarios that can happen, there’s nuance in empathy and emotions at play.

And we immediately as a team decided that just a chat, if it’s just an agent and just an LLM, it’s not enough to capture that.

We actually took a step back and saw how the human brain works. Not to get too much into it, but there’s a psychologist called Daniel Kahneman. He talks about System 1, System 2 thinking. System 1 is your LLM brain. Think about it that way. Your predictive brain. If I ask you, “Hey, Lindsay, how are you doing?” your almost automatic answer is “good.” You’re predicting what you’re saying next. You’re not having intense thoughts about, well, I’m doing good because I had pancakes today and I feel great.

Lindsay Busfield: I’d be great if I had pancakes today.

Jacob Adversine: I know, I know. But that’s not what I’m actually asking. But if I tell you, “Hey, what’s 17 times 24 and do the mental math?” unless you’re some math Olympian that knows it perfectly, it’s going to be on your Type 2 brain. You’re going to have to be like, wait a minute.

Whenever you have those type of tasks, even for us humans, they even feel physically uncomfortable. If you were to sit there and try to do the math in your head, it feels physically uncomfortable because you have to hold data, memory, and context. Like, okay, I have to carry the one or whatever.

There’s a lot of that type of situation happening in phone calls that the human brain can process and do the reasoning a little bit on the fly, but just an LLM, at the current state, is not the best at. Sometimes a person calling in for a case, they might say something, but what they really mean to say about that something might be slightly different. The LLM is just responding in a predictive way, the statistical next best option. It might not capture that well.

What Makes People Want a Human?

Lindsay Busfield: I guess the biggest challenge that I think of is that empathy piece because there is nuance to empathy and there’s linguistic nuance to empathy. As you said, people might say something, but it might not be literally what they mean by that.

And when somebody is contacting especially a personal injury attorney, they are in crisis mode. They might not be at their top critical thinking capacities. They need to be met with, first, empathy and then the resources that they need in order to move the case forward.

So talk to me about the response that people are having when they talk to the AI employee. Because I know personally, I will sit here, if an AI agent or employee answers the phone, and I just want to talk to somebody now, the most frustrating thing I can do is hear, “Go representative, representative.” So how do you meet that challenge?

Jacob Adversine: Yeah. And we get that all the time, but I’ll actually flip it a little bit to you. When you answer that phone call and you say you’re talking to an agent, what about it makes you think that’s an agent?

Lindsay Busfield: I can pick up on it pretty quickly because they don’t necessarily understand what it is that I’m asking, and I can’t have a back-and-forth conversation with them.

For example, I had to reach out to a medical provider with an issue that I had with a prescription. I said, this is written in a confusing way. Because it was written in a confusing way, I took the wrong dosage amount, and now I am in a crisis situation. I’m talking to an AI agent, and not only are they asking me to clarify what happened when I thought that I presented the information in a very clear way that a human would understand, but they also can’t take action on it independently. So it’s going to eventually have to go to a human in order for a next step to happen, and I would just rather talk to the human immediately.

Competence Creates Trust

Jacob Adversine: Right. I think a lot of the things that you said are true. We recently had a talk about this specific thing, mainly on trust. What makes someone say, I want to talk to a human? What does it actually mean to talk to a human?

It’s not just that the voice sounds nice. In my opinion, that is like 5% of the equation. The real thing, I think, boils down to competence. Is the system smart enough to actually make me feel heard, understand what I’m truly trying to say, and then act on it?

Imagine you have a person that, for some reason, the voice sounds really robotic, but it’s a person. You call, and you’re like, okay, well, you sound robotic, but they get you helped out. They understand what you’re trying to say. If you’re asking questions about your case or you want a status update or you want to talk to your lawyer or leave a message, they can actually execute on your need and make you feel like the action suits what you said.

How much of that impact was on how the voice sounded versus on the competence level?

That’s what, from the beginning, we started work on by building our own reasoning engine. The ability to tackle that competence side, because the models for the voice itself are going to keep getting better.

Even right now, I can tell you, we ran a quick study on some of our calls. We have thousands of calls per week. We anonymize them, and we saw that only 1.6% of the callers asked, “Hey, are you an AI?” Only 1.6%.

That doesn’t mean that other people didn’t notice. That rate at which people might have noticed AI might be higher. But they were helped. They engaged with it and had no problem because of the competence side, because of the reasoning side. They, at the end of the day, are getting to where they need to get, and they’re calling for help for some reason, and they’re getting to that point where they’ve been heard.

I can see that by the conversations, the amount of people that say thank you at the end. “Thank you so much, I appreciate it.” You don’t really say that to a machine, but you say it here.

It’s scary, but I think more than how a voice sounds, it’s the underlying cues, both on the language you use, the word choice, the verbiage, the intonation, just a little bit of empathy throughout everything to signal to the caller that, hey, I actually understand what you’re trying to tell me, and I actually can help you.

Every Caller Feels Unique

Lindsay Busfield: Again, you come back to my example. I know that the case that I was presenting this company with was a unique situation. This isn’t something that they’re going to see 10,000 times over. So yes, I’m going to need some specific help with this that an AI agent just can’t accurately predict.

But on the flip side of that, everybody calling a personal injury attorney is also going to feel like their case is unique and that there are special things about it. They don’t go, oh, a car accident is a car accident is a car accident. An attorney can look at that and say, yes, this is going to be 90% the same case to case to case. The person calling needs to be able to feel like they are unique and be told, yes, you are special, and empathized with on that level as well as having that competency piece moving forward.

So how are you able to adjust for that?

Jacob Adversine: I agree. We as humans, it’s part of our psychology that we feel unique. It’s only me. I’m special. When I call, my case is whatever. A lot of it is being heard. Empathy, more than just word choices, is giving the right verbiage back, the right words to make them understand that I hear you. That’s the biggest part.

It’s also very easy to compare AI at its worst to people at their best. That’s the other flip side.

You’re talking about unique scenarios where maybe it’s not the ten-thousandth one. It actually is something where, could a call center in another country help you with that specific edge case? Well, they could at least know that this is outside of their understanding and then direct return. And exactly that type of level of behavior AI can do.

There are limits to what AI can do. I equate voice AI and intake a lot to self-driving cars. There is the 80%, 90%, whatever you want to call it, of the path that is predictable. Like, hey, go straight, turn right, turn left. But then that 20%, being in a city where there’s going to be buses, pedestrians, people messing with the drivers, all of that, you need to have a system in place to reason through those scenarios and then triage them.

Like you said, “Hey, I understand that this is outside of the scope of my knowledge,” just like a person would. And this is something that should actually then be escalated.

The current AI systems, if architected well, can do that. I hear the limitations. Here’s one example. We sometimes have people call the reception of a firm, the AI answers, and they’re like, “Hey, I’m in the parking lot. How do I get upstairs?”

The AI knows the address of the firm, but it doesn’t know the inside layout, how to get there. So it says, “I am working remote. Let me transfer you to the staff that is in the office and they can probably come out and help you.”

“Okay, thanks.”

Just like you would train an employee, even if you had a VA, they don’t know the layout to your office. So by approaching this through the lens of employee, what do you currently do with your employees if they’re remote, you can actually start carving out a lot of the behavior and how AI can actually handle more than people think.

How AI Intake Has Improved Since 2023 and 2024

Lindsay Busfield: I think that a lot of the reservations that people have is because they are comparing it to AI at its worst, and they saw AI a couple of years ago. For those who weren’t really paying attention to AI in 2023, 2024, can you talk us through a bit of the timeline of where it was, where some of the shortcomings were, and how we’ve adapted through that to where we are now?

Jacob Adversine: Yeah. The first thing is latency and speed. The models have gotten better to the point that you can intake almost in real time. They can intake more data faster at cheaper rates, and when we’re talking about legal intake where someone tells a full case or something like that, we can process that information much faster and then you can build systems around it to make the decisions.

That’s on the reasoning and intelligence side of the models. It’s really come a long, long way. The average population can even see it with ChatGPT and coding. As the years go by, it gets better. Even the images it creates, it’s getting better, better.

Lindsay Busfield: I don’t have eight fingers on one hand.

Jacob Adversine: Yeah. That same progression people see from the consumer side on images and ChatGPT, you can then translate to reasoning and the speed at which it can process things.

Then the other really crucial part is the voice. In order to complete even if it’s 5%, it’s still a very important 5% of the conversation. It needs to feel realistic and it needs to feel in sync with what we expect. If you say something sad, you can’t answer back with a happy tone. Even if you say words that are empathetic, you have to answer in the correct tone.

The voice models keep getting better and better, pretty much month to month. I’m in tech, so I obviously have a biased opinion about voice AI, but I generally think that give it another year, give it a year and a half, most people will not be able to tell they’re talking to AI.

Even right now, I can tell you, out of thousands of interactions, there’s thousands of people that can’t tell. But in the years to come, it’s going to get increasingly, increasingly hard to tell.

Lindsay Busfield: Which is a little terrifying.

Jacob Adversine: Yeah. There’s always going to be people who use it for the worst side, but I get excited because especially in legal, nine out of ten times it’s not a good scenario, especially in PI. You’re in pain, maybe there was a death or something, and having the firm actually help you matters.

Where AI Can Be Better Than Current Human Intake

Jacob Adversine: We do intake audits, so we go and it’s like an undercover-boss test. We ask for information and then we go in. You’d be surprised the number of firms that I call and I’m just on hold for five minutes before someone answers.

Imagine that was just a horrible situation. I lost a loved one or something. I see your ad, I see your marketing, and you say, you can help me. But then I call you and nobody picks up the phone.

I would much rather have an AI tell me, “The attorney can’t help you right now, but we care, and that’s why we’re here, and I’ll take all your information. I will make sure that he gets back to you,” instead of sitting on hold.

Somebody in crisis mode who wants to talk to someone now, they’re not going to sit around.

Lindsay Busfield: Oh yeah. I’m not going to wait.

Jacob Adversine: The other scenario is language. I’m originally from Mexico. I know my people. I know, for example, my mom’s personality. She’s going to want to go and talk to someone in Spanish. She speaks English, but whenever it comes to something emotional and critical like this, she’s not going to feel comfortable fully expressing herself unless it’s her language.

When doing this intake, I’ve been testing firms in Spanish, and they’re like, “Hey, nobody speaks Spanish right now. You call back tomorrow.” Not even, “Hey, we’re going to call you back tomorrow.” Just, “Can you call back? Our Spanish-speaking person is going to be here between one and four.”

So I’m like, “Hey, I have work until five. I can’t call.” It was like, well, sorry. Really? Is this really the sentiment of a place that’s going to take care of me?

That’s a door that opens. A firm where a Spanish speaker, a Chinese speaker, a French speaker, whoever it is, calls a firm in their own language, where they’re comfortable, and hears, “Hey, we will help you. You’re being taken care of. I will make sure that the team has all the details. We’re with you.” Much, much different.

Sure, AI is scary, but when you start comparing it to the reality of how it works with humans, it’s no longer black and white. It’s a little bit more gray.

Humans Aren’t Always Great at Intake Either

Lindsay Busfield: The other flip side of this argument is that a lot of times humans aren’t great at intake either. We have monitored some of the calls from the law firms that we work with to see how their intake is doing, especially when we’re listening to LSA calls.

Nothing is worse than having a real human answer the phone when somebody’s in crisis mode: “Hey, my husband was just in a motorcycle accident and is paralyzed.”

“Can I get your name and number?”

This is coming from a real human and just lacking that empathy and having to train those skills.

Jacob Adversine: And it’s not their fault. Part of how we are, we get dull. Your person answering these phone calls, the first hundred, sure, you have empathy and you’re like, “Hey, I’m so sorry.” Once it’s the thousandth call, it’s just another call. You’re like, okay, let’s see, what’s the name and number?

That’s what we saw as well. We have an empathy score when we do these intake audits, and the amount of empathy that is expected of a human, and what they usually criticize AI for, a lot of these humans don’t even have. And it’s not their fault. It’s hard. Your job every day in and day out is to hear people complain about the worst scenario possible and try to console them. It’s not an easy job for the intake staff either.

Lindsay Busfield: Oh yeah. That’s a huge challenge.

What Law Firms Should Look for in an AI Intake Vendor

Lindsay Busfield: I know that there are a handful of good agencies out there, a handful of good AI workflows out there and AI agents out there. If somebody is listening today and they are first of all deciding whether or not to use an AI employee or AI agent for their intake, and they are evaluating different companies that are out there on the market, what should they be looking for?

Jacob Adversine: The first thing that we tell everyone is whenever you go into a demo, do not let them show you a recording. Do not let them do the demo through a website or a web form. They should be able to give you a phone number that you can call and you can test right there on the spot.

Lindsay Busfield: That’s right. By yourself.

Jacob Adversine: Yes. Because in the age right now, with everything and ChatGPT and Claude Code and all that, it’s very easy to make a flashy demo. You can angle it just right. But like we mentioned in self-driving cars, that 20% of edge cases, that type of stuff, you need to have a system that is well built in place that can handle those things because 100 calls is one thing. Once your firm has thousands of calls, you’re going to run into a lot of these edge cases, and some of these edge cases could be million-dollar cases.

The first thing we always tell the firms, and you don’t have to choose us, we know there’s a lot of stuff out there. I know we do great work. If you come with us, we’ll take care of you. But if you choose any other company, fine. Just make sure that whenever you test, you try this as close as possible to the real deal.

Call it. Throw it some curveballs. Really push it to the limit whenever you demo it. I think that’s probably the first thing, or the most important part out of all of this.

Lindsay Busfield: As you said, you’re not going to know how it handles until you push it yourself, ask it the hard questions, throw it the curveballs, throw it the unique situations and see how it handles, throw it some gibberish and see what it spits back out at you.

Just like you wouldn’t buy a car without test driving, really get in there and see how that baby moves.

How to Get in Touch

Lindsay Busfield: I loved the phone booth that you had at PILMMA. I thought that was a really great booth setup and a way for personal injury attorneys to engage with you guys and really test drive the product.

If somebody is listening today and wants to get in touch, wants to take it for a test drive, how can they get in touch with you?

Jacob Adversine: You can obviously go to our website, `reflectlegal.ai`. If you go and book a meeting, it’ll be straight with me. The reason why we don’t put the phone numbers live is because we want to make sure that if you’re going to come and schedule a meeting with us, I can give you a demo that is going to be to your firm.

If you’re a slip and falls firm, you want to test a slip and fall scenario that pertains to you. You don’t want to just do a generic MVA or whatever it is. You get to see a little bit of the other things that the AI does, how it is embedded into the CRM, the case management system.

You’ll get to talk directly to me or my brother, my co-founder. We’ll be an open book. Any other tips and tricks, if you even want to talk to us before you talk to a computer, go for it. We’ll help out.

The other option is LinkedIn. Just my name, Jacob Adversine.

Closing

Lindsay Busfield: I will make sure that we put the links to those in the show notes so that people want to dig in a little bit further and see the capabilities and how far AI has come in terms of intake.

Thank you so much for joining me on the show today, Jacob. It has been an absolute pleasure. That’s going to wrap up this episode of the *Personal Injury Marketing Minute*.

If you want to learn more about Jacob and Reflect Legal, again, we’ll put those links in the show notes here. And if you are a law firm that is interested in growing and bringing in new cases so that AI intake can talk to them and convert them for you, feel free to reach out to me through `lindsay@optimizemyfirm.com`.

Thank you so much, Jacob. It’s been a pleasure.

Jacob Adversine: Thank you, Lindsay. Appreciate it.