Join hosts Jay Ruane and Seth Price in this exciting episode of The Law Firm Blueprint as they kick off a groundbreaking AI series. They are joined by guest host Michael McCready and special guest Jay Madheswaran, Co-Founder of Eve Legal, a cutting-edge AI platform tailored for the legal industry. Jay shares his insights into how Eve Legal is transforming law firm operations by streamlining case evaluations, generating medical chronologies, and automating discovery responses. Learn how lawyers can integrate AI into their daily workflows to save time, improve accuracy, and increase case capacity.
This episode covers practical strategies for utilizing AI effectively, overcoming common challenges, and understanding how to frame queries for optimal results. The team also discusses the future of AI in handling non-digital data like body cam footage and handwritten documents, as well as the immense computing power driving AI advancements.
#LawFirm #Blueprint #AI #SethPrice #JayRuane #LawFirmBlueprint
Links Mentioned
Hello, hello and welcome to this edition of The Law Firm, Blueprint. I’m one of your hosts, Jay Ruane, and joining me, as always is my man Seth Price over there in the DC, Maryland, Virginia, headquarters of BluShark Digital. Seth, you’re rocking the BluShark wear, today you got your whole outfit going here, but we’re also joined today by Mike McCready, who is all things AI in the legal space. But then we’ve got Jay here to talk AI in the legal space. And I kind of feel like the newbie here. I’m just dipping my toe into it. So, Seth, why don’t you talk a little bit about what we’re going to talk about today and how this series is going to go? Oh, Seth, I can’t hear you.
Happy New Year, everybody. We are thrilled to have you here as we start our series on law firms and AI and we have, we’re very pleased to have Michael McCready here as our co-host during this this season of AI, but we’re kicking off the AI series with none other than Jay, who has sort of rocked the legal space with his Eve product. And wanted to welcome you, Jay and just hear a little bit about how you came to, to what you’re doing here.
Fantastic. Thank you so much Seth, it’s great to meet everyone, and Happy New Year everyone as well. Yeah, I’m Jay, one of the Co-Founders of Eve legal. I’ve been in this, in the generally AI space for 15 years now. Started early at Facebook, now called Meta, and saw that, how it evolved over time. And what’s really exciting and new for us in the legal space is, for the first time ever, AI can actually do what we all expected and wanted software to do forever, right? Which is, how does it actually help us with tasks that we’ve done once before? How does it just continue to do that again and again with ideally the same level, if not better, of quality, as what humans can do? And to that end, we basically spent the last four and a half years now, getting into the depth of fine-tuning and making these generative AI models. It used to be called transformers back then, work for the for the legal space. And what we found is we basically made Eve capable of understanding a case and case files related to one particular matter, and be able to assist in every single part of document generation, inside gathering, summarization, and all the tasks that you would typically do in moving that case forward. And since we launched that product, what was it like February of last year? We’ve been fortunate now to have pretty crazy adoption. I think we’ve 5X’d our revenue in the last year alone, and added more than 100 law firms that are using it every single day, which is exciting. And so that’s that’s what I’m super excited by. And as you can imagine, I spend all my time thinking legal and AI and super excited to continue the conversation with my kindred spirits here.
So Mike, I want you to sort of jump in, but really, you know, our purpose here is sort of thinking about how lawyers should be looking at AI. You know, the first iteration came out, we saw the scary news from Jersey where somebody, just like threw an AI produced document in the courts, and it was all scary, but at the end of the day, it is a way to run a firm more efficiently if used within certain parameters. And Michael, I’d love to get your thoughts as it’s, talking to Jay about, where are the places that the easiest, lowest hanging fruit are for lawyers as they start to take this journey of figuring out how AI fits into their whole business model?
Yeah, sure. I mean this, there are so many things that even non tech savvy lawyers can be doing today. You know, the best time to start was yesterday. The second best is today. And you know, you do not need to be a programming genius. You don’t need to even have a computer background. AI makes it so easy for you. Now, what I think is great about Jay’s product, Eve and and something to keep in mind about AI in general is you’ve got to limit your data field, okay, you have to limit what AI is applying to and if, you because, if you don’t do that, then you get these nightmare situations of hallucinations and cases that don’t exist. And all the nightmares, all the reasons why people are saying, oh, AI is not ready yet. Well, no, AI is ready yet. You’re just not using it properly. And and you cannot wait for adoption. You know, I’ve been stressing that to everyone, you can’t wait for this to, to improve or to get, it is here now, and if you don’t, if you’re not on the on the forefront of this, you’re going to be left behind. So I answered about five questions.
Mike, what I’m sort of curious about what I want to sort of focus with Jay here today, since we’re going to have you back week after week, is what are the what is the lowest hanging fruit? Jay, from your perspective, where do you see like if you’re lining up all the different tasks? Because part of it is as lawyers see results. You know, the first time you use Chat GPT to write a thank you letter, you’re like, wow, this is really cool. That saved me time. But from what you’re seeing where there are certain things that take a lot of effort to get up to speed, like, what are the easiest, quickest wins that you’re seeing lawyers adopting AI for?
I think there’s a few classes of problems. So the first class is, what will help save me the most time, so I can increase case capacity, you know, go to my kids, etc, right? So, like, what would save me the most time? And from that perspective, we’re seeing a few patterns emerge already, right? So in in places like labor and employment and personal injury, case evaluation tends to be like the very first place where it’s a very low hanging fruit. Oftentimes, you know, people are following these scripts and deciding whether or not to move it forward to a next step or not. But some of the frontier firms right now are experimenting with, hey, I have this call transcript in places where I can record and how do I just use that to create insights and summaries instantly, right? Because otherwise it would take a long time, and in L and E and some other practice areas where you have to do a bit more claim analysis before you sign a retainer or represent a client, that work can actually take a good chunk of time, depending on how you evaluate and what types of cases you qualify, that could be 40 hours of work, because, you know you, let’s say you’re taking on a new client, and you really want to get in depth of is this really a valid wrongful termination case? And now you can change up your processes to gather more information up front right away, right because now incremental time it takes to crunch, let’s say another 100 pages is basically zero. And before you had to have all of these processes to limit how much information you get up front, right, before you spend more time into it, but now you could actually do more, better, faster. So that’s like the, I think, the first lowest hanging fruit, because people understand summaries, they already have some process in place to, like, summarize, or, or, you know, upload some sort of intake form into their CMS. So that’s like the very first basic level I’ve seen people do, and I can keep going on. There’s probably 10 more that are pretty high value.
Well, Jay, let’s, let’s go off of that. Let’s say that somebody has a transcript or has as a significant volume of data that they wouldn’t be able to, you know, crunch in their head, and you feed it into AI. Can you talk a little bit about how to get the best results out? Right? It’s not just what you feed it, but how you explain the query. Can you talk a little bit about that for us?
Absolutely. What’s interesting is this has turned into a whole field of engineering in the last year and a half in general, like people have probably heard the term prompt engineering, but what this refers to is a lot of what Michael, I think, is getting at, which is, how do you instruct the AI to do what you expect to do? Right? So that the challenge with AI is, like, it’s both really, really clever and really, really silly at the same time, and what we have to get really good at is understanding its limitations and directing it to do exactly what you expected to do. And that’s not easy, and this is obviously where, you know, players like us have done a lot of fine tuning and product improvements to make this easy for people, as simple as you know, on Eve, you can just click a button and analyze your case, right? But without it, what you would have to do is the following: What AI is going to be, still not as good at, is effectively the reasoning side of things, right? So, meaning, you can tell it to summarize a case, but it’s not going to know for, you know, let’s say, when you’re evaluating the case, you identified a lot of bad facts around this particular client. You know, maybe they were doing whatever it is that that results in bad facts. AI is not going to know about bad facts until you tell it, right? So it might summarize things like, oh, this is the name of the client, you know. This is when they got into an accident, yada yada yada. It’ll do the basics, but it won’t have the intelligence to know that is useful to have in your summary. And as a result, when you’re interacting with AI, it’s better to be as precise as possible in terms of exactly what you want out of it, right? So that is what trips up a lot of people, when they start dabbling with raw tools like Chat GPT, is, if you just ask it to summarize, it’s going to do its best. And if you ask incorrect follow up questions, that gives it freedom to go outside the boundaries of just summarization, it’s going to start making up things just to satisfy you.
And with with the query itself, like, like you said, I mean, with Eve and some of these other programs, there’s been a lot of, a lot of work that has gone in to train that AI model. But even with a, you know, with a simple custom GPT, if you frame the query properly, for example, you know, you want to know what bad facts are in in in a medical record. You know, there are ways of creating that query by saying you are an expert orthopedic surgeon reviewing medical records in favor of an insurance defense adjuster. You know what, what are you seeing in here that could help defend the case on medical causality? So, so framing the question, but, but with Eve and other programs, you know, you’ve done that 1000s and 1000s and 1000s of times, but talk a little bit about how you frame the question as to what response you want to get out of it?
So this is a very fascinating topic, as you can imagine, we’ve hired engineers, OpenAI and other places to work on this problem for us, and it is surprisingly challenging from a few different places, right? So for providers like us, what matters is accuracy is everything that customers care about, right? Law firms don’t want hallucinations. They want it to do what they expect it to. It’s simple to say, but it’s really difficult to do, because, as a result, what we have to do to make that happen is break down. Let’s say you want a medical record summarized, or some sort of medical chronology to be produced, right? A medical chronology will have multiple sections to it. It’ll have a timeline section. It’ll have a section that goes through, analyzes each and every single medical record, pulls out, you know, provider summaries, you know, bad facts, yada, yada, yada. And the challenge here is, for every single one of those sections, you might have to do a different thing, right? So that the type of strategy you’re talking about Michael is known as like impersonation. Effectively, you’re instructing the AI to be a person, but what actually happens underneath the hood is, when you tell it to do that, it starts correlating similar topics in creating that summary. So for example, if you tell it to analyze a medical record as a personal injury attorney, it might know, hey, maybe this person, you know, there might be some facts they care about that normal people wouldn’t care about. It would kind of know that. But the challenge is, their data set still isn’t good enough to know the ins and outs of how PI attorneys, or L and E attorneys actually do their day to day work, which is the type of the last mile we solve for so that’s, that’s one part. Second part is this reasoning side. So what can actually help is you can give in your prompts. You can give Chat GPT, Open AI, these AI models really, examples of past summaries you’ve done. So that’s a very, very low hanging fruit you can try to apply if you’re using these simple tools, right? You can just say, Here’s how I’ve summarized these cases before. Try to follow in my example, right? And you have to be a bit careful in how you do it, because it can hallucinate. But if you do it right, it’s actually able to understand the format, kind of, the format you wanted to follow, and what type of facts you’d like to pull out, and how you’d like to summarize your documents. And this is like the beginning stages of being able to use your own data set to get the results you want.
So something that we’ve been doing for quite some time are deposition summaries. And like you said, it’s a perfect opportunity, a perfect tool for AI, but if you upload a deposition and say, summarize this deposition, all you’re going to get is a summary that you know, that’s not what I want. So the next question I asked was, what types of deposition summaries do lawyers typically ask for? And it’ll give me a list of 10 different types of deposition summaries. And then I can ask, you know, well, give me a deposition summary such as this, focusing on liability. But that only gets you so far. But what we ended up doing is exactly what you had recommended, Jay, is we uploaded a whole bunch of our deposition summaries and said, please summarize this deposition using the style of these prior deposition summaries. And then we went one step further. We’ve got, we’ve got 12 lawyers, and everybody likes their summaries a little bit different. And so, you know, each person was able to customize the way that their deposition summary would come out. But what you’re doing is you’re teaching the AI what you want it to produce, and continually giving it feedback, you know, on Chat GPT, you see the little up and down thumbs? You know that, that’s somewhat helpful. But if you can, if you can really explain to the model what it is that you like or what it is you dislike, it’s going to learn every single time.
So Jay, my question really relates to the non PI, non labor and employment field, because I can see having a little bit of experience in those areas, just the voluminous digitally created documents, 1000s of emails, 1000s of pages of medical records. And it’s great to be able to upload these digital documents and be able to spot trends and inconsistencies and that type of thing. In my world and in the world of other areas of law, I’m getting handwritten reports. I’m getting handwritten statements. I’m getting body cam videos. You know, seven different body cams at the same situation, but I’m getting different body cam perspectives. Where do you see the AI part of AI going to allow me to be able to benefit from models such as these? Because right now, if I upload, you know, a narrative report, a couple of handwritten reports, you know, statements from the scene, I’m not getting quality back out even with the best prompt. So I, you know, it may be, I may be too ahead of the time in the criminal world compared to other areas of law. But obviously you have a roadmap, so you can see where these things are going, where guys like I can’t. So where do you think that, that the technology is going for stuff that I need to do?
Yeah, this is a really, really good and important question. I mean, this is the thing to keep in mind: AI is still technology, you know, it’s going to make improvements in specific areas, and you need a lot of engineering to connect those components together to provide value to you, right? And the types of things you’re talking about, Jay, where you have, you know, body cameras, video, you know what, this, basically different modalities of data types, you know, you have videos, audios, and you all have to connect them together. That is still not within the realm of, call it general-purpose AI today. So today, where AI is farthest along in is text-related information. So where you have a lot of text, it’s able to help you. And the tricks people are doing is converting audio and video to text and then helping you further. It’s actually, you know, what we do right? When you get a transcript, a call audio, we convert that into text and then deliver insights from it. So AI is still not, it’s barely getting there right where it’s able to respond back in kind in audio. But this is a huge area of research that people like OpenAI and others are spending literally billions of dollars to figure out and, yeah.
You know, so it’s interesting, because I think we’re all aware of the fact that AI can create video, right, and create all, you know, basically almost anything that you want and mimic your voice and your mannerisms, but, but connecting from, from creating the video and analyzing another video, that connection just hasn’t been made. Is that a fair statement? I mean, at the level that we’re at right now.
That’s right, there’s, they’re related, but very different properties when it comes to AI like they’re kind of related, but to actually understand a video requires a different type of AI model than one that’s capable of generating a video. And what’s interesting is we’ve found a monetary reason to generate images and video. It goes after and actually unlocks a pretty big market, and that got a lot of funding behind it, but we’re now, there’s actually a lot of companies coming out that at least I’m familiar with in the Bay Area having been an investor before, that are quite promising in being able to understand video and images to the same level as people. But the main challenge we’re running into is people might not realize this, but once we’re born, we’re looking at video and audio all the time, right? And this sounds simple to say, but this is petabytes and petabytes of information and data. So AI still hasn’t been trained on this level of data yet to get it to human level of expertise in video and audio. And you can see this happening. Look at how long Wayve has been trying to sell trading cars, right? It seems like an easy problem, but it really isn’t, because this last mile of understanding what it sees is just so difficult for AI, but it’s one that I think will be solved in the next 10 years.
And what about, what about the computing power that all of this takes? I don’t think that people understand, everybody’s got a laptop or a PC, and they think, Oh, I’ve got the fastest Mac. There is. Talk about the exponential factor of how much computing power all this AI is taking.
Yeah, maybe this is a, maybe spicy prediction for 2025, 2026, I think AI is going to be the largest contributor to nuclear power coming out, because it needs so much power to run. And we’re starting to officially get into the bottlenecks of, you know, power grids themselves, not having enough power to run these AI models, which is crazy to think about, right? That we got there this quickly, but it takes a tremendous amount of power, and the main reason is, is GPUs, right? So GPUs, or graphic units, are the building blocks of how AI works. AI is effectively, at the end of it, tons and tons of math that looks like matrix multiplication, and you need graphic cards which can do that way, way faster than CPUs. But unfortunately, all these graphic cards take up a lot, a lot of power, not to mention cost. What is interesting, though, is in engineering, we’ve now found a way to convert these large models into much smaller but equally capable models, which is why, if you guys have used Chat GPT, you’ll see a little drop down for something called O1 Mini or Floro Mini. And these mini models are, effectively take a fraction of the resources and allow them to not just be cheaper to run, but also way faster. And this is a tactic that’s proven to be very popular right now. It’s called distillation, but it’s where you take a large model and kind of train a simpler, smaller model to be almost as good. And there’s a bunch of tricks you can do to make it equally good.
You know. And at the end of the day, for even us lawyers, you know, we don’t need all that computing power. Okay, it doesn’t take a lot to summarize a deposition. Even, you know, even more complex uses of AI in the legal field, you know, are not the same as running coding programs and scientific experiments and so forth that AI can use. So, you know, we always try. Everybody in my firm always uses the mini model. There’s no reason to use up that much bandwidth.
It’s actually kind of tricky. Michael, like people may not realize this, but there’s many cases where you actually want to utilize that extra compute. So an example actually is, let’s say, deposition cases, right? I’m sure people are aware of, but you cannot just upload an entire, depending on how large it is, an entire deposition into Chat GPT, until it is summarized, because it only has so much it’s able to pay attention to at a time. And as a result, there’s many tricks that result in significantly higher levels of accuracy and quality. An example of this could be, you tell AI to look at like page one, do some analysis. Look at page two, do some analysis, and combine them at the end, right? And here, it actually would make sense to utilize, you know, call it either larger models or make way more compute calls than you would think to get higher quality results. And this is why, like, it’s not, as you know, we shouldn’t take it for granted, right? So, like, this is why there’s still a lot of investment going into how do we spend even more compute to get better levels of accuracy out? So that, I think that’s what’s going to result in, like I predict, honestly, even this year, we’re going to see bigger and larger, larger data centers get built out, even ahead of commercial traction. And I think it’s really next year when real commercial traction. Hopefully players like us show that it makes sense to spend money on AI for it to keep continuing, to make monetary sense to reinvest in it.
But you know, for lawyers that really haven’t dipped their toe into AI, talk a little bit about the companies like yourself and other companies that can do the work for you. There are a lot of things that you can do yourself internally, but there are certainly lots of, you know, people in this space that can get you up to speed faster.
Absolutely. So a lot of what we help with is really any labor-intensive task, all the way from, you know, case intake to resolution. So the types of high-value problems Eve helps with is like we talked about case evaluation, but really big ones in PI are medical overviews, right, producing really in-depth medical chronologies out of really 1000s upon 1000s of pages of medical records. And the key difference now is it returns back in minutes, instead of taking days by having, you know, people really do it overseas, right? And creating demands now is really not a big issue at all, right? Now that you have drafting that’s basically solved, and being able to utilize your data and insights into how you resolve cases, is now like an area that’s making a lot of progress and would take a lot of time to do properly. And the next layer that we’re starting to invest a lot in is every firm has some form of SOP, right—standard operating procedures—they want to get their team to follow. And now AI can actually, you know, it’s, it does what you tell it to do, so it can follow it and actually provide a guide for your employees and yourself, honestly, to follow your own guides effectively. And those are all like the immediate value you get from players like us. The second value you get is actually training, which is non-trivial and important, right? So we’ve invested quite a bit in our customer success team and our product, in being able to take lawyers that aren’t used to AI and giving them a path towards becoming good at utilizing it, while avoiding a lot of the traps like hallucinations and being super precise about the questions you ask to the point where it looks like engineering. You don’t do that on our platform, right? And we provide validation frameworks. So when, when Eve actually suggests responses, we use non-AI techniques to validate its response and mark out, hey, did we find this quote in the actual document, or did Eve just make it up? Right? It kind of tells you all of this, and these are all safeguards to give you kind of an on-ramp into safely adopting AI. And what we’ve seen is people that are using it for a few months, they just start making it part of their normal working procedures. They get used to, this is the type of thing AI is good at. This is what it’s bad at. They form a mental model that’s difficult to explain, but that’s the next level of jump, which honestly sounds like Michael, you understand, right? If you’re talking about things like different ways of summarizing depositions, you’ve kind of gotten that, I think.
Yeah, but going back to something that is very simple that people can do today, you know, uploading your SOPs. Listen, we’ve got SOPs all over the place. I wish I, I’ve got a, you know, concrete handbook, but, or the office, our Office Handbook. So everybody’s got the Office Handbook, and let’s say that we had a question on it. You could go to the, to the, to the index, and kind of flip to page 35 and see, oh, yes, we do have Juneteenth off, or if you’ve got your, if you’ve got your handbook in, in an AI Chat GPT, you can just type in, does the firm allow a day off for Juneteenth? And it’s going to give you the response right away. But here’s, here’s what I like is that you can, you can upload your handbook, and even if the precise answer is not in the handbook, it will give you an answer. So for example, can I be disciplined if I swear at my supervisor? Okay? You know that answer is not in there, okay. But you know what,
It’s in Jay’s handbook of course. Comes up a lot more in Jay’s office.
We swear all the time in my office!
Well, well, the answer to you is, it’s not, no discipline in your office, but in my office, it would cite you to the point that, you know, insubordination is something that can be, you know, subject to discipline. It’ll also cite you to the part that office decor does not allow swearing that we should, you know, we should work in a professional environment. So it will take those two separate distinct areas to come up with an answer that says, if you swear at your supervisor, it is more likely to result in discipline, and it will give you the cites from where it came to that conclusion. So, so, you know, that’s an absurd example, but you know, that’s what AI can do when you upload, you know, a handbook, or something like that.
It would seem to me that the best use cases for AI in the market, as we see it right now, is for discrete tasks done perfectly, if you can give them the proper prompt and the proper materials to do the task. You’re not going to get the, hey, this, you know, connects with something else in another case that, oh, by the way, this arresting officer for me, this arresting officer in this case was a backup officer in another case, and guess what? He’s saying he was at the same place, two different places at the same time. That’s something that maybe a human needs to stitch together, because you wouldn’t necessarily ask that in a prompt, but if I need to say, give me the timeline based on these reports, it can absolutely do that better and faster than a young lawyer or a paralegal can do that right now, and we are, I mean, AI, I know you’ve been working in the field for 15 years, but consumer grade AI really is in its infancy, right?
Yeah, I can, I can maybe take and build on top of that, Jay. So based on, you know, I agree with what Michael is saying. You know, I think there’s one other case apart from discrete tasks that is immensely valuable. So generally speaking, Jay, I think you got it, which is effectively AI over time, and honestly it’s getting there now, is going to get really good at if you can define a concrete task, it’ll, it’ll do it right? So that’s, that’s a given, AI is going to get there. It’s going to be good for common tasks that we can define, writing complaints, discovery requests, responses, etc. All of those things, I think will get to the point where that’s more or less a button to push. And on Eve, most of those are buttons to push already. But there’s a second component, which I think Michael talked about, which was pretty crucial, which is, no matter what we do, where we as humans add value is at the edges. You know, we’re trying to apply our reasoning to try to figure out, you know, now that all these tasks exist, where can we add incremental value? And I think this is inevitably going to result in a lot of ad hoc tasks or questions. And that’s where, what Michael suggested is really valuable, right? Using this type of search capability that AI has, that’s more than just normal search, you know, it’s really difficult to describe, but the example Michael gave around, you know, can I swear in the office is, is a really good one, because no one is going to document that. But it is possible to use the documentation of everything you have inside your company to be able to answer that question, and that is going to be equally useful, Jay, in the type of work you do, because while you can’t define the task, you actually could ask AI a question like, you know, was there another officer with a body camera, and it would probably help you find based on this data set, here are all the officers that had body cameras, right? It wouldn’t know to do that for you without you asking it, but if you ask it, it is capable of going and searching and finding the information. And I think that’s a skill that’s very worth building up, because I don’t think it’s going away.
So I got a, sort of a concluding question for me, and I’ll throw it back to these guys for what they have. But as lawyers and business owners are trying to figure this out, Jay, we have three options, so to speak, right? Our own GPT that Mike’s been great at, sort of playing around with. There’s the AI that’s being included with existing software. And then there are these new, amazing companies like yours that are out there. How should we process what we use, which for and how you sort of prioritize that? What’s the order of operations for that?
I think I recommend everyone at some point to start with OpenAI, Chat GPT, Anthropic, these type of solutions, just like Michael is doing, because it’s useful to know exactly what these things are capable of, right? And second thing is, when deciding any AI, at least, I’m a very logical person, as probably it’s coming across. But, you know, I would highly recommend there to be very clear determining factors on what the core value you’re getting out of AI is, and it should be trackable, right? So if, if you are spending x hours, and I recommend tracking it, because it’s useful for your business anyway, if, if you’re spending x hours on medical reviews or creating demands and jumping back and forth between these tasks. How many hours is it? And how many hours can we get it down to potentially? And that is valuable, whether that materializes in revenue increase is further down the line, but that is trackable, and you can compare the quality of what vendors like us do to what you can do by yourself on Chat GPT, right? That’s very, very simple to see the differences. I think that’s probably where I recommend starting. So like, start small, start with something free or cheap, and then figure out what are the areas where it can add the most value. Talk to vendors like us that have found value, that other people are finding in our software, so at least you know what others are doing and see if the same thing applies to you. It may or may not. And of course, like we have a free trial, so, you know, might as well try it out and compare.
Absolutely, free trials. I love free trials. I’m looking forward to when Eve Legal is doing police reports, and reviews that for me. You know, it’s funny. One of the things that we have found is our bio. One of the things that my paralegal said was the biggest time suck for them was interviewing the client about their life history, because we like to put together mitigation packages. And so what we did is we used an AI product to place an outbound call for the client. Run them through 75 questions. We get those answers. We turn it into text. We’re able to upload it and create a client narrative, a bio, basically their life story, very easily. And AI can do 90% of the heavy lifting I probably need. I could probably get to 99% if I, you know, used some Zapier or some other things right now. We’re uploading it ourselves, but, you know, it’s amazing when you look to, and I think one of the things Jay that you said is great, turn to your people and say, Where are you spending an inordinate amount of time? And then see where you can use AI to cut that time down, reduce it, shrink it. I think that’s one of the best use cases across the board for lawyers, because, you know, you only get a certain amount of hours in the day, and so if you can buy back time by using these tools, like your tools at Eve, like for creating those medical chronologies. And I remember back in my PI days, you know, the training we had to do to get somebody who was up to speed on how to do this. It was just, it was overwhelming, and then that person would leave for another job, and you have to start somebody new, and now you have a product that can do it for people. That’s a home run. I just, I’m super impressed with the product that you have out there.
Yeah, one of the functions that Jay didn’t mention about Eve, which is probably my favorite, and my team’s favorite, is answering discovery, right? So we have interrogatories from the defendant, and such a pain to answer these. You know, we answer them ourselves. We send them to the client, and, you know, they’re basically answered for you by accessing all the information in the database. Now, it doesn’t have everything, of course, AI, you know, it’s never 100% but when, when, when a tool can get you, when AI can get you, 75% of the way there. 80, Jay 99% of the way there, you know, you have to use that to, for your time savings. It is, I don’t think there’s any AI product. There’s no computer product that comes out of the box perfect, right? You always can find something to improve. And that’s the whole idea. Is AI gives you that foundation, you know, and the rest has to have that human interaction and involvement to make sure it’s exactly what you want to get out of it.
Jay, what is final words for us?
Please try Eve Legal.
That’s a great final word.
eve.legal. And you know, mention the podcast, we would love to bring you on board and kind of show you and go into a lot more depth as well. And the cool part is, in our pilots, we actually let you try your real documents and ideally get some value out of it, even for free. So highly recommend it. And thank you so much Seth, Jay and Mike for having me on.
Awesome.
Thank you so much, folks. That’s going to do it for this edition of The Law Firm Blueprint. Of course, if you want to check us out, you can do so every week. Live 3pm Eastern, 12pm Pacific, live in our Facebook group, The Law Firm Blueprint and also live on LinkedIn, but that’s going to do it for us this week. Jay Madheswaran, did I say it correctly? Close?
Amazing.
Oh, awesome. I’ve been practicing my head since we started, and you know I, you know, I want to be respectful, so I want to make sure I get it right. Michael McCready, thank you so much for being with us. We look forward to seeing you, Mike on future AI episodes. This is the beginning, folks, of a long, long series all about AI. We wanted to kick it off with Jay and his team at Eve because they have a phenomenal product. You should definitely check out. Seth, anything further from you?
No, excited to kick off the year.
Yeah, this is going to be great. So folks, we’re going to have a ton of stuff on AI, be sure to follow us. If you do follow us on the podcast, be sure to give us a five-star review. We’ll see you next week. Bye for now.
Bye.
Subcribe to our newletter to receive news on update
