Tim Kowal
And I'm Tim Kowal. Both Jeff and I are certified appellate specialists, and as uncertified podcast hosts, we try to bring our audience of trial and appellate attorneys some legal news and perspectives they can use in their practice. If you find this podcast useful, please recommend it to a colleague.

Jeff Lewis
Yeah, if you find it unhelpful, run it through Claude AI and put it in your own voice and try reading it again.

Tim Kowal
Yeah, don't put it in my voice because I've got a cold, and so our listeners are gonna have to suffer through some gravelly voice on my part. But today, Jeff, we are pleased to bring back a podcast alum Ross Guberman. Ross was one of CAL Podcast's very first guests last time he was on, AI was not even a part of the conversation. And since then, Ross Guberman's BriefCatch has transformed from a writing enhancement word plug-in into an AI-powered platform, or really a suite of tools, which we'll talk about with Ross today under the theme, Practical AI Adoption. Ross is one of the world's leading authorities on legal writing and communication. He founded and is the CEO of the legal editing platform, BriefCatch, and president of Legal Writing Pro. He's conducted thousands of writing workshops across three continents for law firms, judges, agencies, and corporations, and for nearly a decade has been the judiciary's choice to train all new federal judges on opinion writing. And Jeff, I think it's important to remind our audience of Ross's bona fides as a legal writing professional. He's not just an AI tech bro who is going to try to disrupt the industry just for disruption sake. He actually has a passion for linguistics and legal writing and reasoning, and he is leveraging that background into legal AI tools. We wanted to...

Jeff Lewis
Yeah, he was criticizing my writing well before AI did it automatically, you know, going to a seminar and reading his fantastic books. He's long been a critic of lawyers writing and helping us all sharpen our swords, and AI is just gonna help him do it more efficiently.

Tim Kowal
And many of our readers have probably already read his books Point Made, Point Taken, and Deal Struck. So welcome back to the podcast, Ross Guberman.

Ross Guberman
Those were the good old days. That's good. Good to be back with you guys. We have a lot to catch up on.

Tim Kowal
Yeah, we do. The last time you appeared was, gosh, I think it was about five years ago, four years ago. And since then, BriefCatch 4 launched just last month in March, 2000. No, was that 2000? That was a year ago. BriefCatch 4, an AI-powered Bluebook citation correction, and BriefChat, a writing advisor trained on your own body of work. Just last December. In 2025, you raised a series A raise of six million dollars led by Folio, a couple of months ago, in February 2026. I guess this is all Legal Writing Pro, the company that acquired WordRake and its 12 editing patents. So something is cooking there, Ross, and then just last month, March 2026, you launched RealityCheck, an authority verification tool that detects hallucinated citations, misstated holdings, and fabricated quotations. It's being offered to courts also, not just law firms. And I had this product highlighted in my notes, because Jeff and I talk about the steady drumbeat of hallucinated cases and the associated sanctions on them on a weekly basis or biweekly basis. I mean, they are just coming down all the time. And all we can do for our listeners is just tell them, well, you just ought to be just scared to death. And that's about all we could say. Don't try to avoid AI, but also be scared to death of using AI. And this RealityCheck seems like just what the doctor ordered for that. So tell us about what's been cooking, you know, these past, this past year or so. It seems like BriefCatch, when we last talked, was a pre-AI legal tech tool, and it was indispensable at the time. With AI, everything has changed, but it sounds like you have been catching up and staying ahead of the legal tech AI curve.

Ross Guberman
Yeah, I mean, I've been doing everything but sleeping. That's the price I pay for all that excitement you just recounted. Yeah, so, you know, I don't regret, by the way, creating BriefCatch in the pre-AI era because having to do everything like deterministically, as they now call it, like through code, you know, it's kind of like the struggle, the struggle that you need to really, really figure out, you know, your own approach, in my case, to taking legal writing, which can often be a sort of mushy and ephemeral and turning into something that's hard-coded. So we still have that product. We will always have that product. I mean, you do want some predictable, completely possible to manage tools that help you with your writing. But yeah, we've jumped on the AI train. It's been quite a ride so far. As you said, we have three products out. One just came out, RealityCheck, which you mentioned. And then we have a couple of goodies in the store as well.

Tim Kowal
Yeah, so just to catch up our listeners on where BriefCatch stands, it's used in at least 45 AmLaw 200 firms, over 100 agencies, and courts. It's used to train all new federal judges on opinion writing. It's got 99% customer retention. So it's a phenomenal product, and it allows attorneys to put an objective marker on where their writing stands compared to some of the top writers, which is what BriefCatch is trained on, and give you a sense of where your brief can improve and where it's doing okay, and make sure your own skills as a legal writer are improving. So it's a great tool for that. But I know that to use the term deterministic is that the word for the old style of what we used to think of technology was coded?

Ross Guberman
It's, yeah, it's really sad. It's really sad. Like knowing how to code now makes you sound kind of old-fashioned. Yeah, that's sort of the divide right now. You have deterministic products that are code-controlled. And then you have, of course, Gen AI. And you know, we, like I said, we've made the shift, but I'm still a big believer in coded products, because for all the reasons you gave. So, you know, we tried. We've tried with BriefCatch to really nail down how those top writers write. But we did that systematically. So you couldn't — you can take a brief or an opinion and throw it into Claude or a ChatGPT and say, you know, make this sound like Scalia, make this sound like Kagan. And it will do it, but it will do it in sort of impressionistic and unpredictable ways. And it will never be the same twice. And you won't really know what happened, right? You'll just see the result. We kind of did it, you know, again, a little bit the old-fashioned way by really analyzing. We used a lot of tech tools to do the analysis, by the way, but we really tried to analyze wording patterns and the like in those great writers' writing.

Tim Kowal
And are you able to use AI now to assist in improving your own products? Before, I guess it was dependent on Ross Guberman reading all the great writers and extracting lessons from the great writers, and then trying to figure out how to code them deterministically. How does AI factor into the continued development of a tool like BriefCatch, which was envisioned during the pre-AI age, but can probably be supercharged with AI tools?

Ross Guberman
Yeah, a hundred percent. So, you know, it's an exciting time to have a tech startup for a lot of those reasons. So a lot of our internal work is done much more efficiently, much more effectively because of AI tools. And then going back to these, this kind of giant deterministic rule set we have right now, about 14,000 different legal writing and legal editing rules. We still, to this day — I mean, I did it last night — we still use code for them, as I said, but what's really exciting is we also use AI to look at user behavior and which suggestions are the most popular. And you can also use AI to look at rules you've already developed, and it will suggest new ones. You can feed it really great briefs, really great opinions, and do the same. So yeah, right now I'd say we have a really good balance between our kind of legacy coded analysis, which really, by the way, dates back to when I first wrote Point Made in an earlier tech era, really I had a bunch of printed out briefs on our floor, you know, on the carpet with tabs, with a big ugly desktop. So we have that, but then we're just really thrilled about all the possibilities going forward with AI as well.

Jeff Lewis
So Ross, I have a couple of questions. One of the things I love about BriefCatch is it lives inside Microsoft Word as a plugin or add-in, whatever you call it. Is the new RealityCheck gonna be standalone or is it gonna live inside of Microsoft Word?

Ross Guberman
Yeah, great question. People really love staying in one place, right, when they work. They love having the UI in, as you said, Word, you know, maybe sometimes Outlook, but primarily Word for the vast majority of lawyers. So the answer is yes. The other thing is, you know, the typical lawyer doesn't say like, now I'm going to spend 20 minutes editing for style, and now I'm going to spend six minutes fixing my Bluebook, and eight minutes making sure everything I say about the citations is accurate. You often want to do those things all at once. And then of course, once you change one part of your writing, sometimes other problems arise, or we make typos, you know, mess up the Bluebook, and the like. So for those reasons, the answer is yes. But as Tim was suggesting, we have massive interest from the courts in RealityCheck, and courts, of course get briefs in PDF. So, not to get into the gory details, but we have a slightly different tech stack for the courts. So they, because they don't want they don't work in, you know, Word when they're actually reading and evaluating briefs. But for the courts, it's within their own workflow. And then for the lawyers, it's in their interests as well.

Jeff Lewis
That's great. That's awesome that the courts are going to be able to keep up with the lawyers. Let me ask you this. Did I read that your tool is going to help you mimic your own voice, that you can train these products to match your own narrative style? So you don't write like Hemingway or Scalia, but you actually write like Tim Kowal.

Ross Guberman
Or beat them, maybe beat them in some cases, we'll see. Yeah. So, you know, even today it's worth experimenting for those listening. If you just take, you know, your last three meaty products, whatever they were, briefs, motions, letters, memos, and you put them into, you know, an LLM, and you give it, then you give a draft. It actually does a pretty good job by itself of imitating you, but not a great job. So we're actually working on a tool that really knows. So let me back up for a second. The problem with what I just said is it presumes that somebody knows how your style is different from the style of your average lawyer, right? So in order to do that, you really need data. What we're doing is figuring out kind of what baseline legal writing looks like, and then it will take yours or Tim's, and it will compare it, and it will actually discern what the real differences are. So, for example, an LLM might think, " This guy really likes to write about California a lot, but that happens. That's just because that's where you practice. So things like that, that are not really germane. So it's really, I think, one of the ways we can bridge this divide between the excitement around AI and all the anxiety. If you show people that, hey, we get why you're concerned about being usurped. But the truth is, and we try to do this with BriefCatch, frankly, too, a really good legal tech tool can make you be a little bit more like yourself. It can actually push you to sound more like yourself and be more kind of more pointed and more specific, you know, if it's done right. And I think that, you know, that's where the best tools are really headed. It's making you be a little bit more like you and making it easier for you to be more like you and not, not, you know, turn you into this homogenized anonymous sterile voice.

Tim Kowal
Yeah. And that idea excites me because I've been thinking along the same lines from my own efforts in trying to develop our own AI tools, you know, developing Claude skills, for example, and things like that, and figuring out what AI tools we're going to be using. And a lot of attorneys are scared — everyone is scared that AI is going to, beyond just, you know, taking all our jobs, it's going to make us less human. For example, I've been hearing about how do you reduce the load on your inbox and get to inbox zero? We'll just have AI write all the responses to your emails. So now everyone's inboxes are just going to be managed by AI, and responding probably to someone else's AI agent managing their inbox. And then inboxes might as well not even be in English. They could just be in binary as some sort of computer code language. And we'll just take the humans out of it entirely. I thought that's probably the — I mean, there might be some element of that, but there is a way, you were saying, Ross, to make AI let us be more human than we were before. We can strip away some of the rote tasks, the mechanical tasks that humans were wiling away at, so that would free up our time to do the things that only humans can do.

Ross Guberman
That's right, and we've got to be careful not to romanticize ourselves too, though, right? Because if you take the scenario you just gave me, the inbox, what are a lot of emails that you get? There are things like, oh, sorry, I'm running late. I know we had a meeting at one. Can you reschedule for 1:30? You know what? Be my guest, AI. Come in, look at my calendar, and figure out a good time. I'm not giving up a chance to express my soul there, right? That's a sort of logistical email, and a lot of emails are logistical. AI is great for that. Nobody's going to miss that. With writing again — we're not writing sonnets here, right? People do have distinctive styles. They do have voices, but it's not every word, right? You're not going to usually have your own kind of pseudo-patented standard of review or certificate of service. Or when you have a standard under California law for a cause of action, AI can really help you there. And again, it's not crushing your soul. If you make that divide firm enough, right, and you do draw the line somewhere, both with your kind of logistical tasks and your writing tasks, then you actually have more time and more energy for the things you really care about. So that's kind of how I look at it, right? We don't want to romanticize ourselves as humans either. We don't want to reminisce about pre-AI, but let's not pretend that the average lawyer spends all day, sometimes 10 hours a day, pontificating and theorizing and expressing a voice. There are usually quite a few tasks for most lawyers in the workday, and frankly, for most judges, that are a pain in the you know what? And they're not personal, they're not enjoyable. Let's put those in one pile and then keep the rest in the other pile. That's what I say.

Tim Kowal
Yeah, I love that idea, you know, leveraging AI to maximize your pontification. So let's talk about how you're...

Ross Guberman
Yeah, right. So like citation format, right? Like, that's not a good use of law student time. So much of the first year writing classes is spent memorizing all those Bluebook rules — that doesn't make you a better lawyer. Right. I mean, that kind of thing, I think we should celebrate. But the problem is people aren't drawing the line there, right? They're using it for everything now.

Tim Kowal
Yeah, absolutely. That's true. And it's going to require us to, I think, change the way we work. So, for example, most attorneys now are probably doing their drafting and revising of their work product, doing formatting and citation work, simultaneously with doing high-level legal argumentation. And they need to learn to separate those things out and just let the AI tools do all of the drudgery work, and it frees up more of your IQ points to do the latter work.

Ross Guberman
Yeah, I mean, I just this morning I spent time noticing that some of my paragraphs had justified right margins and some didn't and I was, you know, highlighting those paragraphs and going up. Yeah, not fun. Not fun. I think that's a very good example.

Tim Kowal
Yeah. So let's talk about your AI suite of tools. We've got BriefChat, WordRake, and RealityCheck also within that suite. So tell us a little more exactly what BriefChat is. I guess it's trained exclusively on your own books, articles and teaching materials. So it's a different approach from the general purpose AI tools that many lawyers are already using. What's the argument for training a model on your body of expertise versus letting it just draw on everything, the whole universe?

Ross Guberman
Yeah. So BriefChat was actually a fun product to develop from start to finish because it's true that it's trained on my writings, my books, my posts, my LinkedIn posts, but I also have a lot of other things in there that I've approved, like other articles, tips from judges, so the advantage — you're curating a library of authorities that you trust, but you don't have to actually go through and tag every single sentence or create a massive index. AI does that for you. And the result is twofold. So one is we behind the scenes control the format of the output. So I think it's really important to give examples, as you probably know from using my products or reading my books. If you ask a question about what's a good way to write headings in a reply brief so they don't sound just like the ones in the opening brief, it will actually give you examples that I've also chosen. So there's the form and the user experience element. And then the other is just reliability. I actually did take the same kinds of questions people ask in BriefChat and ask them to ChatGPT and Claude. And as I think most people know, not only do you get sometimes sort of off-the-wall answers, but if you ask the same question twice within five minutes, you get very different responses, which is not all bad. But for someone who's trying to learn, it can be incredibly frustrating and confusing. So there's a little bit of that too — having some authority and reliability can help as well.

Jeff Lewis
Hey Ross, I have a question. I'm sorry, I'm a little schizophrenic with my work product in that I write appellate briefs with the tone of Masterpiece Theatre and people wearing white wigs. And then I write anti-SLAPP motions in the trial court where I let it sound a little bit more like Better Call Saul. Do any of your products help me distinguish between those two voices for the two uses?

Ross Guberman
I think, you know, that's what are they called? They call that code switching nowadays, right. And it's hard, cognitively hard. So that would actually be — to go back to our earlier topic — that'd be a really good use case for uploading prior work and having it. I think Tim mentioned turning into a Claude skill. I don't know if everyone knows what that is. It's actually very easy in Claude. So a Claude skill basically takes something that you're doing. So in this case, anti-SLAPP motions, number one, appellate briefs number two, and it takes the input you give it, which could consist of your own briefs or maybe your own thoughts. Maybe you could even spend a minute like, what's the hardest part of switching between one and the other and tell it all that as well. And then every time you have an anti-SLAPP motion, I think it's a backslash and you do like anti-SLAPP, it'll pull up the anti-SLAPP skill. When you have an appellate brief to work on, you have the appellate skill. And it will actually, as time goes on, get to know you better, if you know what I mean. It will start to kind of see more in how you work. And it will help you go from one to the other. And it might even be able to identify for you that you're sounding a little appellate-like right now in your anti-SLAPP motion, or maybe a little too professorial in your anti-SLAPP motion as well.

Tim Kowal
Yeah, and that could be a real thing for the difference between appellate briefs and trial court briefs. Trial court judges are busy judges. They might not want the deep dive. They just want the takeaway, the executive summary.

Ross Guberman
Yeah, especially in the state courts. You know, you can't prove it, right? But I think most people just from observation sense that a lot of the time the trial court judges don't seem to have read the entire brief cover to cover, savoring every word, to put it mildly, because their workloads are just so crushed, their dockets are just absolutely out of control.

Tim Kowal
Yeah, that's one of the sometimes one of the difficult things to break into with attorneys, especially appellate attorneys who really love the craft of the written word. And they assume that their reader is savoring every word choice that they put in and would relinquish that role to AI, you know, through their cold dead fingers. But in reality, there's only so many ways to write a legal argument. You don't want to be too innovative. You just need to be clear and precise.

Ross Guberman
I was going to say one thing that's changed since we last spoke on your podcast is that now there's been almost complete transformation in the judiciary to reading briefs on devices instead of in hard copy. And that's something that started during the pandemic. So, you know, a lot of judges aren't crazy about this. They tell me, you know, off the record, that they prefer the old way so they can kind of thumb through it. But the corollary is that, as we all know, people don't read as carefully on screens, right? Their eyes want to go faster and faster and faster. So you have a really bad combination right now in the courts. I don't think a lot of litigators realize this. You have, number one, absolutely shorter attention spans at all generations, right? Not just young people, but number two, you have all these judges who are already staring at screens during the day when they go home and they have to read the briefs, you know, when they're commuting or at night, they're staring at screens even more. So those attention-grabbing skills are probably more valuable than ever.

Tim Kowal
Yeah, I think you had mentioned somewhere about how when someone is an advocate and they announce themselves as an advocate or they're identified as an advocate, their ability to persuade someone is only about 6% or something. If they're speaking normally, if they're not identified as an advocate, that percentage triples to like 18%. And it's just being identified as an advocate, as someone who is trying to sell me something, that automatically depresses your ability to do the one thing that you're supposed to do, which is to persuade. I don't know, something you said just reminded me that you had mentioned that.

Ross Guberman
Yeah, if it's even more so nowadays when we're bombarded with messages, people trying to change our views on current events, on popular figures and culture all the time. So maybe we're a tougher audience and our defenses go up. More jaded and more suspicious, even though intellectually we know that an advocate is doing their job. I mean, that is not something we should be running away from, but yeah, we do, because we feel like we're being sold on something.

Tim Kowal
But when you mentioned that maybe some judges are using devices more to read, and you mentioned that there's a different process of ingestion for a reader who is reading on a paper copy versus on a device, is that something that we as writers, legal writers, should be taking account of? Is there something that we can do in the style or tone, punchiness of our briefs to make it cater to a device-using audience?

Ross Guberman
Yeah, I mean, I could suggest a couple of things I've picked up on mainly just from talking to judges, because I don't actually read briefs all day long the way they do. So one is to be really, really careful about footnotes more than ever, because it's a very practical thing. When judges see a proposition and the lawyer puts the citations in footnotes, or there's something substantive in footnotes, of course, as you would do in hard copy, you look down at the footnote. The problem is on devices, it's really hard for them to remember where they were. And it's really irritating, because they have to go back to the screen. And there's something about hard copy — I don't understand all the neuroscience here — but with hard copy, your brain does remember where you were on the page. That does not happen on the screen. So I'd be extra cautious about footnotes these days. And then the other, which has always been true, even in the paper hard copy era, is that wide margins and just general white space are more important than ever in this kind of digital reading era. The last thing is that it's interesting — you'd think one good thing about devices is you'd be able to access the record in the cases and the other side's brief more readily. But that sort of presumes a lot of facility with tabs and hyperlinks that a lot of judges don't have. So actually what happens is in the olden days, they used to kind of go back and forth to have the different briefs. Maybe I'd see them sometimes around their desks or on the airplane and they could just pick one up. So now they don't really do that. They kind of force themselves to finish one brief before they go to the next. So you have almost in a way like a higher burden to sustain their attention and not make them get restless. I wouldn't necessarily say the briefs need to be shorter in the sense of the number of words, but a lot more paragraph breaks, section breaks, wide margins, white space, larger fonts — certainly, even though a lo... [truncated, 2376 chars]

Tim Kowal
Yeah, I'll have to bring that up with our friend Ryan McCarl, who is on team — I think Brian Gardner supports using footnotes.

Jeff Lewis
Yeah.

Ross Guberman
Oh, I love that guy. We went to the same law school, by the way.

Tim Kowal
Yeah, that's right. University of Chicago. And also speaking of which, your new product will...

Ross Guberman
Which is by the way now ranked number two — I saw we're tied with Yale as of last night. Not that I'm competitive.

Tim Kowal
You had mentioned that coming soon will be a cite check tool for state court style guides such as the California Style Manual. That's another one where Ryan McCarl and I disagree. Sometimes we work on a brief together and I always have to make sure to dust off my Bluebook because he hates the California Style Manual with a passion, but it is the format of choice in state courts. So Jeff and I always use it. And we always find that it's given a short shrift or no shrift at all by many of these cite checking tools. But it looks like you're going to be covering that basis for us finally in California.

Ross Guberman
Yeah. So here's the interesting thing about legal tech. There's a point where we've moved, like almost all legal tech startups, from deterministic products to AI, but there's also a point where you move from just general AI products, which we have now, to context dependent AI. You might've heard this is kind of trendy to talk about context engineering instead of prompt engineering. I don't really get into the wording game — I'm not sure those two are as different as people say — but it is absolutely true that our next product, once I get one good night of sleep, I'll start our next product, maybe tomorrow actually. And our next product is going to — I can't say too much about it yet — but one of the things it's gonna do is if you upload a brief, it's going to figure out a lot about your particular situation right away. So it will certainly know the jurisdiction. It will know the citation style that you follow. It will know who the judge is. It will probably know a few things about the judge. Before you even know what happened, it will have already looked up the court formatting rules to see if you comply. It will probably know something about other briefs you've filed yourself in that court, which again can help you. And it might even know a little bit about writing preferences in that court. So there's no way, even if we had a team of a thousand here, we couldn't code all those permutations and combinations. It's just not possible. By the way, we also hear from Florida lawyers — they want Florida citation style. New York lawyers want New York citation. So what's going to happen is it's going to figure out all the things you need and then it will draw on all the resources at once. So again, not just pull up — if you're writing an appellate brief in the California appellate courts, it will not just know what citation style to choose. It will know writing preferences and quirks and know about California law and the like. So that's the next wave, at least here at Brief... [truncated, 2632 chars]

Jeff Lewis
Let's stick with the Black Mirror theme. I want to ask you a question about something we haven't talked about yet. You talked about assisting judges in courts with catching up with AI. We haven't talked about ADR yet. And I get an email, I don't know, once a week from the American Arbitration Association about how excited they are about AI decision making and how we should all be thrilled by it. Do you have any concerns about either bias or too much delegation in ADR use of AI?

Ross Guberman
Yeah, I mean, it's a fascinating topic. And I think the ground is shifting right now, maybe just based on reality. So, you know, first of all, I would just distinguish between, let's say, mediation and arbitration. So for things like mediation, I can totally see the use case. Again, you'd need both sides to opt in. It could probably help you reach some common ground. It could at least help you understand the other party's position. When I was in practice, I did one mediation. I remember how much those mediators charged. It was a lot. So I can actually see that. But that's not what you're probably concerned about, right? Because that's a voluntary process. It's non-binding. So that one, I think, is an easy sell, especially for all the people who can't really afford lawyers, right? And really don't have the resources to go to trial. Now, when you talk about binding arbitration, or of course there's also talk about AI judges in the district court or the trial courts too. That's where, of course, people get quite animated and all these concerns about bias and delegation come up. So I don't have any answers for you, but I can tell you some things I wonder. So what about all the times people talk about judges being biased, right? Or even panels being biased? And I don't even necessarily mean politically biased. I mean, for example, I was reminded the other day that there was a study — I think it's an academic study — that criminal defendants get the longest sentences if they're sentenced right before lunchtime because the judges are not in as good of a mood because they're hungry. I mean, that's not political bias or ideological bias or any other kind of known bias. It's hunger bias, stomach bias. And I believe juries are known to almost always reach a verdict on Friday afternoon, right? They don't want to come back on Mondays. Suddenly the light bulb goes off because they want to go home for the weekend and not have to come back on Monday. So yes, of course, I'd be conce... [truncated, 2818 chars]

Tim Kowal
Yeah, I've barely scratched the surface of my outline. I'm on question two of like 20, so we're not going to make all the questions, but I wanted to make sure we cover all the products in your AI suite. So we talked about BriefChat a while back. And by the way, does that live in Microsoft Word or is that outside?

Ross Guberman
It does. Yeah. For the same reasons we discussed, you know, people want to ask questions as they're actually working. And then RealityCheck is our newest product.

Tim Kowal
Yeah, and then let's make sure we cover WordRake also. That acquisition brings 12 editing patents into BriefCatch. So what does that actually mean? What are these patents about? How is it going to enhance the product? And is WordRake a separate product from BriefCatch or is it enhancing BriefCatch?

Ross Guberman
Yeah, so we did acquire WordRake recently, the company, and it was quite exciting. I've always admired Gary Kinder, the founder. And, you know, WordRake was the earliest by far — the earliest legal editing tool — and had, speaking of patents, what at the time was really innovative, automatic track changes and editing. They kind of have their own way of doing that. So yeah, we did get the patents — their 12. We also have our own patents and we have two pending, so we could have a nice suite of 20-something patents soon. But frankly, that wasn't what excited me. What really excited me was the customer base and just the general passion for legal writing and the work they've done to create algorithms in a different way than we had. I'm very happy — we have their chief editing guru. She works for us now full time and she's fantastic. It's been really great. We're migrating their customers over. People still have contracts, but they'll all be on BriefCatch very soon. Because we offer all that and a lot more. And plus, we've made our own product better already by integrating some of the great things that WordRake was doing.

Tim Kowal
So our existing BriefCatch customers are going to see those WordRake enhancements with the existing BriefCatch product.

Ross Guberman
Yeah, they already have. That's what I did while the negotiations were going on — the M&A part. I went to my true love, which is legal writing rules.

Tim Kowal
Excellent. Okay. RealityCheck. I want to talk about RealityCheck. You're offering this to courts as well as law firms. Walk us through what it does. It's obvious why firms want it because it's going to help stop hallucinations. So tell us how it does that and why do courts want it?

Ross Guberman
Yeah, so really exciting brand new product, lots of work, lots of thought went into it. So essentially what happens is, let's say you have a draft of a motion or brief. So we'll do two things right away. We'll take every citation you have and it will figure out exactly what you're saying about the authority. So that's already something different from what people are expecting. So when you hear hallucination, you're usually thinking someone cites a case that doesn't exist, which is true in many of the cases that dominate the legal news. More and more, by the way, the LLMs are better at not doing this, but there's a new problem you're hearing about a lot more, and that is people cite an accurate case — they got it through AI — they cite an accurate case, but what they say about the case is wrong. The quotes don't exist. They have the holding wrong, or maybe they say the judge relied on facts. Maybe the party's names, if they mention the party's names, they're completely made up names. It might exist somewhere in the common law, but not in that case. So once that's all kind of figured out — like what are you saying about each authority — then RealityCheck finds the authority and then we have a magic process I can't get into. And that process matches everything up and it lets you know whether everything is on point or not. So what's exciting about it, besides that it solves the classic hallucination problem, is it actually catches a lot of mistakes that incredibly diligent lawyers make because we're human, right? So a tiny little error in copying a quote, a little bit of a misstatement on a holding, maybe you say the judge relied on a fact but the judge didn't really — it was just said in passing. So these were issues way before Gen AI and way before everyone was talking about hallucinations. So one of the things I do for fun since I don't sleep — I've actually taken a lot of briefs from the 1990s, obviously did not have AI hallucinations — and it actually finds a lot o... [truncated, 2187 chars]

Tim Kowal
What are those kinds of classic human errors?

Ross Guberman
There are often quotations that are a little bit off, not the kind where people are obviously lying — not completely changing what the court said — but one word is missing or you kind of truncated something. It finds those kinds of errors. But the most interesting ones it finds, at least to me, the most valuable ones, are slightly off renditions of the holding. We don't have time to trace all the history here, but in a couple of cases, I noticed that the court also, in its opinion, found that same error in the brief. If they'd had our tool back in 1995 — which is impossible since I was in law school — they would have found it. So that's my answer. And people say, well, Ross, you're going to make it easier for people to use Gen AI to write briefs and cheat. What I point out is that's kind of not the headline value of the product, but the real value is it just helps you with your treatment of authorities all around.

Tim Kowal
The way RealityCheck does this, I guess, is through a two-layer verification process. It starts with a deterministic layer. So we're circling back to that word that we talked about earlier on. This is the pre-AI human code that checks reporter volumes and case names against databases without using any AI. And then the AI layer evaluates whether the quoted language actually appears in the cited opinion and supports the proposition.

Ross Guberman
Yeah, let me explain why that's not just mumbo jumbo, although it sounds like it, even though we wrote it. So the reason you want a deterministic way to find the cases is it's true that ChatGPT, Grok, Gemini, Claude — they have a lot of cases. I think Musk even said he put all the American cases into Grok. But the problem is it's just out there in this neverland. And it gets mixed up with commentary about the opinions and the like. So if you use AI to find the cases, you'd actually be creating the problem you're trying to solve. You'd probably be telling people that cases that do exist don't exist and messing everything up. So the reason we do the harder thing — which is to actually find the actual case in an actual database — is it's much more reliable. Now, maybe in five years it will be different. I don't know. I don't work at OpenAI or Anthropic. But right now I would not recommend just going to an LLM like ChatGPT or Claude to verify whether the cases you cited exist.

Tim Kowal
You mentioned the Elon Musk example where he uploaded or ingested all of the American cases. Are those just what you'd find on like Findlaw or one of those free sources on the internet?

Ross Guberman
That's a really great question. I don't know if he even really did it — I'm not sure I'm getting the quote exactly right about what he said he did — but they're not uploaded. That's the problem, right? They're not uploaded in the way we think something would be uploaded into a database. No one really knows how these LLMs are formed, but it's not methodical the way you're describing. They're just put into this giant collection of zillions of bits of data and characters.

Tim Kowal
Yeah, and I get it because most of — Jeff and I have used AI and most of our listeners have used AI to see what it looks like if you try to get the answer to legal research prompts using ChatGPT or Claude or whatever. And you can see that sometimes it's right, sometimes it's a little off, but it's always pulling from those just publicly available sources and sometimes blog posts or characterizations in articles and tweets and who knows where — you don't know exactly. So RealityCheck is what you go through and scan — or is it uploaded through?

Ross Guberman
No, we use existing databases that have already been vetted. We're not scanning them ourselves, right? We're going to the actual court databases that are pre-scanned.

Tim Kowal
Actual court databases, so not like Westlaw or Lexis who have their own copyright on them.

Ross Guberman
Correct, we use other ones that are just as reliable. You can't really go through Westlaw or Lexis, nor would you really want to. That would increase the cost considerably.

Tim Kowal
Yeah. Well, you could imagine the prospects — if they would just release APIs to their databases and you could use a foundation model just to connect to Westlaw to get all of their tools, all their content.

Ross Guberman
I think we might be happy, but they are successful corporations and I wouldn't wait by the phone for that. But I don't think that's probably in their business model.

Tim Kowal
So that's the benefit of having the deterministic layer where RealityCheck is drawing on actual reporter volumes. Because what I've been doing by way of explaining — because I will use Claude or ChatGPT to do some legal research, just some initial canvassing. And then I'll read the memo it comes back with and take it with a grain of salt. And then I'll tell it, extract all the authorities so that I can pump those into Westlaw Find & Print. And then I'll extract all the actual Westlaw authorities and then pump it back into the model and say, now tell me — read the actual cases and tell me how this changes things. And sometimes they'll say, yeah, actually reading the actual cases...

Ross Guberman
You need RealityCheck. It'll give you an extra hour or two of freedom a day. That sounds like a lot of work to me. And it's $9.95 just today.

Tim Kowal
You told Above the Law that the question isn't whether to verify your citations. It's whether you want the court to find the errors before you do. And is that just provocative framing or is that the reality?

Ross Guberman
Pretty good line — I think our marketing director liked it. But yeah, it's a provocative line, I won't deny it. But to be more serious, I was actually stunned — I really mean it — how many courts instantly contacted us when that Above the Law story came out. First of all, even the very fancy courts, there are three or four court clerks — it's still a huge burden. It's a huge time cost for the clerks to have to check every single thing that every single lawyer says. But they know they have a lot of power at the courts, and they take their job seriously. And they need to be able to rely on accurate discussions of case law. I don't even know if the Supreme Court clerks are truly able to check every single nook and cranny of the discussion. So yeah, the courts are very excited about this. I'll just say — this kind of product, I mean, of course they're excited about our product, or I wouldn't be telling the story. But let's just say there's a demand for this kind of thing, because courts can see that it will also make everything go faster. If they don't have to spend so much time worrying that the advocates are misleading them, they can push cases through faster and maybe even get better results.

Tim Kowal
So the courts are gonna be using this tool. Some of the court personnel I've spoken with in California say that the California court system does not use AI at all, full stop, at least as of right now. Maybe things will change. What's happening in the courts that you've dealt with? Apparently you've mentioned that RealityCheck is going to be used in the courts — or is it currently being used in the courts already?

Ross Guberman
Yeah, it's going to be used in the courts starting this coming Monday. By the way, when courts say we don't use AI, I just nod — I'm not going to argue — but you should ask them, do they use Westlaw? Do they use Lexis? Do they use any of those products? Those all have AI in them. So I think the divide is shrinking very fast. I mean, when the hallucination scandals broke out, as you know, a lot of courts issued standing orders saying you've got to disclose. So then they applied that to themselves and they were nervous. But I actually know a fair amount about what the courts are doing with AI beyond what they're doing with our products. They are — I don't want to say all of them, but a lot of courts — I think they're using AI products much more than they say publicly and also much more than most people realize. And not necessarily for writing opinions, but for processing the record, for combing through deposition transcripts, for checking to see how far behind they are in their dockets. Even I've heard — some of the savvier judges want to see if they're slower than the other judges on their court processing motions. Some of them are quite interested in predictive analytics. They want to know what AI says they're going to do. Those products that say with this judge, you have a 60% chance of getting summary judgment granted or denied — they want to know too. You can't blame them, right? So they're using it. I think I understand the public statements, but trust me — in six months, I think most courts are going to be using AI.

Tim Kowal
And when they say we don't use AI, I've kind of assumed that, okay, I don't see the courts using ChatGPT or Grok or Claude anytime soon, but targeted tools — specifically tools that are specific to the legal industry, such as RealityCheck and other vetted tools — it seems foolish not to use them for it. And even just to use any AI tool to search through a voluminous record to look for a text string where it might not appear exactly in the form that you search for under a traditional dumb search — it seems foolish not to use AI for that purpose.

Ross Guberman
Yeah, I mean, courts have IT officials, they have tech budgets, they have software budgets. They absolutely know what settings they need and what kinds of accounts they need to have full protection. It's the same with law firms, of course — law firms are very worried about security and privacy with these products. And it all depends on how the account is set up and what protections exist. It's really exciting to see, though. Certainly since I started dealing with the courts, I can't remember any other time of this kind of innovation. It's quite rapid, especially by our profession's standards. Even at one of these legal tech summits I went to — I think it was Judge Scott Schlegel, I'm probably getting his name wrong — he's just so far ahead. I mean talking to him, it's like talking to a tech bro. He's helping all the other courts. There's a lot of energy like that in the state courts and the federal courts.

Tim Kowal
Okay, I have one last question. I don't know if Jeff has anymore, but I wanted to maybe take a little bit more of a 30,000 foot view and ask where you see AI tech moving into the future. And Jeff and I have talked before about this problem of "Sherlocking" — that anecdote from the Apple OS back in the 90s. I think it was called Sherlock. And then someone released some innovative app — the designer under the name Watson — who created a product called Watson that put stock prices and weather widgets on your home page and it was so popular that Apple decided, well, let's just build that into the operating system in our next revision, and then Watson properly went out of business. So it was called "Sherlocking" when the underlying foundation model just decides this app maker had a really good idea — let's just absorb it into what we're doing. Is there a fear? Is there a strategy for making sure that anything that you're doing with Legal Writing Pro and its products don't get Sherlocked?

Ross Guberman
Well, I think it's a very, very valid concern and it doesn't go just to writing — it goes to all different aspects of a lawyer's craft. Number one, we've always been very big on giving people choices in editing. And that goes from my books and workshops and goes through my products. So, you know, if you write something like "the document indicates," BriefCatch is going to give you seven verbs to choose from that are all more precise and interesting than "indicates." So that's number one. And then we try to already kind of stimulate you as you're working, give you things to think about, keep you engaged as opposed to having you turn your brain off. But then the other thing we were also talking about before — the better products, the better tech, becoming more and more able to kind of track you and your own preferences and practices and actually make you more individualistic, not less. At the same time, I will say — once again, we don't want to romanticize lawyering and act like every single thing is self-expression and unique. There are a lot of things that just need to get done. They need to be written about. Documents need to be sifted through. Contract provisions need to be negotiated. And it's always the same song and dance — an NDA negotiated like 100 of them by now. It's the same rigmarole every single time. No one's going to be sad that AI takes over those tasks, as long as we keep these two values I was talking about: giving people choices and also finding ways for the tech to actually adapt to the person instead of the other way around.

Tim Kowal
I got a couple of follow-ups to wrap up. Ross, a million years ago, I attended one of your seminars in person. It was fantastic on writing. It sounds like you're busy. Are you going to continue doing teaching in person? Do you see value in that?

Ross Guberman
To be honest, I sometimes miss those days. I have a little bit of nostalgia. I'm kind of old school. I like people — I like being in the same room as people. I used to love those workshops when I was done, like before I go to the airport, mingling or during the breaks. I do miss that, but the actual answer to your question is I do a little bit still. I mean, I train all the new federal judges. I do a little bit with some of our BriefCatch large firms and some of the others. But even a lot of that is by Zoom, right? So that's good — it's more efficient. But yeah, I'll admit that I do miss the old days of getting on the plane and going into the buildings and being with real people in the room. And I decided to give up the workshop business because BriefCatch was taking over my life. But frankly, I also gave it up because I hated doing all the workshops on Zoom all day. COVID was one of those things where I made money and I think I did a decent job on Zoom, but yeah, I do miss those days. And I worry about the younger generation not getting those opportunities and not even really wanting them, right — being really glad to just do all training in 15-minute TikToks or YouTube videos. I think there was something to that. Even CLE — everyone hates CLE, but CLE used to be kind of social. Now you can get all your CLE without leaving your condo.

Tim Kowal
Yeah, well, now that I've gotten you committed to liking people and liking being in the same room, we've never met. In October, Tim and I are going to be in Boston for Clio Con, which is an annual convention about legal tech. It's just all sorts of things about legal tech. Will you come so we can meet you?

Ross Guberman
You know what, I think I actually might take you up on that. I've always wanted to go to Clio Con. I wish you would have asked me when I was in New Orleans, because I love the restaurants there. I will — if I'm free, I will go up to Boston.

Tim Kowal
Put in the reservation now. They fill up quick from what I hear. Speakers are amazing. It's not just a big sales pitch for Clio. It's a fantastic opportunity.

Ross Guberman
Clio is one of the greatest stories. I don't know them that well, but it's a fantastic story. They literally started in a garage — if you've ever seen the video where they have the pictures, it's just a wonderful story. It's a great product and it's interesting. It kind of targeted small and medium firms. So a lot of us in legal tech focus — and we're a little guilty of this sometimes — on the huge firms. But Clio did the opposite and obviously they've been incredibly successful.

Tim Kowal
Alright, I heard a yes. We'll definitely see him in Boston. Okay, alright, Ross. We'll see you at Clio Con. Thanks for talking with us. Great to see you again. BriefChat and RealityCheck — and to be candid, I hadn't been keeping up on what you've been doing the last couple of years, but I am now. I'm a continued customer of BriefCatch and I'm going to be a new customer of this new AI suite because it sounds like it's going to accomplish some things that I've been going to Byzantine and Rube Goldberg ends to try to make sure I'm not putting any hallucinated cases into my work product. RealityCheck is going to be a godsend, I think. And I think you're right that these tools that are crafted — that are leveraging AI, but doing it thoughtfully, doing it specifically for the legal market, and doing it by someone like you who is kind of straddling both worlds of deeply understanding and appreciating legal writing, but also trying to figure out how to make our writing more human using these AI tools — I think you've got just the right vision.

Ross Guberman
I love it. Thanks for the kind words and I look forward to discussing all this in Boston.

Tim Kowal
Sounds good. Look forward to Boston. Well, that'll wrap us up for this episode. If you have suggestions for future episodes, please email us at info@calpodcast.com. In our upcoming episodes, look for tips on how to lay the groundwork for an appeal when preparing for trial.

Jeff Lewis
See you next time and see you in Boston.

Ross Guberman
Thank you. Have a good week. Bye bye.