Loading summary
Transcript51 lines
- [00:01]
Steve Poor
Hi, this is Steve Poor, and you're listening to Pioneers and Pathfinders.
- [00:17]
Narrator
We're pleased to welcome back Ed Walters, the thought leader in legal tech and innovation. Since Ed last spoke with us, he's become the Chief Strategy Officer at Velix, which merged with Ed's previous company. Fastcase V Lex is a platform using AI solutions to streamline workflows and provide lawyers with greater access to knowledge and resources. About a year ago, the company launched Vincent AI, a legal intelligence platform that references real cases and materials from Velix's law library of over 1 billion legal documents. The tool is used by law firms, legal departments, and law schools around the world. In addition to his role at Velex, Ed continues to teach as an adjunct professor at Georgetown University Law center and at Cornell Law. He also writes on various topics about innovation in the legal industry. It was great for Ed to rejoin us and we talked about how Fastcase merged with Felix. The response to Vincent AI and I hopes Generative AI will enhance access to justice and improve lawyer training. Thanks, Ed, for the conversation. Thanks to all of you for listening in.
- [01:22]
Steve Poor
Ed, thanks so much for taking the time to come back.
- [01:25]
Ed Walters
Stephen, thanks for having me.
- [01:26]
Steve Poor
It's great to see you again. When we last Talked was like 18 months ago. And you've been a busy guy since then.
- [01:32]
Ed Walters
Yeah, I feel like a lot has happened in that 18 months.
- [01:35]
Steve Poor
Well, just some of the summary is Velex and Fastcase merged with each other. You rolled out Vincent AI, including a whole series of really cool enhancements. You brought out Velix Labs, along with a whole other suite of products. So you've been busy for the last 18 months.
- [01:53]
Ed Walters
Yeah. And what an 18 months has been. I mean, what a change in the industry, like in legal tech, in law firms. I think our entire outlook about what legal technology can do has changed so much over the last 18 months. It's an unprecedented time.
- [02:09]
Steve Poor
Yeah, it really seems to be, doesn't it? You started Fastcase around the turn of the century, if I recall, 20 some years ago.
- [02:16]
Ed Walters
Yeah, just in the past century. November of 1999.
- [02:21]
Steve Poor
There you go. So you span two centuries expertise. And it was. Was one of the more fascinating. Of course, you started it before there was such a thing as legal tech.
- [02:30]
Ed Walters
That's true.
- [02:31]
Steve Poor
But it's one of the more fascinating business models in this segment. You are independent, you had incremental growth, you built steadily, you were profitable. And then 18 months ago, you merged with Valex. So what was it about the growth trajectory or the market or the business strategy that that caused you to merge and why Velix?
- [02:54]
Ed Walters
Yeah, well, so I think it's safe to say that right before the merger, just about every state bar association was offering Fastcase to its members as a free benefit. And that's great. It was like a. Democratizing the law was the essential part of the mission of fastcase. And making sure that lawyers in the United States had access to the law was an important part, an important component of fulfilling that mission. There were a couple of pretty big challenges beyond that. We shouldn't pretend like that really accomplished those goals. I mean, first of all, the same problems that existed in the United States existed everywhere else in the world, in some cases in a more dramatic fashion than in the United States. Second, we also shouldn't pretend that lawyers having access to the law solves all of the problems of democratizing the law. Certainly for their clients, it makes it much easier. They're practicing in a much more competent way. But that certainly doesn't solve all the problem. And then, you know, I think it's safe to say that we were right at the dawn of the large language model revolution. And everyone at fastcase had conviction for a long time, really, that artificial intelligence was going to be a key part of the legal services industry, not just legal tech or legal research or anything else. And I think it's safe to say that we had built a competency for data collection and for legal research. We did not have a team of great AI engineers.
- [04:24]
Steve Poor
Few people did.
- [04:25]
Ed Walters
Yeah, well, I think that's right. And so we had bootstrapped fastcase. I think we'd done a good job of growing incrementally, but I think we realized we were coming into a time where we needed to think bigger than the United States. We needed to think bigger in a way about the problem more comprehensively. And we need to get good at AI really fast. So we were actually out raising money. We had decided to, after 24 years of being self funded, to go raise outside capital to address some of these issues. And in that process, we had a conversation with Velex. I often think of Velix as like a peer company. Tifaske started about the same time, about the same size, same mission. While we were going deep in the United States, they were going broad across the rest of the world. They were founded in Europe, but had pulled together the law of Europe and Latin America, the Caribbean, they were starting to move into Asia on a similar track. I think they were also in the market looking to raise money. And the idea was they wanted to expand into the US and they wanted to add some of the capabilities that the Fastcase team had of data science and product. And so we had a conversation in the middle of this process where we basically realized, look, all of the pieces that Fastcase needs, basically Velex already has and vice versa. We've known the guys for 15 or more years. We'd worked together, we licensed data back and forth. We had like a very common mission. And so it made all the sense to the world that we should build to scale. The name of the, of the company is Velix, but it wasn't an acquisition, it was a merger. Phil and I didn't sell out. We put all of our stake in Fastcase into the combined company. We're both still there, we're both still actively engaged, as is the Fastcase team. But I just want to say that the animating idea, that if we put together this expertise in data, the Velux team's extremely good kind of engineering and AI engineering in particular team, and the Fastcase team's great data acquisition team and product smarts, along with the distribution of working through state bar associations, you really could address a lot of those seemingly global challenges. And so that's sort of the idea. The idea of combining these two companies was building strength on strength. Things that they were really good at and we were really good at. And now it's pretty exciting to see it come together. Like the Velex team, a combination of the old justice team in the uk, the Velix team and the Fastcase team. I mean, there's so much talent. It's kind of incredible to see all of these amazing legal tech legends in some way under the same roof. It's an amazingly interesting experience to be a part of.
- [07:10]
Steve Poor
Yeah, I can imagine. Now you had some expertise in integrating companies with the Fastcase team. You'd acquired a number of companies. The integration, though, must have been at a different scale. When you, when you're talking about a merger of Velix and Fastcase, how did the integration go? You always had success in bringing products out. Was it as smooth as you would have hoped it to be, or did you encounter bumps along the way?
- [07:36]
Ed Walters
Yeah, I would say it was really hard. I think we expected it to be hard. You know, Velix was a remote first company and Fastcase was sort of an in office company. We're combining cultural teams of really different cultural backgrounds from around the world. And so I think all those were challenges. I mean, not to be too much of a nerd about this but just from a systems perspective, integration often means unsexy things like your HR system, your phone system, your customer relationship management system, your ERP and accounting system, when you're bringing those things together in a merger. And we've done this seven times at Fastcase before the Velix merger, but with much smaller companies. You know, in the Fastcase and Velix merger, these were global companies and they were quite sophisticated, quite large. And so I would just say that, you know, for the first, like maybe nine months to a year, the team was working really hard behind the scenes, migrating things from one finance system to another finance and reporting system, from one CRM and marketing automation system to another. And man, I mean, I didn't actually do a whole lot of that work. I wasn't super involved in it. But I have so much admiration for the group inside of Valex who did. They did work that doesn't get seen a whole lot that doesn't get celebrated. But ultimately, if the company is successful, it might be because of the great hard work and time they invested just to get the backend mechanics of both companies working in sync. They did an amazing job.
- [09:14]
Steve Poor
Yeah, that's, that's great to hear it. You make a good point. The backbone technology systems of combined companies are often overlooked in terms of the big picture and the piece, but they're fundamental to the operation of the company.
- [09:29]
Ed Walters
I'll point out one other thing, which is, you know, I've been the CEO of Fastcase for 24 years. People on my team would tell you that I am a pretty tough customer. Like, I'm hard to please. I set the bar pretty high for my teammates. Luis Faust has done an extraordinary job as the CEO of the combined business. And I would be as tough on him as I would on anybody else. He's done an incredible job of managing through change and bringing the team together. He's an amazingly effective CEO. It's really great to sort of work with him as the Chief Strategy officer, but him as the Chief Executive officer. I really admire the work that he and his brother angel, who is the Chief Technology officer, have done as leaders of this now large and scaled company.
- [10:17]
Steve Poor
What does a merger tell us about the change in the global legal market? I mean, obviously you made a decision to grow more rapidly to meet the global needs. Are there overarching messages that this merger sends to whether legal tech companies specifically or the legal legal industry generally?
- [10:34]
Ed Walters
Well, I don't know if there are lessons we can abstract from this for everybody. I'll say that at the Time that we did this, the table stakes for using the foundation models like GPT4 or Claude from Anthropic or Gemini from Google were very, very high. So it was insanely expensive to use those models to do the kind of development that we did in Vincent AI. And what that meant was that companies who wanted to do that kind of R and D needed to spend a ton of money. And you, I think you, I'm reading between the lines. I don't have, you know, special information, but I can imagine that casetext use of GPT4 was very, very expensive for a company at that size and scale. And so they either needed to raise like a lot of money or go be acquired by a company that could afford to pay what it costs to build at that scale. I don't think that's necessarily true. Going forward, the costs of those models have gone down. The cost per token now for sufficiently powerful large language models might be 1/2 of a percent of what it was even 18 months ago. I expect Moore's Law to continue to drive those costs down. So that might not be true for challengers, new companies, startups, but certainly in the spring of 2023, if you wanted to operate in that market, it was going to be very, very expensive.
- [12:00]
Steve Poor
No, I get that. Talk to us a little bit about Vincent AI because it's really a great product out there.
- [12:07]
Ed Walters
Yeah, well, thank you.
- [12:08]
Steve Poor
Tell us a little bit about sort of the strategy behind it, the development of the project and sort of what value it brings to the market.
- [12:16]
Ed Walters
Sure. If you look at the press release where we announced the merger, even in that press release 18 months ago, we said it's our thesis that over time the large language models and the foundation models would sort of become a commodity. There'd be a lot of competition, the prices would come down. It really wouldn't matter whether you're using Claude from Anthropic or Gemini from Google, you would get comparably very good results. But what would be differentiated, what would never become a commodity, is the proprietary data. And at that point, like, you know, Fastcase and Valex had put together a global database of the law, the law of more than 110 countries, more than a billion with a B, documents of structured legal data, very deep library of complaints and answers and dispositional motions from docket alarm. And so the idea was that even though you could prompt engineer your way to interesting structures of answers with any of the foundation models, the structured data itself would be a really differentiated product. And so that was the idea behind Vincent AI. To use that structured legal data to create something that nobody else could create. And so just from the start, it was multi jurisdictional, multi country from its birth. And today it works on 13 countries. It's US, UK, Ireland, EU, Spain, France, Portugal, Mexico, Colombia, Chile, Singapore, New Zealand, and Brazil, if I've got that right. I don't know if I got all 13 of them.
- [13:53]
Steve Poor
I didn't count as go through, but I'll take your word for it. It's an impressive list of countries.
- [13:59]
Ed Walters
Yeah, and we're, I mean, we roll out new countries like just about every other month, you know, every other system in the world, if you want to do research in multiple jurisdictions, you would search in the US and then you would log out. Then you would log into that product in the UK and run that search, and then, you know, log out and run that same search in Spain or, you know, wherever else. So Vincent was designed to be global right off the bat. And the idea was that we could use retrieval, augmented generation, something that is not probably new to your listeners. But the idea is instead of creating answers based on the World wide web on YouTube comments and Reddit threads and trolls on Twitter or something.
- [14:40]
Steve Poor
Hence the hallucinations.
- [14:41]
Ed Walters
Yeah, right. We can instead use authoritative law to answer the questions. And so that's, you know, that's really where Vincent started. It started with global research and drafting. I have to confess, I was a little skeptical about whether it would work. It works amazingly well. It's incredibly powerful to ask a question in natural language and to get back beautifully structured, authoritative, cited answers, to have full transparency to see where they came from. In Vincent, we show, like, the answer to the question on one side of the page, and then like a full transcript of the sources on the other side with summaries, confidence scores, direct quotes, hyperlinks so that you can open all of them. And that's really where Vincent started. I'll just say that today, you know, In October of 2024, after the autumn 24 release of Vincent, it's really now more of a platform. People think because of fastcase and Velix, that it's a legal research tool, but really now it's so much broader than that. So the workflow skills are transactional. There are contract analysis skills, there's litigation skills, like analyzing depositions, certainly research skills that are amazing, but not just like, ask a question and get back really authoritative answers. Also, like, I've received this pleading, this complaint, this motion. What comes next? How do I think about this? What sorts of questions would a good law Firm ask, help me spot the issues. And we can do that with contracts or really almost any document that a lawyer would get. And so from these sort of beginnings of research, Vincent today is like something that you would have sitting right next to you as a lawyer for almost anything that you're doing. Any drafting exercise, any research exercise, any analysis exercise. I would say anytime lawyers are working with documents, but that's like 100% of our day. It's thinking about the next steps. And then, I mean, I'll always say this. These are good first drafts. They're not last drafts. I would never cut and paste anything, even from authoritative legal AI systems and drop it into a motion or a brief. I always think that legal work is sort of like a chain, not like an event. We might replace some of the links in that chain, but the last few links of the chain always have to be human judgment. And so I fear that now I'm starting to teach my class with you, I'll just step back. But that's the big idea behind Vincent.
- [17:09]
Steve Poor
AI and what's been the receptivity of the market? And how has that changed over the last. Let's see, Vincent's been out a year.
- [17:18]
Ed Walters
Yeah, that's right, about a year.
- [17:19]
Steve Poor
I recall it being about this time last year.
- [17:22]
Ed Walters
Yeah, that's right. And this is the kind of third major Release, the autumn 24 release. I think we'll probably go on like a quarterly release cadence now. You'll see, like a Winter 24 release as well. But I think that the response has been amazing. We're really gratified that people like it as much as they do. It was the 2024 new product of the Year from the American association of Law Libraries, the first generative AI tool to win that honor. And this was a year where they had more applications, more nominations for that award than they ever have before. And out of all of. I mean, everyone's got an AI tool right now, but out of all of those AI tools and all of the people who have spent hundreds of millions of dollars building out these tools, the people who know best, law librarians, who are a pretty tough crowd, said, we think that Vincent is the best of all of them. And then in the market, I mean, people are buying it like crazy. It's law firms, yes, but also corporate legal departments, law schools, small law firms, medium sized law firms, and around the world. So at the same time that some of the biggest law firms in the world in the US and the UK are subscribing, small firms are subscribing in Spain and the U.S. in Mexico, like all around the world. It's really cool to see.
- [18:40]
Steve Poor
Yeah, that's great. I've seen the argument that sometimes people think the generative AI tools are even more important for the small firms, the solo practitioners, the not for profits as they are for big firms, although obviously they have less resources to spend on it. Do you see that same dynamic?
- [18:57]
Ed Walters
Well, I don't know if it's more important for small firms than for big firms, but one thing I will say is that it is part of the mission of Fastcase and Velex to democratize the law. If we built a tool that was only affordable and only offered to the biggest law firms in the world, I think that would be a failure. I want it to be very useful to the biggest law firms in the world, but I want to make sure that we price it and package it in a way that makes it available to small firms as well, that it levels the playing field in some important ways. So we have migrated the fast case legal research benefit that's offered through all of these states onto the Velex platform. And so now when people are logging into their Fastcase benefit through their state bar association, what they're seeing is Velex Fastcase. You can think of it as fastcase8, the successor to fastcase7, I guess, but the business model is the same. One important difference, we are including a couple of the Vincent AI skills for free in that member benefit. Things like find related documents or translate this or summarize this. The AI generated headnotes. When you're reading a judicial opinion, we're making sure that the smallest law firms in the world are getting access to those as a part of their member benefit. And we're also offering some of the more advanced skills for a subscription. And lots of small firms are subscribing to them because it's so helpful. But so I don't know if it's more important for small firms than for large firms, but it's very important to us that we're making it available to everyone at prices that makes sense for people.
- [20:26]
Steve Poor
Let's broaden that out a little bit because as you said, part of your mission has always been to democratize the law. There's obviously a huge access to justice problem in this country and around globally. How do you see the evolution of technology fitting in as part of a solution to that problem?
- [20:44]
Ed Walters
Yeah, well, so, you know, it's interesting that not knowing the law, not understanding the law, not having transparency into what the law Requires of us is such a major facet, a major part of what legal services do today, like ignorance, is the assumption, even though ignorance of the law is no excuse. Like, there is this epistemological step where we just don't know what we're supposed to do under the law that law is dealing with. And I don't know if we should be assuming that, like, what if we lived in a world of intelligence abundance instead of intelligence scarcity, where understanding what the law requires of us were more common? I don't think it would mean that we would have the end of lawyers. It would just mean that this epistemological step, like explaining to people, here's how it works, here's what law is required or requires of you might be different. You know, we might spend more time counseling people about what option they should take among several options they understand are already legal. And so my hope is that AI, if it, if it really works here, helps us to move from a kind of a information scarcity market to an information abundance market where what it means to practice law really changes. It's not. It's not as much. It'll probably be some, but it won't be as much like trying to figure out what the law is, what the law says, what the law requires, and then kind of communicating that to clients. It might be more understanding immediately or almost immediately exactly what the law is and what it requires, removing that kind of, I don't know, stage of fumbling ignorance. Or I almost understand, but I need to do more work to understand fully and moving beyond that to things like counseling and structuring and helping people to assert their rights better. So that's my hope. There's a million places where, I don't know, I have kind of concurrent anachronism. I see things today the way they're going to look in the future, and they look really stupid. Like, how much of legal practice right now is finding a similar document and going through and stripping out the pronouns and the company names and then trying to review it to see if there was a weird legal provision that was custom fit for a different matter, but doesn't fit here? That's crazy. In hindsight, it's going to look like madness. You know, the idea that law firms trained associates in the past by going through warehouses full of cardboard boxes of paper, that seems crazy to.
- [23:12]
Steve Poor
Careful. That's how I was trained.
- [23:13]
Ed Walters
No, I mean that, you know, it's crazy. It's crazy, right? You hated it. I hated it. That's how I was trained too. And in hindsight, we say, like, oh, my God, I can't believe clients paid for law firms to train their associates that way. And by the way, it's not super great training. We're doing the same thing today. The kind of research and drafting pieces of this that we have associates doing. Not great. You know, it's not really training them to become the lawyers we need them to be. They might accidentally anecdotally learn to become the lawyers we need them to be, but not because we have them doing this kind of grunt work. And so this is part of what I'm doing with my. With my teaching right now. I'm teaching at Georgetown this semester. I class called Genai and Big Law. And the idea is, how will law firms look? How will they train young lawyers? Who will pay for the training of young lawyers? What skills will they need to have in a future where AI is just in the background, where it's kind of pervasive throughout law firms and corporate legal departments?
- [24:16]
Steve Poor
Have you figured out the answers to all those questions? Because a lot of us are struggling with exactly that, particularly the training part for lawyers. You say, okay, the technology will take away the grunt work and will allow us to operate at the top of our license, applying judgment and wisdom. Okay, but people don't come out of law school with judgment and wisdom. Experience is what teaches you that. And so how do you accelerate that learning part of the learning curve, either in law school or in the practice, so that you're delivering value to your clients?
- [24:49]
Ed Walters
Well, I think that's exactly the question. I think it's extremely astute. And one of the things I want to explore in this class, we're doing it right now, tomorrow in fact, is what parts of the grunt work do we believe trains creates integrity and wisdom in associates? What parts are valuable and what parts are not valuable? Are there other ways that we could teach young lawyers the skills necessary to speak truth to a client who doesn't want to hear it, to tell someone that they're going to lose a motion, they should file it anyway to help people understand legal risk. I just think that, you know, it's. It's possible that they will pick these things up through osmosis, but osmosis shouldn't be the training model, you know, So I think there's much better ways of doing this. By the way, you know, there's new tools like Hotshot, where the idea is to train people not in boring cles, but like in very practical, like, here's a checklist. You have to work through. And here's why the things are on the list and how to make sure they're done and why we do them and what they look like. And I think lawyers are much more likely to learn these competencies, the skills, the wisdom, if we intentionally create programs and methodologies to train them in those things. And not by saying I think they'll magically pick it up because of the collegiality of our law firm. You know, they will pick it up in law school by studying torts. So I really think we have an opportunity to use AI as a lever to re examine how we train lawyers, what we train them on, what we want them to be good at, and then not to have it be a faith based initiative where we have faith that one day they'll acquire these things or the ones who do, we keep and the ones who don't, we just won't be partner. Instead to say, look, we'll intentionally figure out what things we need to train and then have a training regimen, a curriculum that is designed to teach lawyers these things and the expensive grunt work that used to be our excuse for training. If we can do it with software, we owe it to our clients to do it with software. Especially if the software is better at it.
- [26:59]
Steve Poor
Absolutely. And how do you see the generative AI technology featuring into the training itself? Because a lot of this training ought to be by senior lawyers who've seen it all, who are transferring their knowledge and their experience to the associate, not through osmosis, by looking at a box full of documents, but by a conscious effort. But I think we know that finding a senior lawyer who can do that effectively is not a common thing. So how do AI tools begin to replace the human in this learning loop?
- [27:33]
Ed Walters
Well, you know, I'm not sure that there's really a good answer for that. I don't think that AI is going to do the training, for example. But I do think that as we use AI tools to do more of the work that used to be our ad hoc training, we should take that as an imitation, an imperative to rethink how we train lawyers. If clients in the past paid for associates to do expensive, laborious, hard, often inaccurate work and that helped them become better lawyers, if clients are willing to do that, maybe that's okay. But in our future, if software is more accurate and faster and lower cost for clients, I don't think clients are going to agree to pay to train associates in the same ways. That is going to likely become an investment that law firms need to plan on. Making. And I think they'll have to hire people whose job it is not just from an HR perspective to keep people moving up the ladder, but but whose job it is to intentionally train people with the kind of judgment and experience and wisdom that we want from them. I don't think it looks like anything in law firms today. I think it's really going to be different. I don't think AI is going to do it. I don't think we should trust AI to do it. I don't think AI would even be very good at it. But AI might be the catalyst that makes us think more intentionally about it.
- [28:53]
Steve Poor
We're at an interesting moment, aren't we?
- [28:55]
Ed Walters
We really are.
- [28:56]
Steve Poor
You made the observation early on about how the world has changed. It's a great time to be in the profession.
- [29:02]
Ed Walters
I really agree. There are people who say, like, I wouldn't send my kid to law school right now, but I have to tell you, like, I've got 90 students at Georgetown Law right now who are very enthusiastic, who are fired up about being lawyers in the next five to 10 years. And I love that enthusiasm. So I'm very hopeful about what the next five to 10 years are going to bring for the legal profession.
- [29:26]
Steve Poor
As am I. And you guys are leading the way. Ed, thanks for making time for us today. I appreciate it.
- [29:31]
Ed Walters
Thank you for having me, Steven. I really appreciate that as well.
- [29:35]
Steve Poor
Thanks for listening to Pioneers and Pathfinders. Be sure to visit thepioneerpodcast.com for show notes and more episodes. And don't forget to subscribe to our podcast on your favorite platform.