Lois for Word: Filevine’s Embedded AI Drafting Vision (Full Transcript)

[00:00:00] Speaker 1: So welcome everybody to Filevine's latest webinar, Lois for Word, formerly Pinsights, for those of you that are Pinsights users. What we will be doing today is we're going to talk a little bit about our perspective on AI and legal, that will come from our CEO, the acquisition of Pinsights and how that works into the greater Filevine story. We're going to have Airslate's perspective, so a customer of Pinsights, Misha is here to share about his pre-AI journey, as well as his perspective post-Pinsights, or Lois for Word. And we are also going to show some true real-life use cases on how this is working, how Lois is working for firms and legal professionals today. So that will be both corporate use cases as well as litigation. We'll then open it up for Q&A, as we know many people have a ton of questions about how this might work for them. And then lastly, for those of you that joined for our giveaway on how do we evaluate AI, or are ready to meet with our team one-on-one and figure out how this might work for you and your legal team, we'll get that set up at the end here. So with that, I would love to introduce our speakers. We have our CEO, Ryan Anderson, will be here talking about the vision and how Filevine and Pinsights are coming together in the future of Lois for Word. We have Sona, who is the CEO of Pinsights, the recently acquired company, as well as the VP of AI drafting here at Filevine. She has come on to continue the vision and make sure that we are creating a great path forward for this AI product and so much more. I've heard so many ideas that are coming out of this acquisition as well. We have Misha, a senior legal operations manager of Airslate. He will be here telling his side of the story, how this really works in real life. You will also notice we have quite a few subject matter experts in the Q&A. So as we are presenting, feel free to send as many questions as you want through. They will be helping answer those questions as we go. We have found on webinars like these that people have very specific questions that they want help with. So feel free, send those through as we go, and it will help you kind of stay on your journey of learning how Lois for Word could work in your firm or legal team. So there are quite a few folks that have not been Filevine customers before that are joining us on this webinar. So we wanted to just quickly frame up who is Filevine. Filevine is a legal AI company that's been around since 2015. We have over 6,000 customers, and this is across enterprises, firms, government agencies, and legal teams everywhere. And so this is a great mix. And so what we will do within these use cases is try and show an overview within the litigation space as well as the corporate space to make sure there's something for everybody on this webinar. We're headquarters in Salt Lake City, so great mountainous city, but also have offices in Chicago, New York, San Francisco, Prague, and Bratislava, and we'll continue to grow and expand. But it's been incredibly exciting to get a footprint across globally to make sure that we are finding the best AI products, building them, and giving them out to our customers and audiences. Our team is growing, awards are growing, and we're just really excited to tell you about kind of the vision we have here for AI and where we're headed. So with that, you don't need to hear from me anymore. We have our CEO Ryan Anderson here who's going to kick this off.

[00:03:12] Speaker 2: Thanks so much, Keegan, and absolutely thrilled to be with all of you today, especially those who haven't joined us in the past. If you're running a corporate legal team or a government legal team, and maybe you're not a Filevine customer today, but you're particularly interested in contract lifecycle management or redlining via AI tools, thrilled to have you. To our amazing customers who have been with us for a long time, we can't wait to show you what we have, what has been built by Sona and her incredible team, just a beautiful, fast, slick product that I think is going to make an immediate impact in your workflow. So this is going to be a fun day for all of us. So just a couple of things, for those of you who have heard me speak before, and the vast majority probably haven't, but for a few of you, this might be a repeat. We feel very strongly that AI should not, and I don't believe will, replace legal professionals. I do think that the legal judgment will be dramatically impacted on a day-to-day basis as AI takes over more mundane workflows and also gives you more optionality and helps you ideate. The best lawyers I have found in my career are the most creative lawyers. I don't care what the endeavor is. Lawyers get a bad rap for not being creative. So you might be in transactional, you might be in corporate or government, you might be a litigator. The best lawyers are finding creative solutions to cut client problems, to get them the result that they want. I don't think AI will ever be as creative as you, but it can be a thought partner and help you achieve results that you've never achieved before. So we believe that AI will empower and amplify you. It will work with you to bring better results to your clients. And I think that is one of the most important parts. We actually think like, sure, AI might make you faster, but I think there's far too much emphasis on time savings as a result of AI. We want AI that literally makes outcomes better for you and the customers you serve. So with that being said, I just kind of want to give you a little bit of feedback into how we think about AI as a market. Many of you have probably heard of Jevon's paradox. For those of you who haven't, it is the notion that it's a principle that the effect of increased demand predominates and the improved efficiency results in a faster rate of resource use. That is a very long winded way of saying that when the price of something goes down, we actually wind up using more of it. And let me tell you why this is so important for our industry. There are a lot of folks in the sound of my voice who might be saying, I'm a little worried AI is going to take over part of my job or part of my junior associate's job or part of my paralegal's job, and I worry that maybe there won't be enough legal work for me. But what we have seen over and over again is that the opposite is true. People thought that people would take less drives in cars when Uber came along and that people would own less vehicles. That's actually not true. People just took many more drives than they did before. They were more mobile than ever before. People who couldn't maybe get in a car now had access to a car. That's what we're seeing in legal. Lawyers are actually just doing much more in their day. And let me tell you why that's so exciting for me. As somebody who knows your pain, who used to practice law, I can tell you, my word, increasing the throughput of a lawyer so you can help more clients means A, your impact is greater and B, let's be candid. It means you can probably make more money with less time, which is an incredible place to be as a lawyer. And we're seeing this play out in the ecosystem. If you look at sort of, we'll call it the zeitgeist on Twitter, here we have Adam Osmani, who's the director of Google Cloud AI. And he wrote the following just recently on Twitter, on X. Every time we have made it easier to write software, we've ended up writing exponentially more of it. When high-level languages were placed. I won't get into the rest of the quote, but the bottom line is instead of saying, Hey, there's a set amount of code in the world. And if Claude or Anthropic is going to help us write code, well, there's going to be less work for engineers that hasn't ended up happening. Instead of course, engineers have simply written more code. So I have some good and bad news for you. You're actually going to be busier. There's actually going to be way more work for you because there are so much going on in the world of legal AI. You're going to actually have a lot more inbound and things to do than ever before, but you're going to need AI tools to handle it all. It is the golden age of being a lawyer, but that also means you're going to need to use golden age tools to win. So given that exciting future, we actually see it in the data. Litigation demand is up 5% year over year. That doesn't happen very often, but it is up 5% year over year. Litigation, labor and employment grew alongside transactional work. That is very rare. Historically, we are seeing something different than we have seen in years past. And again, corporate M&A real estate all grew 4% to 7% year over year. Q3 was the fourth highest demand for legal services in corporate in 20 years. We are seeing a sea change in the amount of legal work you all have to deal with. And that is in part because of AI, in part because lawyers are becoming more efficient and taking on more tasks. And so that means we live in a world that gets very exciting, but more will be required of you. So what does that mean? It means that AI is not a nice to have. Unfortunately, it must be part and parcel of your daily work. If you've heard me speak about this before, I like to talk about the fact that maybe a year or two ago, it was like, okay, I'm going to do my legal work and then I'll do some AI stuff or AI assisted legal work. And we saw that reflected in the Filevine product. We had Filevine and then we sort of had some AI tools. What you will see in Filevine and Lois is that AI will be assumed throughout the product. Everything you do will likely have an AI component. You won't choose to use an AI feature or a non-AI feature. We will embed naturally, natively, AI where it makes sense throughout the product. Sometimes you won't even notice it. You'll forget you're even using it because it'll be just part of the water you swim in as a legal tech worker. So Lois is legal intelligence embedded and not bolted on. For those of you who don't know, Lois stands for legal operating intelligence system. We believe that Filevine has the tool set to build the best legal operating system in the world. And because of AI, it can be an intelligent legal operating system. So a little bit about why we decided on Penn Sites. For a long time, we had heard from customers, particularly in corporate, but throughout government and throughout our law departments or our law firms that we would sell to, we would like to have a better experience in Word, particularly with redlining and negotiating, but throughout the drafting experience in Word. And let me tell you, we looked at everybody. We looked at all of Sona's competitors. I made my team try all of these competitors' softwares. We bought licenses to them all. If you look at some of our paralegals, they've got all these little apps on their Word, all these little add-ins for the other competitors I made them buy and try out because we wanted to buy the best product. At Filevine, if we've made an acquisition, you can be sure of one thing. We have bought what we believe is the very best product in the industry with the most highly technical and excellent elite team of engineers and product people we can possibly find. And that is exactly what we saw in Sona and Merriam, the founders of Penn Sites. They're rigorous in their thinking about design. You will see that in the product. It is beautifully designed. It is incredibly fast, already trusted by some of the best companies in the world, not only Misha and Airslate, who we are so honored to have today, but also groups like Ramp, Redis, Glean, some of the best and AI-first companies in the world already use Penn Sites and rely on it every day. And of course, to make that happen, you have to have a quality-first engineering discipline to bring a product like Penn Sites about. We can't wait to show it to you. It's going to be awesome. So let me just tell you a little bit about the vision. Lois for Word unlocks a drafting experience that we think is totally different. If you are going to draft a document, the things that you need to bring to bear are the facts of the case, the case law, templates that you already use, and your creativity. Lois helps you with all of those. It's going to pull the facts of your case over from Filevine. It's eventually going to pull case law in. That's a big initiative at Filevine. You will see that come over the course of this year. And of course, with FBDA, you can actually pull, and Penn Sites, you can pull the actual templates you need to start the drafting experience. And then Lois is going to help you ideate and give your documents the most creative, thoughtful solutions, well-written, and perfectly stylized, so you can be confident you will have awesome documents that leave your office. When we think about legal work, when we think about sort of what it means to do a legal job, what I have found is that it is almost always the case that legal work is looked at as a lot of different things you do throughout your day. So whether you're on the phone with a client, or you are talking to a judge, or you're with an opposing attorney, all of that work that you're doing is encapsulated eventually in a document. You think about, like, if you're managing a case for the litigators on this call, you're having conversations with your paralegal, you're having conversations with your other attorneys, you're receiving inbound from an opposing lawyer, or maybe an insurance adjuster, you have to deal with court documents, you have to deal with expert witnesses, and all of that information is probably going to make its way into draft interrogatories, a motion for summary judgment, or an opposition to a motion for summary judgment. Almost all of our work winds up with a document being the defining artifact of the work we do. And so Lois for Word, we felt, was almost the capstone product for all that Filevine has ever been. There's all these things we do so well to help you categorize, catalog, and understand your data, but none of it makes sense if we don't provide the best drafting experience in the world, and that's what Lois for Word will do for us. All right, so having said all that, I cannot wait to introduce you to the founder and CEO of Pinsights, Sona Sulakhyan. Thank you so much, Sona, for joining us today, and we're thrilled to announce and show you Pinsights.

[00:13:58] Speaker 3: Thanks, Ryan. We are also super excited about this. As Ryan mentioned, the big reason we wanted to partner together is for that vision of legal AI embedded where lawyers already work. I think the biggest takeaway today I would love everyone to leave with is most of us who are using AI, and can you remind me what that number was at the beginning of the poll?

[00:14:21] Speaker 1: Oh, I'll have to tell you in a second here. Let me pull that poll back up.

[00:14:24] Speaker 3: There's a lot of you that are already using AI. You're probably using it in some way that's like, give me, tell me what I should do, and then you go and do it. Having AI in Word turns that around, so now you can also ask AI to help you issue spot, help you figure out what you need to do, but also actually execute with AI. So that's the big takeaway I want everyone to leave today of how can you take AI and go from just asking what you should do to asking and guiding it to what you need to do to get that final result out. And because we're embedded in all of the file-bind data and information that Ryan covered, there's so much information that we can use. There's a ton of information we already know about you when you get into the tool, so you'll see not a lot of the prompting and things that I do today is very heavy. Your goal will be more around how do I become more of a context engineer? How do I architect the right piece of information to get the results that I need? And we'll take that burden off of you of how to use AI and all of that fun stuff. But I'll turn to Misha also. I want him to introduce himself maybe as well. He has been working with us. I'll give a quick overview. Essentially, I cold emailed Misha. He was gracious enough to hop on a demo call, which is also why you should always take those demo calls in response to cold emails once in a while. That's blossomed into an amazing friendship. He is so smart, an amazing technologist. He's been a customer of ours for about two years now, maybe going a little bit longer than that. We are really grateful to work with him and his team, but I'll let him give a quick intro of himself as well, too.

[00:16:01] Speaker 4: Hi, everyone. My title says I'm a senior legal operations manager, but in practice, it basically means that I'm responsible for essentially all of the modes and all of the practice areas in which only four of our team members for over 800 headcount in the company work, supports. That includes virtually everything and the documents and any kind of reviews are part of it. So, yeah, that's it.

[00:16:35] Speaker 3: Yeah, and to hammer that home, he is basically organizing a team of four and has orchestrated the whole legal department to manage a unicorn startup that is growing at an incredibly fast pace. So, he's done a lot of work for us. He's managed a unicorn startup that is growing at an incredibly fast pace. So, he's done a very good job at everything he does.

[00:16:51] Speaker 4: So, I only want to add from scratch, there was no legal operations before that. Thanks.

[00:17:03] Speaker 3: So, Misha, I'm just going to ask a couple questions. I think it'd be helpful for the audience to understand, I guess, your approach to evaluating AI, figuring out what you need, what that kind of looked like. But I guess if we're just mapping out how folks did contract review specifically before AI came around and you guys were very early adopter there, but if you could just paint a picture of what contract review looked like pre-AI and the process you were working with.

[00:17:30] Speaker 4: Yeah, you're very right. We were very early adopters of AI as such. I would say company-wide, we were basically not waiting for executive teams to tell us that we should start thinking about AI. We were the first team in the company to actually start using AI for the actual work and outsourcing the part of the work to AI responsibly, of course. Our contract review process looked basically like we just knew all of the business rationals and kept them in our head, just as simple as that. And because most of the lawyers in our team used to work with the word primarily, they've been using all of it functionality. And that's why our very first focus on deploying any kind of AI was about the contract review. But in summary, it was just all individual review by each console consulting with the business with no additional tools except for word.

[00:18:47] Speaker 3: Gotcha. I think that is sometimes the challenge as well. I think everyone thinks that they have a process, but in legal, it isn't. It's usually just a massive data you've accumulated over your 10, 20, 30 years of expertise, working with a lot of different businesses, understanding what that judgment looks like in different scenarios and what the right language and templates you want to pull from for that is. So definitely understand. I think that is also one of the big transitions with AI is how do you take the gut feeling that we have as lawyers and turn that into a repeatable process that something like AI or some kind of computer program can automate consistently to a high level of accuracy. So that's really helpful. I guess, how did you guys decide that you needed to change something and what were the success criteria that you created to evaluate different solutions?

[00:19:41] Speaker 4: The primary reason was that a couple of months before we actually onboarded Insights at the time, we were working on the vendor contracts playbook with outside console to kind of absorb the wider expertise on all the possible scenarios for the vendor contracts because we essentially decided that we will not be playing the game of pushing vendors to adopt our paper for their services because this is some sort of illogical from the business perspective. So we decided to at least work internally with the legal team in order to compile a rules book for what we allow or what we try to avoid in a vendor contract. And that was one of the worst experiences because we basically came up with nearly 100 pages PDF and the contract review process in detail looked like you open the new contract in Word, then you just simultaneously open that vendor contracts review playbook and then you browse through that playbook to find the applicable clause that you are currently reviewing. Then you try to think about whether this completely works for this agreement, regardless of how small it is. And then you compare it to the tiers to which this full fallback language applies to, which is not. And then you might want to double check with business and then you want to double check with a general console, et cetera, et cetera. So as you can imagine, it was just not so comfortable in terms of the actual process. So we kind of, once we understood what AI can do in that realm of contracts review or virtually any document review, we realized that we need a boost and we need to look for something that would not make a human spend all of this time of tracking, navigating the applicability of certain fallback languages or even finding the appropriate rule for the tier. And I would say, I would call it the main reason, which boils down to the conclusion that even though you kind of came up with and agreed upon all of the rules for different vendor contracts type, it was still a very time consuming process, even just for the one agreement, regardless of its value to actually review it and be confident that you did your best in preventing any sort of risks or satisfying the business.

[00:22:47] Speaker 1: Yeah.

[00:22:47] Speaker 3: I think one thing we can agree on as attorneys is we don't have unlimited time. So anything we do is time boxed and how you spend that time matters a lot to the quality of work product that you get. I think what you mentioned as well, Misha's talking in playbooks for in-house, that makes a ton of sense. I used to be a transactional attorney at a law firm. We had checklists and we had clause libraries I created for all the different deal types that I did. And I knew that I liked, and I could find it. It's the same idea where you're trying to create a process and something that you do over and over again. And that creation process can be really difficult. And for Misha's team, that was made sense given they're a very small team to try to outsource that. A big problem with outsourcing such workflow creation ends up being just very difficult for someone to manage. If you're jumping into a giant workflow, a document someone else has built, it's very difficult to understand what's in it. How do you modify it? How do you apply it to different scenarios? So part of the things we'll show you today on playbook creation is all the different AI features you can use when you're in your workflow or when you're thinking about building that workflow, where you can interact with it and just keep that ownership over your own workflows to yourself. Before we jump into the demo, and before one last question for Misha.

[00:24:09] Speaker 1: I'm just excited to see what you got.

[00:24:13] Speaker 3: I guess for the evaluation process, how did you go about filtering all the different tools out there? And for reference, some deals that we're in right now have valued up to 40 different tools. The market has so much noise that it's difficult to understand like, why is this one different? Why is it better? What do I get out of this? So just would love, I guess, Misha's take on how he went about it.

[00:24:41] Speaker 4: Yeah, this is a very specific question because we are also a SaaS company and part of it, we could like anything that is basically related to the document generation itself, like when you need to automate MNDAs generations per specific country or any other criterion, we can cover it for ourselves. And our issue when comparing to among, I'm thinking up to 10 other vendors, was that we didn't look for a vast functionality where that contract review would be included. We didn't look for a CLM. We didn't look for document generation. We just looked for something that would do that, what we had the gap in, which was the contract review powered by AI, which we could control, meaning set up our own rules where the data will be processed securely and independently from any other customers, obviously, which is now, I do believe, the market standard when it comes to AI. But yeah, our process for the vetting of vendors looked like we were very narrowed to the AI powered functionality specific for the document review. I would not even say only contracts because for us, adoption of insights resulted in changing the mode of any and more types of documents because that empowered us to much, much more expand beyond just contracts.

[00:26:32] Speaker 3: Thanks, Misha. I think with that, I can jump into maybe the long-awaited demo here.

[00:26:38] Speaker 1: I will give you screen sharing, Sona.

[00:26:41] Speaker 3: Awesome. Thank you. I will do two use cases today. We'll review, see if this is the right. I think everyone can see a VA. Cool. We're going to review a corporate use case, so marking up a SAS contract that comes fairly often, and then we're going to switch over to more of a litigation-focused use case and how do you pull from documents in Filevine? How do you talk to them and use context about you to draft new parts of, in this case, we'll do in interrogatories. So, starting with VAs, and I'll just note, I like doing VAs over an NDA or an MSA or another contract type because they are standardized in a way where you do get VAs fairly often. There's a lot of third-party paper that comes in for this contract type. Everyone has a different approach to VAs. Some customers that we work with will mark it up, just bloody the hell out of it. Other folks will be super light, almost treated similar to an NDA when it comes in. The other complexity on top of it is you have to comply with all these other laws and regulations that are out there. There's HIPAA, there's HITECH, there's template model VAs that you can reference from the HHS. A lot of other documents are the same where you can reference those laws and regulations, the models that are put out there, and then customize to your specific use case and approach. So, here, we'll open up Insights on the right-hand side, and I'll make this a little bit bigger to give us some more real estate to work with. And you'll see two flows up here. We've got Playbooks, which are the workflows, and we'll come to in a second. But we also have Composer, which I think of as a very powerful agentic chat that does a lot of fancy stuff on the back end when you ask questions. We'll come back to this, but you can do things like research, you can talk to your playbook and knowledge base that you set up, you can draft things, you can redline things, you can generate comments for other things that you've marked up. And all of that is done with the full context of the document that you're looking at. So, we'll come back to this, but starting on Playbooks, we do have a ton that's out of the box. So, if you're reviewing things even as simple as a hotel agreement, that comes up often, but I don't know anyone who's an expert in hotel agreements. We've got you covered there. Feel free to use one of the ones that are out of the box. And from here, let's run our VendorCo BAA Playbook and see what that looks like. Now, when I run it, I can always give more context if I want. I'll often mention things like, what is my leverage here? Maybe I have high leverage. So, feel free to mark up whatever you want. Or I'll mention I have low leverage. I will note sometimes you want to do both. So, if you are in a rush, low leverage is nice and have it mark up very small things. Sometimes folks do want to get every single suggestion possible on a document just to understand where they could have missed an issue, where they could have marked it up. So, depending on however you go about it, you can customize that. Over here, for the Playbooks on the back end, I think of Playbooks as a series of prompts that you set up at a very high level. It'll talk about what you like to redline, clauses that you like to use, how you want to redline it, and then how you want to talk about those redlines when they come about. Behind that Playbook, we want to keep those Playbooks as light as possible. I think of it as principle-based prompting. And as we work as lawyers, we tend to think in concepts more than just actions that you need to do in different scenarios. So, we've added knowledge bases behind it. So, you can upload redlines you've done before on BAAs. You can upload your template. You can upload other reference documents, which might be your checklist, your clause library, your Playbook you're using today. And Misha, that's a new feature. So, if you haven't done that yet, you can upload that 100-page Playbook to the Playbook and have it think about it. And then from here, if we filter this down, you can see we have a nice little checklist we can work through. The AI will risk score all the issues in that checklist for you. You can hard code that and then use that to get a good sense of what escalations need to look like. But we can table that for another conversation. And then staying in the document, let's just scroll through this and click through the document and see where we can improve this doc. So, here's one issue. Just clicked into these purple highlights. Everything we do in Pinsights will be explained and sourced to where that language came from. So, on the Playbook side of things, you can see there's a nice explanation up top. That is embedded. We have verified these sources actually exist in the document. If you uploaded other documents to chat and you were using that, we will source that as well. But that really helps reduce hallucinations. And I know that's often a concern for folks, especially on the litigation side of things. Below that, we have the suggested fix as well as a comment to go along with that fix. I will note, everything is customizable. You can change the tone of your comment. There's a section you can do that in. You can say what your comment is. And as we make changes to what the AI suggested, you can see the AI will learn from it. And this is where the value of having a workflow comes from. You can get that checklist. But our goal is to learn from everything you do in the tool. And I'll show you in a second. But we'll also try to learn from everything you do outside the tool. So, we keep tracking those trends and give you suggestions to the Playbook and keep it as kind of like a living, breathing document going forward. And then from our perspective, nobody likes creating workflows or maintaining them. So, we try to get AI to do as much of that as possible. And over time, we keep adding those features. So, there's stuff here. If you're like, hey, I don't like this suggestion, fix my Playbook for it. You can quickly do that. As you use learn from button, you'll get suggestions to the Playbook over time in the web application. And then there's other features around, hey, I like this red line. Can you generate a rule off of it in one click? That's the Playbook side of things.

[00:32:38] Speaker 2: I'm going to take- I just interrupt just real quick. This sounds so silly and simple. But if you've looked at other products, they often are not nearly as slick on commenting. As you note here, the comment is actually coming from Sona. So, when you send red lines over, it looks like it's really the team, your legal team who wrote it. And of course, they did indeed approve it. But it just feels so much more personal and human.

[00:33:01] Speaker 3: I will also mention everything we do, I don't get this question as often as I should, but you want AI-generated stuff to come from you to maintain attorney-client privilege and work product. So, everything we do, whether it's- there's a new email integration as well. So, if you wanted to, hey, I've created an NDA Playbook. I never want to look at those documents again. Can I just email it and have it magically come back redlined to me as if an associate did it? You can do that. But again, all of those will feel like it's coming from the user and has the human built into it. So, when we're building AI, that's always a consideration for me, is we want to keep it still attorney, work product, whatever we do. And the attribution is really key to that.

[00:33:43] Speaker 4: Cool.

[00:33:44] Speaker 3: That's Playbooks. From there, anything you guys think I should add? I'm going to move to Composer otherwise. Awesome. Let's go to Composer and try to do a couple things here. I'll keep it simple. We can maybe, let's say, let's highlight this. As I highlight the clause here, you can see now the whole clause is in scope. Even the track changes and the comments are in scope here as well. So, if you are trying to respond to comments that are already in the document, you can just highlight and do that in one click if you want to respond to redlines. And that's throughout the tool. We take the whole history of the document. So, we get the question of does AI know the redlines came from me? I received a second pass. Does it know who did which redlines in the document? The short answer is yes. And now we can say something like revise this clause to be more favorable to the vendor. And I'll let it do some thinking here and give me a nice little response. As we're waiting for this to load, I'll note that we do have a giant prompt library here as well. So, just like we have playbooks built out of the box, there's a ton of prompts. So, if you want to learn how to write really good prompts, what formats work really well, you can always come to this library, select the one that you want, customize it to your use case, and use it. But here we can see revise the clause for me, give me an explanation again, made some nice light changes. I can generate a comment to go along with that. And again, that comment will use the tone that I've either described or that it's learned that I like over time. And then if we're happy with this, we just drop that in. One click goes into the document. We can also continue to iterate with this redline. So, if we wanted to keep going with it, we can say, hey, chat with the redline, make it lighter, make it heavier, add a carve out. That's the beauty of AI. You can just talk to it and have it do what you want. And we'll see in a second, once you pull in more of your documentation and context, it should do an even better job at that. Awesome. Let's try one more prompt here. I'm going to take something from the prompt library in this case. Let's do BAA reviews. Let's assume we had no playbook for this document. It is the first time I'm seeing it. I've never negotiated in a BAA before, but I wanted to take a first pass issue spotting and then marking up this contract with things that I could mark up. Here's a prompt that I use fairly often. So, I'm just going to ask the AI to revise this BAA to align with pharma market standards, ensure compliance with HIPAA regulations. I'm giving it some information on how to prioritize and what my leverage looks like. And if all this feels like super overwhelming, there's a little fancy prompt enhance button down here. You can ask it, write a very simple prompt, ask it to inject some more context, and then it will help you prioritize and focus that review a bit more. I almost think of it as each prompt you ask the AI has X amount of brain power to do a task for you. So, you want to maximize the brain power that's used on different things. And then add qualifiers. And in this case, I'm going to ask it to look at the whole playbook that we have and all the knowledge in there to be a little bit better. Actually, let's take that out. So, let's say it doesn't have context of the playbook. And let's add a review step here. So, the review step is going to ask it to plan first, essentially, and then execute. I usually, whatever I do, try to add a planning step in all my prompts so that, just like a human, I ask it to think first and then do what it needs to do. And you can see here, whenever you ask it very loaded prompts like this, it will break it down into a nice little checklist for you and then execute them one at a time. Again, I think of it as AI has X amount of brain power for each task. So, breaking that up will get you better results for every single task. It also means you don't have to become a very expert professional prompt engineer when you're using this tool. Just say what you want. We'll break it down. We'll add the right context. And then you can just focus on the format of the response that you want to get. Awesome. So, here we've got the output. Here's the explanation. Everything will be sourced. If it went to the web, you'll get a little note saying, hey, this is a web source. Click on it. It'll take you to the link. And we'll let it do some more thinking here. Great. It's given me some edits I should probably make. And we'll let it finish some thinking before it gives me some red lines. Take a little bit more time here. That's a lot of reasoning. Awesome. Cool. So, we've got our red lines here. Here's the explanation. The red lines should be fairly surgical in the changes that it makes. And it'll use the right defined terms that it pulls out through the contract. So, whatever language you're using. And if you're drafting new stuff, it'll use the format, style, and structure of the rest of the documents. That should be all consistent as you go through.

[00:38:55] Speaker 1: Cool.

[00:38:56] Speaker 3: Well, that's our corporate use case. From this, I'm going to shift over and we can maybe draft a couple interrogatories and see what that looks like. I will note I was a transactional attorney. So, I am very good at that. But I'm going to defer to Ryan to add color commentary.

[00:39:12] Speaker 2: I would be more than happy. Do you want me to do so now or do you want to tee it up first?

[00:39:17] Speaker 3: I can tee it up first.

[00:39:19] Speaker 2: Okay. Cool.

[00:39:21] Speaker 3: Or actually, do you want to add it now?

[00:39:23] Speaker 2: Yeah, yeah, yeah. So, here we're looking at basically a typical interrogatory document. So, this has a caption. It has a first set of sort of like I think it's like seven or eight interrogatories that we've drafted. And now we want to say, okay, you've got this set of interrogatories. Now we want you to go back into Filevine and look at data inside of Filevine and help us draft more interrogatories specific to my particular case. So, that's what we're going to show here. Obviously, like this is the big vision here, folks, is to take data and documents that we have kind of extracted out via an agentic AI layer and allow you to draft interrogatories that are very specific to your case. Of course, we'll do the exact same thing for answering interrogatories. We're not going to show that today. But this is a sneak peek. It's not ready like tomorrow, but it is coming very soon. And we've got a live demo to show you today. I think you're going to love it. But here we're going to be able to ask more interrogatories. You can see there's like, what, there's maybe 10 that we've asked. So, we say, okay, look, we can add up to 10 or 20 interrogatories here. So, Sona is going to ask Pinsights to look into the case that we've selected, which looks like it's Owen Klein versus Lion Airlines. Again, that's like a connection to Filevine. And ask it to produce interrogatories based on what Lois knows about this Filevine case.

[00:40:50] Speaker 3: Awesome. Thanks, Ryan. Yes. So, you can select the Filevine project. All your projects will be listed here. You can select the one that you want. Right now, we've got Owen Klein queued up. And then from here, let's maybe ask it to draft some more interrogatories. So, I'm just going to say, based on this document and the information in our knowledge base, can you draft further interrogatories that are relevant to my case and which help prove the defendant is liable for plaintiff's injuries?

[00:41:18] Speaker 1: Cool.

[00:41:24] Speaker 3: Let's hit run and see what happens. And this is going to probably kick off our drafting flow. So, we have a lot of different tools on the back end. You can kind of see a few of them listed here. But we try to do everything hidden from you. So, your main goal is to prompt what you want. And then our goal is to figure out what tools need to be put in place to get you the results that you want. And I'll give it a couple more seconds here to run this. But it should pull the format and structure of the other interrogatories when it's drafting this. It should use the right words. So, in this case, you, plaintiff, defendant, as the defined terms when it's drafting. And we can see it's not done yet. So, it's doing a little bit more thinking. It should give us a nice little draft. Awesome. So, anytime it gives you a draft, sometimes it might ask for clarifying questions. I just had this happen earlier today. I was demoing a different document. And I was asking for a DPA to add late payment penalties. If that happens, it'll say, can you tell me some more information? Because that doesn't make any sense. So, it'll ask you some follow-up questions. Here, it'll give you a summary of what it's done. If there's anything it's unsure about or it thinks it could do better, it will ask you follow-up questions that you can answer and iterate basically on whatever draft it gives you to get to that best final result. And again, here, you can see it's using the right terms. U looks great. The start of it as state also kind of aligns with how the rest of the document is also set up and how the other interrogatories are set up. And then we can put our cursor.

[00:42:57] Speaker 2: It's correctly numbered, too, which is kind of nice, right? You can just pop it right in.

[00:43:00] Speaker 3: Yep. And then you can set your cursor wherever you want in the document. And then let's just drop it right in. Awesome. And then we've got those interrogatories already dropped in place.

[00:43:12] Speaker 2: Amazing. I have to imagine this would save a lot of our customers a lot of time. There's a lot of questions we're seeing in the QA. But yeah, this is super exciting. Thank you so much for showing this. This is just a quick example. You could do this with any type of document, though. I think earlier on, we had a question about like requests for admissions or reviewing an answer to a complaint or taking a complaint and drafting an answer. It would work with all those kinds of documents. The beauty of Pin Sites and what we now are calling Lois for Word is that you have access to all that file line data that's sitting right there. And you can just pull it into your document and draft bespoke documents right in Word without ever having to leave Word directly here. I think you all are going to love it. I will tell you, I have used this feature a ton. I have played around with Lois now for, I don't know, quite a bit. I am sort of blown away by how good it helps me with reviews. Another couple quick examples of kind of good things to do if you're a litigator, if your team sends you draft complaints for review and you say, OK, I want to make sure we've lined up every defendant with the appropriate allegations, that is such an easy playbook to create in Pin Sites, a really easy one. You can be so much more confident. Your complaints have all the necessary language. Confidential settlement releases. So if there's a settlement release that you want to review, I have a suspicion that all the attorneys on here who do that kind of work are really sick of drafting those form settlement releases, but you all know you need to read them and revise them and redline them, especially, I mean, pretty, you should do it for all cases, but particularly for big cases, Pin Sites is going to be awesome for that. It will make sure things like confidentiality and identification when the check is set to arrive, date of receipt of funds, all of that can be put into playbooks. So you make sure not to miss anything. So for those of you who are in a reviewer's position, I think you will find Lois for Word to be particularly powerful. Anyway, with that said, we are so glad to give you a brief demo of Pin Sites. There's so much more to show you in this product, by the way. Sona only showed you a very sneak peek, but she showed you maybe 10, 20% of the product. So lots to show later on. And we have 10 questions in the Q&A, so there's a ton to answer here. With that, do you want to pass it back to Keegan?

[00:45:34] Speaker 1: Yep. And actually, I think Sona is going to handle this Q&A because we'd love to now hear Misha's experience as the last, what did you say, about two, maybe a little over two years, Misha, and what was Pin Sites, now is Lois for Word, how that is impacting your team. So Sona, I'll let you take it from here.

[00:45:53] Speaker 3: Yeah, I'll turn it over to Misha. I don't know if you have thoughts of working with us. Has it been good? I hope so.

[00:46:01] Speaker 4: But yeah. It was virtually nothing negative, and I'm not even shy to say that because I am just personally very, how to say it, not in a good way. I'm very demandful of the services that I deploy for my team because I some sort of feel personally responsible for what and how it is convenient for my team to work, do their work. How easy is that? Because that's essentially my work. And I do it as well. So only the best experience and no regrets whatsoever. One of the key highlights that I probably would like to share is that a very, I would say, common example of MNDAs because that's still the most volume of documents that we receive, thank God, and especially customer MNDAs, which is always great, right? But before Pinsights, our practice was that we some sort of had a policy that we would try to train our account executives to utilize our own one-page MNDA at all times so that we do not have to review our customers or prospects MNDAs because they could be very tricky, and I bet everyone on this call will know that. But with Pinsights, we realized that the time savings was just that huge that not only allowed us to implement playbooks on Pinsights for virtually all the documents we work with, like our partner agreements, our different and various types of vendor contracts agreements, and BAIs and DBAs and DORA addendums and whatnot. So it really, the type of document did not matter once we figure out that Pinsights is changing the mode we are looking at building the process around any document, as well as one fancy thing that happened with MNDAs was that we basically stopped demanding that sales use only our template because we became enabled to review more. We just told them that it doesn't matter for us anymore whether the customer offers to us to review their paper. We became available to handle more volume. That's why we decided not to impose that restriction of signing our MNDA. And to say less, were sales happy with that? They were just in the heaven, really. And we received a lot of positive feedback. There is no friction anymore. There are no back and forth about which paper to use. It just became much, much smoother.

[00:49:24] Speaker 3: Awesome. Thank you, Misha. Appreciate it. I know we have, do we have six minutes left?

[00:49:32] Speaker 1: We have, yeah, I think we kick it over maybe to Q&A. What do you think, Sona? Make sure we're getting everything answered. Or do you want to have a few more for Misha? Totally up to you.

[00:49:41] Speaker 3: Let's do a few. Let's do some Q&A. Okay.

[00:49:43] Speaker 1: That's awesome. Misha, thank you so much for your insight and experiences. I know, obviously, this is the conversation live, but I've really enjoyed learning about kind of the iterations and evolutions your team has gone through in this process. And I'm sure our audience loves it as well. So with that, we have a Q&A open. If you all will use the Q&A box, I know quite a few of your questions have been answered, as we have many people in there, Ryan included, making sure that everything is getting answered. I'm very excited. Me too. I'm loving to hear things like this. Someone says, I'm ready to use it. So, Michael, at the end of this, I'll make sure that you have a link to get in touch with the right people to make sure you can. So I think folks are looking a little bit for the litigation timeline. I know it's on the roadmap. We can make sure that we are, you know, updating each person that came to this webinar as new features release. But, Ryan, anything you want to add around litigation timeline or not to put you on the spot?

[00:50:46] Speaker 2: Yeah, if I commit her to a date right here now. So I won't do that. But, you know, look, I mean, you saw a preview. It's going to be incredible. And it is it is a very high priority item on our roadmap. I don't know if you want to speak any more to it. No obligation. So I don't want to put you on the spot.

[00:51:03] Speaker 1: No, I don't want to put anyone on the spot either. I just love a good teaser in these webinars and want to make sure that people know that we will let them know as soon as this is live. So.

[00:51:12] Speaker 3: Yeah, I think the biggest thing is legal tech companies always make promises about product roadmaps. We are actually building it right now. So it is in process of being built. It's not a future roadmap item. I saw some other questions from current customers of Pinsights as well. Yes, we are investing into the Word Add-in. We're hiring more people for our team. So you can expect even faster everything from bug fixes to new product stuff. We have two new features that are pretty big coming in the next, I think, two-ish weeks. So we're really excited for that. And I saw some questions about different jurisdictions. We do work in different languages, so it will match the language of the document for the drafting, redlining, and comments and all that. Amazing.

[00:51:59] Speaker 2: I'll just say from a broader perspective, what you can expect from us is deep investment into Sona and Merriam's team. You know, Merriam's not on the call today. Sona is an extraordinary founder and CEO who, you know, I think most of you, well, some of you maybe know the story, but we used, we were lowest customers, excuse me. We were Pinsights customers for about a year before we acquired the company. And, you know, Sona won't talk about this unless I absolutely prod her to, but she had a lot of opportunities to sell this, to sell her company to many different potential companies. It was a very highly contested, a lot of people wanted to acquire Pinsights. And frankly, I don't, she didn't need anybody to acquire her at all. The company was doing phenomenally well and she could have continued on the journey and built a legal tech company on her own. Fortunately, I was able to convince her that doing it inside the Filevine ecosystem provided some really cool enhancements, not only with the data, but, you know, distribution, but also we really believe in her and we really believe in Merriam and we will invest deeply into her team. They will anchor the office we're opening up in San Francisco. And so her team of engineers and designers and product people will work out of our San Francisco location, which we're opening up here in about 30 days. So you can expect not only a really cool integration with Filevine, they can speak to litigation and all sorts of other uses, but also you can expect that the lowest for Word integration in Word continues to get better across a host of different features. And it's going to be awesome.

[00:53:39] Speaker 1: Awesome. Thanks, Ryan. Sona, knowing you have been in this world a while, obviously, it being your company prior, there have been some questions in the chat around just where this stands out against the competition when you're thinking about Pinsights specifically or Lois for Word. Do you want to give the audience just a few high level pieces around your experience and where Lois for Word has really stood out for the companies or teams that have purchased?

[00:54:07] Speaker 3: Yeah, that's a good question. I think every company has features. Other folks have features we don't have. We have features they don't have. And it's kind of up to you to create that criteria and evaluate things across it. What we have done is instead of building a lot of features, we want to go really deep on making things that we have good. So the core functionality of creating red lines that are light, that are surgical, and everyone always mentions that it's actually very hard to do, to understand the whole layer of context the document has gone through, what other pieces of information you need to pull from, how to do that well. The playbook creation flow, making that as clean as possible so someone can feel comfortable owning it rather than outsourcing the creation of it. And that's very key IP that you should not be outsourcing, in my opinion, because you're building essentially how your company handles a certain type of document. And that should be handled internally. And you should have the expert on your team who does that. But it's hard to build that workflow really nicely where someone would want to do that. So I think those two things are probably where we stand out. Just ease of use also just comes in a lot. And that will depend on you guys evaluating different tools. But I do think we're one of the top just for that.

[00:55:23] Speaker 2: So I know you're obviously unbiased, but when we did our review of competitors to Pinsights, what stood out to us was speed. The product is just so fast. AI products, sometimes if they're calling models and they're doing a lot of thinking and they're adjusting a lot of data, you can tell how AI native the engineering team really is by just how quickly the responses come back and how, of course, how accurate they are. And given that we had been customers for a year, we knew it was really fast. But when we compared it to competitors, it was just so much quicker and snappier to use.

[00:55:58] Speaker 1: One other thing I'll add, and thank you to an anonymous Q&A sender, but one of the areas that stood out to them was that they tested all of the different competitors and they read the Pinsights output and thought it was their own team or it was a human that wrote it. So I think that's an amazing thing to call out here and love to hear from our attendees here at the webinar. So we are here at one o'clock. As you can see here, there's a QR code on your screen that you have had up. For anyone that is interested in learning how this directly applies to your team or how you might weave this into your day-to-day processes, please scan the QR code. You'll create a quick form and we will connect you with the perfect expert to ensure that we are going to get you the use cases that specifically apply to you. As you all know, we have tons of overwhelming excitement around a webinar and we can only cover so much. So we would love the one-on-one opportunity to help answer your questions and develop the right flows for what would work for you and your team. So hopefully you can have as good of an experience as Nisha has had in adding a new product and being able to do more for his in-house team as well. Lastly, as promised, many of you showed up for the ability to download this AI evaluation guide for legal teams. Feel free to scan and download it, all of you that stayed on. This is your guide to help you decide what is the right AI product for you and your team. We hope that this asset is a nice add to your day and as you think about these different products you're evaluating, you have a nice checklist that will help you. So I want to thank you for putting this together with your team. I think our audience will love this as well. So with that, all of our speakers, thank you so much for speaking. All of our attendees, thank you so much for all of the great questions in the Q&A. And with that, we will hopefully be having one-on-one conversations with you on how we can implement Lois for Word into your day-to-day. Thank you, everybody.