NetDocuments CPO Dan Hauck が語る、刷新プラットフォーム・Legal Context Graph・エージェント型DMS

Today on Lawnex, I am live at the NetDocuments Inspire Conference in Orlando, Florida, where I've just interviewed Dan Hawk. He's the Chief Product Officer at NetDocuments. Earlier today, he gave a talk laying out a number of the new products and features being introduced by NetDocuments. I talked to him about that and about developing products in the age of AI. All that and more. But first, please take a moment to learn about the sponsors who support Law Next. BriefPoint reapplies your objections and response strategy automatically.

No data entry, no setup. Watch the 60-second demo at BriefPoint.ai. Streamline your law firm operations with PracticePanther, the all-in-one legal case management software. Trusted by tens of thousands of lawyers, PracticePanther's easy-to-use features, such as customizable workflows, automated billing, and seamless client communications, help you accomplish more in less time. Making the switch to Practice Panther is easy with dedicated data migration, live training, and support. Get your free trial today at practicepanther.com. Now let's get to today's conversation. Welcome to LawNexus. This is Bob Ambroji, and I am here live at the NetDocuments Inspire Conference in Orlando, Florida. And I'm here with Dan Hawk, Chief Product Officer at NetDocuments. Dan, welcome. Awesome. Thank you for having me, and thank you for coming and joining us here at

the conference. Yeah, I'm glad to be here. It's been actually a couple of years since I've been to this conference, and it's interesting to see how it has grown, but obviously how everything is changing so quickly as well at the same time. Yeah. What do you think? I mean, how are you enjoying this conference? Yeah. One of the great things about the conference is it's a chance to get everybody together and to share how things are going and get feedback and understand how people are using our product for us to be able to collect that feedback, for us to have conversations with customers about where we're going, what we're adding. We have multiple keynotes throughout the day or throughout the couple of days. And it's a lot of fun. It's obviously my favorite conference of the year. Yeah. You were on my podcast in 2022, which is the last time you were on this podcast. And the world has changed a lot since 2022. Yeah. You know, I want to dive into some of the more specific product announcements you've been making here and what you've been talking about. But kind of what's your perspective on, you know, the bigger picture perspective of where we are right now and how things have evolved over the last couple of years?

Yeah, I think that, you know, one of the just early recognitions that we all had was that for the first time, we had the capacity through large language models to really understand the unstructured data of what is in all of the legal documents. And at NetDocuments, we store over 23 billion of them. And so helping customers understand what's in them so that they can then use that in more purposeful ways to find precedent, to align with client preferences and things like that. All of that is data that's stored inside of their documents. And traditionally, it was hard to pull out. And a lot of the older methods around machine learning and so forth just weren't very effective. We had explored them very deeply to understand how we could leverage them, but we were pretty dissatisfied with the results. And then even with those early LLMs, while we started investigating them, I think the speed, the cost, the context window, all of those things were still pretty early in terms of us being able to really ingest documents at scale to extract out critical information and to store that data back into that document alongside the document.

But we saw the trajectory pretty clearly. These models were advancing rapidly. They were getting smarter, better, faster, cheaper, so that we could do the kind of large-scale operations that we are now here with our customers doing, you know, classifying millions of documents inside customer repositories, building out rich taxonomies and ontologies, building out data structures that I think we had always dreamed. It's kind of the dream of the firm or the legal organization to understand what's in all of their documents. And now we're at the point where we're doing that. Yeah. You were a former practicing lawyer. You were a company founder. You had founded ThreadKM, which NetDocuments acquired. Given that background, that perspective, you know, when you talk about the kinds of changes that you've just described, what is the practical implications of that? What does that mean for the rank-and-file people who are out there working in every day and using NetDocuments? Yeah. I mean, it's interesting because although there is a lot that's new and there's a lot of unchartered territory, I do create parallels to my own experience when I was practicing law.

I think about like the e-discovery work that I had done. And when I was a litigator, kind of that whole concept of technology-assisted review, which was a much simpler form of the kind of tech that we have today, but it was helpful in being able to identify issues, being able to classify for privilege and things like that. And there was real questions about, well, wait a minute, we have all of these people who are manually reviewing each of these documents. Is this technology-assisted review going to be as good as what they were doing? And there was a real kind of perspective change as people realized, well, the humans aren't perfect and the technology isn't perfect, but together there's these ways that we can put them together to create these greater outcomes. And it required real change and organizations having to think through how do they do discovery differently. Well, now we're talking about that for the entire practice or for the entire, you know, all parts of the legal services delivery. And that's where we're focused on helping our customers be able to leverage these tools effectively.

As you know, we've talked quite a bit about being able to develop the right context so that agents that are working on behalf of our users are able to operate much more effectively. So that is going to create kind of downstream changes to how lawyers work, and we're seeing that happen today. Yeah, it's funny to talk about TAR because I remember that kind of seismic shift when people started realizing this can not only do what a lawyer does, but maybe in some cases do it better than a lawyer or at least faster than a lawyer. But you're right. It was always that combination of the tech and the people. And I guess that's still true, right? It's still that the human loop is still a critical part of this whole equation. And really what we're finding is that more and more you're kind of abstracting up what you're doing as the lawyer. You're not necessarily doing the things that you were doing before. You're directing agents. You're directing some of this work, and you're using it to perform the tasks at a scale that wasn't feasible before. Yeah. Obviously, one of the big announcements over the past six months or so from NetDocuments has been this whole development of the legal context graph, as you've been alluding to.

Talk more about what that actually is and how you construct that. How does it get created? Yeah. So, I mean, if you think about it, I mentioned before that every document is really a set of unstructured data. And in NetDocuments, we store everything from contracts to pleadings, lease agreements, different kind of transactional documents. And it's voluminous, and some of them are dozens or hundreds or thousands of pages long. And so the thing is, when an agent goes to grab one of those things, it might not – that may say, oh, that's the wrong document. And then it goes to grab another one and says, that's the wrong document. And so what we have found is that lawyers are oftentimes having to direct the agents very intentionally to just the right information in order to answer a question like hey here my question I get an answer back Well that not quite right I need you to read these other documents here. And what we saw was lawyers who were manually constructing context,

right? They were giving the information that agents needed to be more successful, but it was very time intensive. It was incomplete. And the problem was you would have a colleague down the street or down the hall who was doing exactly the same thing. And so that starts to create real token costs as well as you're redoing this work. And so at NetDocuments, we focus on how do we make it so that that context is just appears automatically and readily available for an agent. And so what that ends up looking like is when you save a document into NetDocuments, as part of that process in our ingestion pipeline or our document intelligence pipeline, We're helping analyze what kind of document is this? What are the key pieces of information that we would want to extract from it? Who are the parties and people that are referenced inside of this document? What are the key dates and activities that are happening there? All of that information allows us to start to roll up into more of a graph view as to what this matter is all about, where expertise sits inside the firm.

And so that then when you ask a question to an agent of, you know, how should we adjust this part of the agreement, it can look at similar agreements that you've done, similar agreements, you know, from that jurisdiction in that industry with that particular client. And it can do that much more effectively. In fact, we found savings of almost 50 percent of token savings of when you're using the context graph to get to the right answer versus when you don't give it that graph that we provide. Mm-hmm. How do you sort of create a system that understands, so to speak, what's legally meaningful in terms of the context of those documents? Yeah, so we have a great team at NetDocuments of legal engineers who are looking at this challenge every day. They've built out all of our test harnesses so that we can put all of our assumptions to the test, and we use real documents. We use based in real matters that are being evaluated by real lawyers. So, you know, one of the things we found is that sometimes when you rely too much on the synthetics, it's kind of easy to get to an answer that you like, even if it's not the right answer.

And so we were pretty rigorous about that approach. And we've also leaned on some of the kind of foundational work that groups like Sally and Folio and others have done, which has proven to be incredibly effective in the age of AI, where you're basically going from asking people to figure out what's important to now be able to provide a taxonomy or an ontology and have an agent be able to match that up and with very high accuracy rates. Yeah, we had a press briefing just before this conversation takes place. We were kind of talking about this development of ontology versus taxonomy. Have you kind of literally created that, or is it being created by the generative AI? Yeah, so we actually have an out-of-the-box ontology that we provide customers. And so, for instance, our document intelligent pipeline uses it natively. But we also have the ability to work with firms and customers who want to adapt that in ways that are meaningful for their firm, particularly firms that have specialty practice areas or specialty clients,

and there's certain information that they're always going to want to pull out of their documents. We have teams that work with those customers and allow them to do that at a scale and speed that they've not been able to do before. Yeah. You're listening to a conversation I recorded live at the NetDocuments Inspire conference with Dan Hawk, the company's chief product officer. Much more to come in our conversation, but first, please take this opportunity to learn about the sponsors who support this podcast. Law firms lose thousands of billable hours each year because the work they do never gets captured. Ajax uses AI to turn activity across screens, email calendars, and documents into ready-to-bill entries tailored to your firm's billing rules and outside counsel guidelines.

Entries sync directly to your billing system, helping firms capture over 12% more billable time. 97% of firms that pilot Ajax choose to keep it. Learn more at joinajax.com. Practice Panther is the leading practice management software designed to help you analyze, automate, and grow your firm. intuitive design and easy to use features like customized workflows automated billing and client portals will help you streamline your practice plus dedicated data migration and live training will help you get up and running quickly get your free trial today at practice panther.com and see why practice panther is the smart choice for lawyers again that's practice panther.com Welcome back to LawNet. I'm continuing my conversation with Dan Hawk, Chief Product Officer at NetDocuments,

which he recorded live during the recent NetDocuments Inspire Conference in Orlando, Florida. You can see a little bit of that conference behind me right here. Let's get back to my conversation with Dan. you are on stage this morning uh talking about a number of new developments uh one of those was basically the new user experience the new the complete redesign of this platform tell me kind of what it was that was the goal of this redesign and how it is that you've achieved that goal, assuming you have. Yeah. Yeah. So we essentially had been working for the last couple of years on a new user experience where we were going to really capture the way that the modern lawyer wanted to work and all of the different interfaces and design patterns and so forth. So there's been a lot of work that had been going into that already. And then as we really started

to pivot into this agent ready world, we thought, well, what is an agent ready experience look like inside of this new experience. And so it was an opportunity to really rethink how does our AI show up? How does that automated extraction start to appear inside of our platform? How does the graph that we're building really show up in ways that a lawyer can basically look at one screen and say, I'm getting a matter summary. I'm seeing all the people who are here. I know who from my team is working in this particular case. I have everything at my fingertips. And also it's readily accessible to any agent that needs to hit that in a secure fashion as well. And so that was really an exciting opportunity here at Inspire to be able to unveil it. To your point about have we hit the mark, we've been in private preview with a number of different customers over the course of basically since May, and the feedback has been phenomenal. We've been really excited. There's people who are saying we want to use this right now across our entire firm.

So that's the kind of feedback that we're excited and that gives us confidence that our users are going to have a great experience. Is there a conflict? I don't know if that's the word I mean, but is it difficult to both be building these more and more powerful features into the platform, but then also designing a platform that's easier to use and more intuitive to use and more user-friendly? Is that difficult to kind of achieve that balance? Yeah, I think it's something that, you know, it's an important part of the work that my team does in product and design and with our engineering counterparts is that we're trying to build that right experience. And our users are the ones who will tell us if we've hit it or if we need to make changes. And we appreciate them doing that. And I think the other thing that's been interesting as well is how are we designing the legal graph as a work surface, not just for our own solution, but how do we make that accessible through MCP and other solutions, being able to connect to that in secure fashions as well? And so we really have thought about holistically as where lawyers are working regardless of what the work surface that makes sense for them to be in Many times it net documents but we seeing connected applications as well

So what's the timing now for the rollout of this new experience, and how do customers transition from the existing platform to the new experience? Yeah, so we're actually making both our new M365 capabilities, our co-authoring capabilities, and this new experience all available on November 12th. So we're less than a month away. We're really excited about that. And we're also going to be continuing to maintain and improve our existing experience as well. So there is no forced cutoff migration or anything like that. We expect to maintain the two solutions for some time. They actually live on the same platform. And, in fact, the user can switch between the two experiences whenever they want. We want firms to be able to be in control of their transition over to the new platform and have provided the capability to do that. Was it you this morning or one of the other speakers just said one of our customers told us to turn off the toggle switch? We like the new one. That's right. That was Jeff Carter, our head of design.

And, yeah, that's the kind of feedback we love. Yeah. We didn't even talk about the M365 experience. So do you want to explain what that is? Yeah. Yeah. So traditionally, most integrations with Microsoft products were done through what's called a com add-in, which is a desktop application that integrates with those solutions. But Microsoft has for some time now, but really invested heavily in creating a strong integration footprint through M365 add-ins. Those are easier to deploy. They work across any device. So whether you're on Windows, Mac, mobile experiences, you can do all of that. And so we built a purpose-built M365 integration that's all new. It matches all the design language and patterns of that new web experience that we talked about. And it also facilitates the co-authoring experiences as well so that people can work in the same document inside of NetDocuments all the time. So, yeah, that's also becoming available as part of this rollout. And co-authoring means that, as you just said, that people are working in the same document at the same time.

How does that get resolved? I mean, if two people are working the same document making edits at the same time, I use Google Docs all the time. So I assume it's as simple as that. Exactly. It's actually, and if you've used Microsoft SharePoint or OneDrive, we actually are using the same services that Microsoft itself uses in order to facilitate that. So it's a very familiar experience for a lot of people. You can see your colleagues in there and what parts of the document that they're editing. But all those edits go straight back to NetDocuments. So you've got – you're continuing to build out these other features. Tabular review is another one that we haven't talked about yet. Why don't you – you want to kind of quickly explain what that is? Sure. Yeah. As we think about, like, that ability to do more in AI inside of NetDocuments, we've added a lot to our assistant. One of the things we introduced a couple of months ago was our new tabular review. That's now going into general availability this month after a lot of very positive feedback from customers who are running thousands, five, 10,000 documents through to be able to do large-scale due diligence reviews and other types of analyses where you're wanting to look at a large set of contracts and pull out key information from those.

And the nice thing is those documents never have to leave net documents. It's all part of a singular system of records, singular audit trail, and you can get outstanding results with our tabular review. We've also done a few – I'll share a few more things on the Assistant that we're actually introducing here at Inspire. We have introduced the ability for it to now start to do all the document generation, document editing. And so right inside of NetDocuments, you can ask it to update a document, create red lines, et cetera. And maybe my favorite feature is the ability to create those HTML style documents. So if you've ever had that experience of asking an AI solution to generate a report or to put together something, the beauty of those is that they're interactive. You can click on – it can generate toggles and dropdowns so that you can customize a view to what you want. That's really powerful when you couple it with the data we're extracting in our context graph. You can look inside a matter or series of matters and say, show me all the positions we took across all of these different matters in this contract negotiation or something along those lines.

And ultimately, if you want to make adjustments to that, you can just tell our assistant to how you want that report to be adjusted. And it all stays with inside net documents. You can save that report back into net documents as well. Yeah. So does this all become – obviously, this all becomes sort of an integrated environment where firms are able to do, more or less, whatever they want to be doing within NetDocuments. I mean, do you see NetDocuments becoming kind of the core work platform for law firms, or do you see law firms continuing to use a variety of different applications and then integrating them within NetDocuments? How do you see that developing? Yeah, we see both. We have over 7,000 customers, and we always talk about meeting them where they are. For a number of our customers, they want to rationalize their technology spend. They want to simplify their governance and training footprint so that it's easy for them to deploy net documents as a single solution and have access to our assistant, which can do natural language search across their entire repository,

to tools like tabular review and document creation. So we make all of those available. But we also recognize that there are firms who either across their organization or in pockets of their organization might have different AI solutions. And so that's why we worked hard to develop our ND Connect program so that those solutions can tap into net documents in a governed, secure way, and in particular, leverage that legal context graph. because that's how they're going to get better results for their users. When you ask a question inside a matter and it knows all the people who are there, it knows what's been going on in the matter, the timeline of events, and it can navigate across all of the documents inside that matter really quickly to get to good results, that's a better outcome for our partner and the user as well. Yeah. Yeah. Something else you announced this morning was the new, I think, beta version of a MCP connector with Legora. That's right. I know you've already got MCP connectors with Claude and ChatGPT and Copilot and Gemini, I think.

And who else? I'm missing a Harvey. Do you have a Harvey? We do. Is there a Harvey NTP? Yeah. And just kind of curious your thoughts. I mean, I'm fascinated by this kind of what this development portends in terms of how workflows can evolve in the legal profession. And I'm wondering what you think about what this means in terms of how lawyers maybe should be thinking about MCP and how you think it's going to evolve in terms of how they will be using these different kinds of connectors. Yeah, to the point earlier, you know, we're seeing all of those tools being used inside the same firm, right? And that's not uncommon right now. And the beauty of MCP and the reason that we've made really strong investments here is that it does have the ability to – it's a protocol that goes across all of them so that it allows us to be able to build a unified connector that all of our partners can leverage so that those partner applications can improve. And really what we hear about most from our customers is that they want to be ensuring that their system of record stays intact as a single source of truth.

It is where all of the kind of governance is maintained. It's where all of their kind of documents are maintained. We're adding write-back capability as well so that if you have one of these other solutions that's generating documents and other content, you can just tell it, hey, save it back into NetDocuments. And that agent will pick up that work and move it over. So those are the kinds of things that I think you're exactly right. There's a lot of fluidity in terms of how firms are adopting these different technologies. Some are allowing that kind of multiplicity of solutions. Others are kind of being more limited. But regardless, we're focused on serving the core asks of our customers, which are really around how we make their data more valuable and how do we make it easy for their users to work with. Yeah To me it raises an interesting question about again on the workflow side of things it almost makes the UI for a platform less important in some ways If somebody is able to get into the data underlying that platform through any number of different

ways, if I want to just start in Claude and start then, you know, pulling documents out of NetDocuments that I'm working on, in a sense, it makes it less urgent for me to go into NetDocuments and start working in NetDocuments. I think what I heard you and Josh saying at the press briefing this morning is that part of what you want to do is to just make net documents so compelling as a platform that people are going to want to go there versus coming in through some other way. Is that fair to say? Yeah, that's fair to say. In fact, but we've had those conversations about, you know, we're investing a lot in this new experience. There's other ways to integrate. And so that's why, you know, first principle is we want to make sure that the data that we're serving up is the best in class data that allows whatever agent you're connected to to be effective. But we are really excited about the new experience that we've developed because oftentimes, and we're already finding that if you're in a particular workflow, hitting that documents along the way and being able to pull in some information from there is really valuable. But a lot of the work that our users are doing is kind of core managing of files.

They need to be looking at a large set of documents. They need to be moving documents around. They need to be doing a lot of that kind of core activity, and they like to be able to see it. And the interaction patterns are very, very fast. In fact, one of the key pieces of feedback that we've gotten from customers on the new experience is it's very fast. And so it's going to be faster than what you could get in that kind of connected experience mediated through MCP. Yeah. Just before we run out of time, I've been asking you a lot of questions, but what else is really kind of top of mind for you in terms of what you're talking about this week? What are you most excited about? Yeah, I think we hit on a lot of it, but I would say the most exciting thing for me personally, and we had a great example of it yesterday, Inspire is a place where we do get to bring together our customers. And we had a meeting of a number of our different customers in what we call our ND labs, where we have brought some of our leaders in AI professional services, some of our leaders in our customer base who are leading KM innovation, information technology functions.

And they actually started working on building in code enhancements to our existing solutions to help profile better, to do some legal specific domain app work and so forth. And so it's that work that they started outlining yesterday is actually going to be something that we're going to be reviewing tomorrow and potentially even you'll see in the product. So I'm excited to see what they make. But that's the kind of speed and innovation that we are operating at. Yeah. One of the things I love about coming to this conference is, for me, the ability to sit and actually talk to customers because I don't know if I often get to talk to customers. I hear from the marketing and PR people, but not the customers so much. And, you know, my sense is, I mean, NetDocuments is interesting because your customers span almost the full range of firm sizes, all the way from small firms up to the biggest firms. What are you kind of hearing from customers, not just the sort of sophisticated ones that want to come into a lab and sit down and do some work with you, but how are they wrestling with?

How are they adopting a lot of these new AI tools? Because just some of the few conversations I've had here so far in the last couple of days have been people saying, you know, in our firm, we're really still struggling with what AI we want, how we're going to use it. Lawyers are still a little bit reticent about adopting it and using it. Yeah. What are you hearing? Yeah, I think that's a really important call out that firms are at different points in the journey. Yeah. And one of the that, again, just to kind of go back to the legal context graph, which I know we talk about a lot, but it's because we see it as so foundational to getting success with AI. And so a lot of the work that we've been doing with larger firms, we're now bringing down to many of the smaller firms at NetDocuments, which is to be able to really do that rich profiling and start putting their documents through our context graph. As that starts to happen, what you'll see is just better results and more efficient results. You know, there's nothing more frustrating than using a piece of technology and not getting the outcomes that you want or just not delivering the results that you were expecting.

And we've seen that with the context graph, you're able to get to better results more consistently. And so regardless of the solution that they have, and I think for a lot of them, they're going to see those results inside of our assistant. They're going to be able to just have tools like tabular review at hand. As those start to get deployed more widely into those kind of customers, we want to create those natural paths for them to get to good outcomes. And probably even more important for smaller firms, midsize firms that haven't had sort of an advanced KM staff or innovation staff being able to kind of work on creating those threads on their own manually or however it was done in the past. Yeah, no, that's exactly right. And our tools, we spend a lot of time making sure that those tools are easily adapted and that they show up naturally in a user's workflow. Yeah. Again, as you bring more agentic capabilities into NetDocuments, again, how do you see that developing evolving? What is kind of the future of an agentic DMS?

How does it look different in terms of the lawyer, in terms of a practical lawyer's workflow from where we've been in the past? Yeah, I mean, you know, going beyond what we're doing from the profiling of content and so forth, it's really going to go more deeply into those workflows and exposing capabilities that allow you to really talk to the DMS to do the things that you want to do. If you think about, you know, you might have a messy workspace and say, hey, I need you to clean up this workspace. And I've got a bunch of different documents here. Organize them and maybe rename them and then put them into some folders. And if the folders that are here aren't correct, maybe suggest a couple of other ones. That's how we might talk to one of our colleagues or a legal assistant. And that's how we can start talking to agents as well. And suddenly now the workspace is getting kind of set up the way exactly that you want. So those are some of the areas that we're thinking about. Obviously, you know, there's a lot of things that we're working on that we probably can't talk about here. But we're excited to be, you know, things are moving really fast.

and we're excited to be delivering more stuff to our customers. Yeah, how do you keep up? I mean, it's chief product officer. How do you keep up with the pace of change right now? I mean, I remember the good old days when like maybe every year at your customer conference, you might announce some stuff, but now it's just getting rolled out all the time and evolving all the time. I think that the quality of the legal tech conversation, and you've been a key part of this conversation for a very long time, so thank you. It's been awesome. and as that conversation has continued to grow and take place in places like LinkedIn and others, what you're seeing is a really incredible sharing of ideas that are very legal-oriented. I mean, there's some amazing minds who are operating in this space now who are innovating in this space. So I try to digest as much of that as I can. I also find that my own understanding of how to use some of these technologies or how we could use them at NetDocuments stems from, you know, work I do outside of NetDocuments as I, you know, build different projects or try different things where maybe it's okay to fail more frequently

or see what the limits of some of these technologies are, figure those out, and bring some of those lessons back into the company as well. Like wiring your house. Yeah, that was an example before. But, you know, at this point, Claude has an understanding of all of my home networking, my home automation, all of those things so that it has the right context so that it can answer things more quickly. And, again, that was a lesson around context that I started to develop there. Right. For anybody watching, wondering what the hell I'm talking about, Dan, at his opening remarks this morning, talked about his own kind of nature of wanting to fix things and repair things and build things in his personal life as well as professionally. Yeah. Yeah. It's a fun hobby. Yeah, right. Well, Dan, thanks a lot. Great to talk to you. Always great to talk to you. Thanks, Bob. All right. I'm talking to Dan Hawk, the Chief Product Officer at NetDocuments here at the Inspire Conference in Orlando. This is Bob Ambroji. Thanks for watching tonight.