NetDocumentsのHeath Harris氏が語る「タクソノミーこそAI戦略」――最終的に勝つのはデータ

NetDocuments' Heath Harris on Why Your Taxonomy Is Your AI Strategy and Why, in the End, the Data Wins

LawNext35分
NetDocumentsのHeath Harris氏が語る「タクソノミーこそAI戦略」――最終的に勝つのはデータ

NetDocumentsのHeath Harris氏が語る「タクソノミーこそAI戦略」――最終的に勝つのはデータ

LawNext

0:0035:16

要約

LawNextがNetDocuments Inspire(オーランド)で収録した、同社のHeath Harris氏(応用AI・エンタープライズ変革担当VP)へのインタビュー。法律事務所のAI導入は「何でも買う」段階を過ぎ、戦略やブランドとの整合を問う段階に入ったと述べる。同氏は経済合理性が投資姿勢を試すと見ており、タクソノミーによる構造化データがAIエージェントの読む文書量を減らすと説明し、最終的にデータが勝つとの見方を示した。

この回のポイント

  • ●法律事務所のAI導入は「全部買う」段階から、自所の戦略やブランドにどう結びつくかを問う段階に移りつつあるとHarris氏は見ている。
  • ●2年目にパートナーが「AI人材・インフラ」か「顧客基盤を持つラテラル採用」かに予算を投じる投票をする場面で、経済的な制約が効いてくると予想している。
  • ●AIファーストの法律事務所やスピンアウトには、価値ピラミッドを上るほど必要になる保険的性格やグローバルな拠点の面で天井があるとみるが、Am Law 100入りする最初の例には注目している。
  • ●「タクソノミーはAI戦略」とは、事務所の知見が文書に宿るため、それを人にもエージェントにも使える構造化データにすることが出発点だという意味である。
  • ●タクソノミーがあればエージェントは数百万件ではなく数百件程度の文書を読めば済み、コストが下がると説明している。
  • ●MCPをめぐるガバナンスはCIOやAI責任者との対話で常に上位3テーマに入るという。事務所は、外部出身者を採用して実質的な決定権を与えるべきだと提案している。

章立て

  1. Harris氏の役割と経歴

    顧客に寄り添いAI戦略策定を支援するチームの役割と、Fenwick & WestやEYでの経歴を紹介する。

  2. 生成AI時代の市場変化と導入状況

    生成AIはインターネット並みに普遍的で、企業側の圧力もあり、事務所は「全部買う」段階から戦略との整合を問う段階へ移っていると語る。

  3. 2年目の予算と事業モデルの現実

    AIインフラ投資とラテラル採用のどちらに資金を使うかというパートナー投票の問題を論じる。

  4. AIファースト事務所の可能性と限界

    スピンアウトやAIファーストの事務所には価値ピラミッド上の天井があるとみる一方、Am Law 100入りの可能性にも触れる。

  5. 「タクソノミーはAI戦略」

    事務所の知見を構造化データにする意義と、タクソノミーでエージェントの検索範囲を絞る例を説明する。

  6. MCPとガバナンス、データが勝つ理由

    ガバナンスへの懸念、モデルより変わりにくいデータの価値、DMSの位置づけを語る。

  7. 事務所が今すべき準備と締めくくり

    データ基盤の整備、外部人材の採用と決定権の付与を勧め、業界の変化への期待で締める。

解説記事

法律テック分野のポッドキャストLawNextが、NetDocumentsのカンファレンスInspire(フロリダ州オーランド)で収録したインタビューの内容を紹介する。話し手は、同社で応用AI・エンタープライズ変革を担当するVPのHeath Harris氏。FenwickやEYで事務所側・企業側双方の変革を担った経歴を持ち、現在は顧客のAI戦略づくりを支援している。

「全部買う」段階の終わり

Harris氏は、今回の生成AIの波は過去のNLPやRPAの時期と違い、インターネットのように普遍的で、大きなコスト障壁もないと述べる。企業の法務部門ではCEOがGCにユースケースを求め、株式市場もAI導入を評価しているという。

法律事務所では、金融、製薬、ビッグテックなどを顧客に持つグループが先行する。最近の変化として、「とにかく全部買う」段階を越え、AIが自所の戦略やブランドにどう結びつくかを問う動きが出てきたと語る。同氏が見るのは、プレスリリースよりも誰を採用し、実務で何をしているかだ。技術とビジネスの両方に通じた人材が意思決定の席に就く例が増えていると述べている。

2年目に問われる経済性

同氏は「最終的には事業モデルが勝つ」との立場だ。事務所は、パートナーを育てるか市場から買うことで売上を伸ばす。顧客からのコスト圧力が続けば、成長手段はラテラル採用になりやすい。

そこで同氏が投げかけるのが、2年目にパートナーが再び票を投じるときの問いである。2000万〜3000万ドルをGPUへのアクセスに使うのか、Microsoftのような顧客基盤を持つ人材に使うのか。事務所は今も現金ベースの事業だとして、経済的な制約が一定の影響を及ぼすと予想する。AI責任者の報酬が1000万〜1500万ドル規模になるシリコンバレーと競うより、最上位の法的助言に集中するほうが理にかなうという見方も示した。

AIファースト事務所の天井

事務所がAI専業のスピンアウトを設ける動きについて、Harris氏は「プレスリリースはプレスリリースだ」と慎重だ。ただし、フェデレーション型でインフラを再販するモデルは、下位市場では信頼できるブランドのインフラを使う事務所が出るかもしれないと述べる。

一方で、価値ピラミッドの上位では、助言に加えて保険としての意味合いや、複数法域にまたがる実働体制が必要になる。このためAIファースト型には自然な天井があると見る。それでも、売上でAm Law 100入りする最初のAIファースト事務所には注目しており、Fortune 500の契約処理などで年間数億ドル規模に届く可能性はあるとの見方も示した。

タクソノミーがAIの効率を左右する

本題の「タクソノミーはAI戦略」とは、事務所の経験や強みが成果物の文書に宿っているため、それを人にもエージェントにも使える整理されたデータにすることが出発点になる、という意味だと説明する。同氏は元ナレッジ責任者のMark Ford氏と、SALIなどの標準を簡素化したタクソノミーづくりに取り組んだという。

具体例として、「北欧のバイオ製薬企業との2億5000万ユーロ以上の株式取得で、特定のパートナーが関わった案件の株式譲渡契約書を探す」場面を挙げる。タクソノミーがなければ、エージェントは見つかる限りの文書を読み込むことになる。あれば、コーポレート分野、契約書の種類、業種、取引額で絞り込め、読むのは数百万件ではなく数百件程度で済むという。

同氏はNetDocumentsが公表したコンテキストグラフのベンチマークにも触れ、合成文書ではなく公開された取引や訴訟を使い、実際の弁護士が検証した点を評価した。他社が似た評価を出していることも、手法の妥当性を示すと語っている。

MCP、ガバナンス、そして「データが勝つ」

MCP(AIモデルと外部ツールをつなぐ規格)については、ガバナンスがCIOやAI責任者との対話で常に上位3テーマに入ると述べる。「シャドーDMS」への懸念もあり、深い連携を作りつつ、ガバナンスの枠組みを壊さない方法を各社が学んでいる段階だという。NetDocumentsは早期にクラウドへ移った経験から、この点で価値を出せるとしている。

今後については、モデルやそれを包むハーネスは変わりやすく、勝つのはデータだとの見方を示す。必要なのは検索、ワークフロー、成果物の作成という基本機能だという。すべてを一社で束ねようとするのは無謀かもしれないとも語り、DMSがなくなる可能性にも「絶対とは言わない」としつつ否定的だった。「ツールは簡単に手に入る。難しいのは頭脳を整理することだ」という言葉が印象的である。

まとめ

日本の読者にとって示唆が大きいのは、AI導入の成否をツールの数ではなく、データ基盤と組織の意思決定の質で捉えている点だ。Harris氏の助言は、まずデータ構造化に本気で取り組むこと、事務所外でビジネスモデル変革を経験した人材を採用し、会議に同席させるだけでなく決定権を与えることの2点に集約される。

なお、ここでの見通しは同氏個人の見方であり、NetDocumentsの立場からの発言でもある点には留意したい。それでも、法務に限らず社内文書を扱う企業にとって、分類体系の整備がAI活用の前提になるという指摘は参考になる。

文字起こし(英語・自動生成)

Welcome to Law and Exit. This is Bob Ambroji, and I am here live at the NetDocuments Inspire Conference in Orlando, Florida. Today brings a third of three interviews I've recorded here. You've already heard my interviews with Josh Baxter, the company's CEO, and Dan Hawkins, Chief Product Officer. Today's conversation is with Keith Harris, Vice President, Applied AI and Enterprise Transformation. He and his team are out there with customers, helping law firms and legal departments on AI strategy and implementation. And what a team learns there, they bring back into product development. I'll get to Heath in just a moment, but before we do get to that, please take a moment to hear from the sponsors who so generously support this podcast. This episode is brought to you by BriefPoint, the discovery, governance, and automation platform used by more than a thousand law firms. BriefPoint just launched Discovery Playbooks.

Here's the idea. You respond to Discovery like you always do. Save it to your playbook in one click. And the next time a matching request shows up on any case, BriefPoint reapplies your objections and response strategy automatically. No data entry, no setup. Watch the 60-second demo at BriefPoint.ai. Spend more time on billable work, not repetitive administrative tasks. Bill for Time, the most trusted name in legal time and billing software, is designed to effortlessly streamline your firm. Automate your time tracking with multiple timers, standardize your invoicing with custom templates, and get paid up to 70% faster with built-in online payment processing. Discover how Bill for Time can help you work more efficiently by visiting BillForTime.com. That's Bill, the number four, time.com.

Now let's get to today's conversation. Welcome to LawNexus. This is Bob Ambroji, and I am here at the NetDocuments Inspire Conference in Orlando, Florida, I've been attending a couple of days here of programming largely oriented toward customers of the company. But now I have an opportunity to sit down with Heath Harris, Vice President, Applied AI and Enterprise Transformation. Heath, welcome to the show. Thank you so much for having me. It's a long title to have to memorize. It's a mouthful. What does it mean? Tell me what it means. Well, it's a wonderful, relatively invented title. of effectively where emerging technology and customer experience meets how we build products. Our mission is to be as close to the customer as possible and take all of that learning into stuff that makes our technology work better. So does that mean you're actually going out and sitting down with the customers and helping them? Absolutely. We're like a team of ninjas that basically wrap around customers,

whether they're corporate legal departments or law firms. and we effectively help them make sense of and create their AI strategies and then do some capability mapping to those AI strategies between what NetTalks brings to bear and what their business objectives are. We make a deliberate intent to actually map ourselves and what we do to their technology and be a part of their team because at the end of the day, we're only successful because they are. I know that you worked in innovation at Fenwick & West, I believe, and you worked at EY. Yes. Tell us a little bit about your background and what got you here to NetDocuments. Oh, yeah, absolutely. It's a winding road, but it's been a fun one. So I actually started out at what is now Fagery Drinker, running their risk department, and then was sort of at the helm of post-recession, we need to double the firm size over five years kind of stuff, and that got me into invariably technology and organizational design. And then Fenwick picked me up to effectively do that as kind of a first-of-its-kind position,

where I would be both an internal and external consultant to Fenwick clients and then help redesign practice groups to make them more profitable, generate new revenue streams, generally improve operations, things like that. And it was a pretty fortunate, very lucky time for me in terms of the types of clients that Fenwick had and what that created. No surprise. A lot of Silicon Valley are their clients, as well as a lot of the folks that fund Silicon Valley. So we kind of were put into this little de facto role of like a little bit of a legal tech incubator just for the VCs that happen to be firm clients. When was this? What years were you? This was 2000, late 2017, early 2018. All right. And I was with the great Jim McKenna, who is now at Foley. It was exciting. It was effectively the time of NLP. and RPA and only the firms could afford it type of thing. But how do you sort of bring that value to your customers and create new ways of working that, you know,

your customers actually feel? What that sort of built up to was, you know, the age of legal operations becoming sexy and cool. And a good friend of mine at Ernst & Young asked if I would join EY. And I ultimately got the opportunity to lead digital transformation and AI for EY Law Americas. So it was kind of growing up on the firm side, helping firms figure out things from an operations and client service perspective, and then going over to the UI side and helping, you know, the consumers of that find new ways to work with tech and, you know, expect more from their law firms. Yeah. So having worked in law firm environment before and working now closely with law firms, how's the landscape changed with all the changes in technology that have come along since then? This is exciting. I know that, you know, for decades, everybody has said that this change, this inflection point, it's the death of the billable hour and everything's going to change. I don't think the billable hour is going anywhere soon,

but I do think the way that we work is dramatically changing. And I think the difference between then and now, like if you compare the NLP moment to the generative AI moment, this is as ubiquitous as the Internet. There's not really a big cost barrier on the corporate legal side. And those corporations are being rewarded by the street for adopting it. So in my time at EY, these were CEOs giving directives to GCs saying, come up with use cases and I'm going to start evaluating on that. And so all of a sudden, general counsel were like at the seat of the table in informing AI strategy and kind of the gatekeepers to it to some degree. And what I've seen since is there's no escaping it. The technology itself is here. The technology itself is understandable and makes an impact. So I think what I'm seeing is a more earnest embrace from each side of the market. Yeah. There's no escaping it, but what are you actually seeing in terms of adoption?

I mean, obviously, I'm sure law firms are confronting the fact that there's no escaping it, But how are they doing in terms of actually making it a reality? Just like any change in law firms, it sort of starts with the groups that are, you know, forced to embrace it first kind of things in terms of the customers that they represent, so folks that are more exposed from, like, say, a financial services customer base or a pharma customer base or a big tech customer base, sort of leading the way in terms of what it looks like to adopt this technology. I think what I'm seeing that's different is we're going beyond the I'm just going to buy everything phase. We're into the how does this actually map to the strategy of my firm? How does this augment like what my brand is in the market? Like you'll still get your here's a cool press release of how much money that we're going to spend over the next several years or what have you. But what I look for are who are they hiring and then what are they actually doing with their customers from a practice perspective.

And I think what I'm seeing more of is they're hiring people that look kind of different than you would ordinarily see in innovation or AI roles. People that are more technical and business mixed. And then those people having kind of a seat at the table in terms of where directions actually head. And that coupled with kind of embracing a broader, bigger data strategy is something different this time around for me. Yeah. You're working primarily with kind of the larger firm into the market or larger corporate legal departments. Is that fair to say? I do work with some mid firms as well And my teams do support every segment of our customer base So down from small to midsize to large firms Yeah In your work with customers are you primarily focused on the company you work for or are you helping them kind of think about the full picture of how they can be using AI in there? Yeah. The most durable thing we have is trust. And so if I'm thinking

only about what NetDocs brings to the table, I'm missing half of the field. And so I mean, I was that guy at Tenwick, right? Yeah, right. Sell me all your stuff. So like people see through that. I try to focus on outcomes just as sort of the ethos of our team. Like what does this organization have to achieve and how can we help them do that? Sometimes NetDocs will be a big part of that story. Sometimes we'll be one of the players. I just think we should be there. Yeah, yeah. Do they have a fairly well-evolved sense of what it is they want to achieve with AI or are they looking to you to help them think that through and guide them? Both. I would say like that we've been pretty deliberate about the kinds of people that we've hired on our team as well. I like to say that the folks that make up our team are the folks that both had to design strategies and then live with those decisions in real life after they made them. So it's a little bit different from I just have a bunch of educated people that are going to swarm around you. Um, so we often find ourselves in a different level of conversation, um, in terms of like

how to navigate the executive committee, how to navigate the practice group leadership and how to sort of shape and stage things. And in terms of, you know, what may be pressing now versus what means more longer term. Um, and that's the stuff that we really get value out of, um, in terms of like, at the end of the day, you have an opportunity to help make someone's career better. So this is big, scary change. What do I do? If we're going to meet that moment, we shouldn't just be a vendor. We should be someone that has to actually walk beside them and help them figure that out. Obviously, there's a ton of smart people in our industry, a ton of very experienced people in our industry. But I would say, like, overarchingly, the sophistication and comfort is mixed. A lot of what we're doing is just basically giving other professionals tools to sort of shape that together. And that's what we're after. Yeah. I was surprised to hear you say earlier, I think you commented, that they're no longer just buying everything. They're kind of becoming more focused on the products they're thinking about. There was just some story that popped up in my feed this morning about another major firm buying both Harvey and Lagora at the same time.

And we're seeing a lot of that too, I think. So I guess I'm kind of wondering how you see – now we've got – obviously we've got the frontier models all coming into legal. We've got Harvey and Legora. We've got co-counsel. We've got any number of more sort of point solutions in a sense, I guess, for AI products. How do you see that all playing out? Not to mention the documents, of course. No, yeah. No, I tend to think that the business model wins ultimately. So I always ask what happens in year two when the partners get to vote again in terms of where that budget is going to be allocated. So the way that I see it is law firms are in the business of making partners that can sell more legal services. And they can do that by virtue of building them from junior associates up or they can buy them in the open market. And I think as the downward pressure that is inescapable from like a cost perspective that I am seeing across most firms from their corporate clients continues, how do you grow?

And that's typically by virtue of lateral acquisition, which is already kind of a redwater market. And so I tend to think that will increase in intensity. And so if we're in year two and we have previously allocated hundreds of millions in that year to go towards those tech infrastructure projects or what have you, but you're seeing a decrease in billables and losing ground in wallet share, The question then becomes, am I going to spend $20, $30 million on advanced access to GPUs? Or am I going to spend $20, $30 million on someone who has a book of business at Microsoft? And having grown up in these worlds and law firms still being a cash business, I tend to think that the sort of economic barriers will win out a little bit. I'm super encouraged to see a trend of firms spinning out subsidiaries that will be AI first legal organizations. I will equally look for who gets to be the more lucrative business and how those gains are paid back to the partners and who's the winner in that partnership, for lack of a better way to describe it.

I tend to think it will normalize in the sense of I'm in the business of providing superior, pristine legal advice at the top of the food chain. And I shouldn't be in the business and trying to keep pace with the frontier models, both from a technical perspective and a talent perspective. Because then, like, if you look at it, I'm going to go hire a rainmaker for, call it 15 million a year. 10 to 15 is what you're paying someone in the valley to go be your leader of AI. That I haven't seen yet in terms of that kind of a different looking organizational structure. Yeah. One of the ideas I've heard thrown around is that as firms do look at making those significant lateral hires, besides throwing money at them, they want to be able to throw the right tech at them. And so they need to make sure that they have not just whatever tech they might already have had in place, but whatever tech that lateral might be used to using and want to have there.

Absolutely. So, like, this was going back to my Fenwick days, the chairman's mantra. Like, we need to compete in this marketplace, and we need to be a platform to build your career on. I think now that platform is not just the customer base that you have access to and the partners that you get to work with, but equally the tools that you're provided to go compete in that market. And the intensity of that competition still takes me back to the same build or buy calculus of what keeps me at the top end of flexibility in technical platform from a delivery perspective. Well, I can make the most out of my competitive dollars to grow the business that I'm in. Yeah. You alluded to firms spitting out AI first practices. Have you worked on any of those? Have you worked in any sort of AI-first law practices or law firms? I can say unofficially I have worked with captives that the aim is to basically build that technological infrastructure that can be sold, resold, sort of in a federated model.

Yeah. Yeah. And do you think that model is – is it something that can have a real impact or is it marketing hype? So I'm trying not to be jaded. But a press release is a press release at the end of the day. And law firms are still going to compete with each other at the end of the day. I do think down market, you'll have firms that are willing to consume infrastructure from a trusted brand. But how durable that is when that trusted brand is another reseller themselves is a question for me. Yeah. So I, as far as impact goes, I think the impact when you're in the sort of the spinoff environment or the AI first firm on its own, that's maybe has has investment from VCs or what have you. You're sort of capped at how much you can, how far you can travel up the food chain from like the value pyramid perspective and legal services.

I think what I'm looking for is the first AI firm, AI first firm that cracks to AMLA 100 based on revenue. Right. And I think they can. Yeah. I don't, at EY, there are plenty of corporations that are, you know, willing to contract that this machine and this person supervised by my AGC. Yeah. Can absolutely process my contracts for me. Right. And to get to a couple hundred million in AIRR in that context of the Fortune 500 is, I think, achievable. Yeah. Raising some big money. I mean, they seem to be more focused on sort of the commoditized aspects of legal practice. That's honestly remains to be seen. I'm not a fortune teller either. But the nature of the purchase as you go up the value pyramid is just as much insurance as it is I need advice. Yeah. and equally so like true global footprint Like if you thinking about like a you know bet the company multi acquisition

like, where you actually do need boots on the ground cross-jurisdictionally to make that happen, there's only but so many organizations that can actually pull that off from, like, a what's needed to make it happen kind of perspective. So I do think there, at least so far, is a little bit of a natural ceiling. for how far they can travel. But, you know, we can hope that one day the ABA becomes enlightened and allows for a different investment vehicle in law like we see in other countries. But I do think there's a cap. You're listening to a conversation with Heath Harris, Vice President, Applied AI and Enterprise Transformation at NetDocuments, which was recorded live at the Companies Inspire Conference in Orlando in October. We've got lots more to cover, but first, please take a moment to learn about the sponsors who support this show.

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. Capture lost revenue by accurately tracking time anytime, anyplace. Bill for Time is the most trusted name in legal time and billing software, helping firms of all sizes streamline their processes. Bill for Time allows you to easily track time, automate billing, and process online payments, all from one centralized location.

Visit BillForTime.com today to learn how you can take your practice to the next level. That's Bill, the number four, Time.com. Welcome back. This is Bob Ambroji. I am live at the NetDocuments Inspire Conference in Orlando, Florida, where earlier today I recorded a conversation with Keith Harris, Vice President of Applied AI and Enterprise Transformation at NetDocuments. Let's get back to that conversation. Keith, I wanted to ask you about, I came across this article you wrote just in August on your taxonomy is your AI strategy. What do you mean by that? Explain that. What I mean by that is your brand lives inside of the work product that you create. So as a firm, your decades of experience, your secret sauce lives in the product that you created inside of a document that lives inside of a system.

And until recently, that was easier said than done of turning all that unstructured mess into something that is honestly like an organized data model that not only you can access, but your technology can access. So if I'm going to get to this, you know, panacea of a defined ontology of my organization and a layer of agent workforces that support my associates and partners and deliver more and more value for my customers, you have to start with that data foundation. And to get to ontology, start with taxonomy. That's the backbone of what informs it, right? Yeah. And so about a year back, I sat down with a very good friend of mine, Mark Ford, who's former CKO for Clipper Chance, Baker McKenzie, and he was EY Law's global knowledge partner. And he's a like an actual taxonomist. He built the taxonomies for those three firms, respectively. And we just said, well, hey, shoot. But, you know, I've still not seen anybody like have something out there that is both useful and will be used and can be consumed by both humans and agents.

And so that's the little mission we started on. How could we take what are the leading good standards that everybody recognizes as global standards between Sally, Foley, and Oz and sort of simplify that to what reflected our own experiences building knowledge, knowledge organizations, knowledge systems? Yeah. And so that's how that started. And so the reason I think the taxonomy is your strategy is the what you know is why people buy you. And this is making that repeatable and scalable. Yeah. And can you kind of connect the dots a little bit more between how that taxonomy matters to the effectiveness of AI within a long time? Oh, absolutely. So let's just say you don't, you're just like every other firm, you're bribing your associates over the summer to profile documents to go up in the chamber's rankings. And it's sort of an uneven application of what you know inside of your document management system or wherever your work happens to live.

If that is the current state and you go into ChatGPT or Claude and you say, you know, when was the last time I did a reverse triangular mortgage with a bio-pharmaceutical company in the Nordics where the deal was valued at 250 million euros or greater and this partner was associated with it? Show me the stock purchase agreement. If you don't have a taxonomy, ChatGPT's agent will be delighted to read every document it can find to try to give you that answer. But if you do have that taxonomy, that agent then knows, oh, you're looking for a stock purchase agreement, which is a part of the corporate practice. So I'm going to start with corporate practice. I'm going to go a stock purchase agreement. You're looking for a client that is in industry of biopharmaceuticals and a resolved deal value that is greater than or equal to 250 million euros. So instead of having to read millions of documents, I can read maybe a few hundred. And then the difference in that is how much does it cost to get you to, like, you're actually giving legal advice and coming up with a customer strategy.

So I know that, what was it, August or so, NetDocuments published a sort of best set of benchmarks around the context graph and how much it could save you in terms of getting to an AI answer. Were you involved in that at all? I was involved in the sense of admiring it. I mean, I'm wondering how much of that done in sort of, you know, an ideal environment. We select the documents. Oh. How much of that translates to real life? I would say one thing that struck me, a couple things that struck me about NetDocuments when I was deciding, can I keep going in the big four or not? I never met a customer that didn't like them. That's hard to do. Yeah. For any legal tech leader. And that says something to me about how they treat their customers and their people. Because typically happy people treat their customers the same way. But also just like the value of minds that are here is pretty incredible.

Like Scott Kelly's nuts. He's just incredible. And there's nothing he puts his name on that he didn't do himself. Yeah. Like four in the morning. Yeah. Kind of guy. And this organization has the courage to do things right. And I think that's what the discipline that was brought to these evals. So this wasn't synthetically generated legal documents that, you know, cook the books to give me my 50 percent savings and 100 percent accuracy. And, you know, anybody can make their own eval. This was I'm going to take publicly available transactions or pieces of litigation, and I'm going to come up with question and answer pairs that are representative of you would have to be pretty steeped in this matter to be able to give this a quality answer. And we put that up against real lawyers that would have had knowledge of those matters. Like one of the matters that we actually used was a case that Dan Houck himself worked on to evaluate it. So real data, real mess, real lawyers validating.

I would ask that question, and this looks like it's a good answer to me, to then come up with the rubric for, you know, did I actually get the model that's at the economy level to perform at a higher capability? Or did I actually get the model itself to require less consumption to get to that answer? And the math, maths. And why that particularly exciting for me is you know you doing something right when somebody copies it And so you know lo and behold here comes the next context graph Lo and behold here comes a strikingly similar eval basing their analysis off of the same sort of strategy and approach. And so I think it's real. And this is probably the most readily understandable thing that I get to talk about across any size organization. It just logically computes that if you have this stuff, it's probably going to work better. Yeah. It's something that it seems that nobody can talk enough about these days is MCP. And

especially as, you know, the frontier models are now all identified, you know, some legal component of what they're working on and what they're building. And I'm wondering, again, in your work with law firms and legal organizations, are you seeing them kind of embracing that? Are there concerns about, you know, the shadow DMS people talk about or about governance issues or what are you hearing on that? Governance is probably one of the top three consistent themes that I have in conversation with CIOs and chief AI officers and chief innovation officers across our customer base. Yeah. That is very real. And I think that is sort of the gift that we have in this ecosystem that we can use the customer as a North Star for what's required for the moment, particularly as, like, the problem itself will only be larger in the space that it covers. So I think people are learning, you know, what MCP really means

and what it can and can't do in the context of their daily work and what and how you build responsibly to have deeper levels of integration without creating sort of the breakage in the governance model. and I have refreshingly found the experience when we're working with the ecosystem for people to kind of be grateful of, like, that's kind of how we started. Like, this was the plucky guys that decided to go in the cloud before anybody else did. So we kind of had to learn all those lessons pretty early on, and it's kind of baked into how we build anyway. So we add some value there. Yeah. How do you think this evolves? I think it plays out in terms of, again, so many players from the platforms sort of competing for lawyers' attention and eyes. I tend to think how this plays out is the data wins and that the models themselves, the harness, whatever is wrapping around that is very changeable.

That's how I think this plays out personally and that everybody finds a way to effectively make it as frictionless as possible to get to what you need to get to, to do the actual work you're trying to do. I think it could be a little bit foolhardy to try to rule them all. And at the end of the day, I just need some basic capabilities. I need to be able to search and find. I need to be able to do some workflow. and I need to be able to do creating work product itself. Yeah. So if the data wins, then it all comes back to the DMS. Maybe. You never know. But we're else going to go. We're else going to go. I certainly hope so. I think we're just going to. Should the concept of a DMS become outmoded at some point? Or, I mean, you know, obviously it's an evolving concept, clearly. What we think of a DMS today is not what we thought of a DMS five years ago. I don't think it can. Well, we'll see. Never say never to anything. But one of the thrills of my career was happening to be quoted next to other AI companies for building an app.

Yeah. And it was a judge analytics app. And the quote was, where's the moat if a DMS company can build that? My thesis coming in here was if I turn this into a structured data model, what can't I do? Yeah. At the end of the day. And I do think that will be a direction of travel of, you know, the tools are easy, easily accessible. And the hard part is, you know, really organizing the brain. Yeah. Again, not asking other names at all. But, again, in your work with different kinds of law firms and different kinds of legal organizations, as you work with them today, are there obvious areas that stand out for you that they need to be focusing on more to prepare for not what's available today but to prepare for what things are going to be in the next five years? I mean, what should firms be doing today to make sure they are in the right shape in terms of technology and training and, you know, adoption and whatever else to be in a strong position as this technology continues to evolve?

I think I'm seeing a lot of people do the right thing first, which is focus on the data foundation. I think take that as serious as it is in terms of what that creates, in terms of durable competitive advantage that compounds over time. What I also hope that they're doing is hiring people that didn't grow up in law firms, that have gone through sort of business model transformation and how that informs how you shape a product and how you price a product and how you sell a product. and to give that person more than a press release and to give that person a seat at the table alongside the chief AI officer or innovation officer that should have decision rights and not just they get to be in the room kind of thing. That's what I think people need to be doing sooner rather than later as opposed to more sort of reactionary strategies. I'll just spin off an AI first law firm or just go buy a rack of GPU

or what have you, I think this is meaningful change of what I'm willing to pay for and what creates that value. And I think you need someone that deliberately didn't grow up in that space to help inform what that looks like. And that sort of takes me back to the, if I have that $15 million to spend on a lateral or a data scientist, where is that going to go? So I sincerely hope that over the next two years that those are the next press releases we see that we get to say, will they do it? Yeah. Well, I've been asking you a bunch of questions. Is there anything else top of mind that you wanted to talk about or point out? If you'll indulge me, I've been reading you for a while. I'd love to hear where you think it's going and what you think. I'm just a reporter in the room. You're the expert. You've got transformation in your title. I mean, yeah. What excites you, though?

I mean, you know, everything. Yeah, I've been doing this for a long time, and there has never been a more exciting time in the whole legal technology world. I sometimes wish I was just starting out rather than at the tail end of my career right now because it's like I want to be around for the next 30 years and see what happens. It's just crazy. It's really exciting. It's crazy. I mean, you know, we were talking earlier. I am really fascinated by the MCP, sort of the sudden ubiquity of MCP and how many – this came up at a press briefing we had this morning. I was talking about how many lawyers I now talk to who are starting their day in clot. And I just think there's like this whole upheaval of the way lawyers even think about approaching their work. When people say this is all going to change law practice, it's already changed law practice. That's the thing. And I think people are – it's happening in a way so subtly and so quickly that people are almost not noticing it. But it's changing everything about how you design products, how lawyers use products, engage with products.

And so it's an interesting time. Little things and all at once. Yeah, yeah. That's super cool. I personally love how it's creating more time to focus on, like, actual humans. Yeah. Yeah. So the session that you sat in on, like that's the why I do it. Yeah. They get to get up there and tell their stories, but like you get to actually connect with people and you're all kind of in the same moment together. And that's pretty cool. Yeah, that is cool. Well, Heath, thanks a lot for talking to me. I really appreciate your taking time out of a busy conference to sit down and have a conversation. Of course. Lovely to speak with you. Yeah. Yeah. And speaking with Heath Harris, the vice president of applied and enterprise transformation here at Net Documents. We're at the Net Documents Inspire Conference in Orlando. Hope you enjoyed today's show. You can always find us on any podcast platform you use or on YouTube. This is Bob Ambroji. Thanks for being with us today.

Thank you.

番組の概要欄(原文)

Law firms are moving past the "buy everything" phase of AI adoption and starting to ask how the technology maps to their strategy and brand, says Heath Harris, vice president of applied AI and enterprise transformation at NetDocuments, in this episode of LawNext, recorded live at the company's Inspire conference in Orlando last month. But Harris, who previously worked in innovation at Fenwick & West and led digital transformation and AI for EY Law Americas, expects law firm economics to test that commitment. The real question, he says, is what happens in year two, when partners vote on whether to spend $15 million on AI talent and infrastructure or on a lateral with a book of business. Harris also explains why he believes "your taxonomy is your AI strategy," how a structured data foundation lets an AI agent read a few hundred documents instead of millions, and why he sees a natural ceiling for AI-first law firms, even as he watches for the first one to crack the Am Law 100. He discusses the governance concerns firms raise about MCP and what they should do now to prepare, including hiring people from outside law firms and giving them real decision rights. As for where the market is headed, Harris says the data wins. "The tools are easily accessible and the hard part is really organizing the brain." This is the third and final LawNext interview from Inspire. Listen also to the conversations with NetDocuments Chief Product Officer Dan Hauck and CEO Josh Baxter. Thank You To Our Sponsors This episode of LawNext is generously made possible by our sponsors. We appreciate their support and hope you will check them out. Paradigm, home to the practice management platforms PracticePanther, Bill4Time, MerusCase and LollyLaw; the e-payments platform Headnote; and the legal accounting software TrustBooks. Briefpoint, eliminating routine discovery response and request drafting tasks so you can focus on drafting what matters (or just make it home for dinner). Ajax, the AI timekeeper lawyers want to use.   If you enjoy listening to LawNext, please leave us a review wherever you listen to podcasts.

出典・この記事について

番組
LawNext(Populus Radio, Robert Ambrogi)
元の回
NetDocuments' Heath Harris on Why Your Taxonomy Is Your AI Strategy and Why, in the End, the Data Wins
配信日
要約作成日
公式サイト
lawnext.com

この要約・解説記事は、元の音声を文字起こしした内容をもとに AI(Claude)が日本語で作成したものです。 固有名詞や数値に誤りを含む場合があるため、正確な内容は元の音声でご確認ください。制作方法について

X でシェア

関連エピソード