为律师构建 AI - Legora 首席技术官 Jacob Lauritzen — Latent Space

本期节目中，我们采访了 Legora 的首席技术官 Jacob Lauritzen。就在几周前，该公司宣布完成 5.5 亿美元的 E 轮融资，估值达到 88 亿美元。我们聊了聊，为律师打造 AI 智能体需要具备哪些条件，以及其他正在构建智能体的人可以从这项工作中学到什么。In this episode, we chat with Jacob Lauritzen, CTO of Legora, following the announcement of their $550M Series E at an $8.8B valuation a couple weeks ago. We talk about what it takes to build AI agents for lawyers and what other people building agents can learn from that work.

Legora 始于斯德哥尔摩一间地下室，最初的产品很简单：ChatGPT、瑞典法律和一款搜索工具。Jacob 曾带着自己的创业公司参加过 YC，之后帮助 Max Junestrand 申请 Leya（现已更名为 Legora）。后来，他加入公司担任工程负责人，如今是 CTO。Legora started in a Stockholm basement with a simple product: ChatGPT, Swedish law, and a search tool. Jacob had already been through YC with his own startup when he helped Max Junestrand with the application for Leya (now Legora). He later joined as head of engineering and is now CTO.

如今，这款智能体可以起草和编辑 Word 文档，一次审阅数千份合同，并协调律师、客户及其他智能体之间的工作。公司在今年4月实现 ARR 超过 1 亿美元，距离上线还不到 18 个月，如今客户已超过 1,800 家。公司于3月融资 5.5 亿美元，估值达 55.5 亿美元；随后又于10月融资 5.5 亿美元，估值达到 88 亿美元。最新收购的 Wexler，是公司仅今年就完成的第五笔收购。Today the agent drafts and edits Word documents, reviews thousands of contracts at once, and coordinates work between lawyers, their clients, and other agents. The company passed $100 million in ARR in April, less than 18 months after launch, and now has more than 1,800 customers. It raised $550 million in March at a $5.55 billion valuation, then another $550 million in October at an $8.8 billion valuation. Its latest acquisition, Wexler, was its fifth just this year.

我们向 Jacob 请教了：当你无法查看客户数据时，如何评估智能体；哪些任务仍需要更好的模型；法律工程师和前线部署工程师如何分工；以及为什么按小时计费的商业模式正在被颠覆。We asked Jacob how you evaluate agents when you can't see your customers' data, which tasks still need better models, how legal engineers and forward deployed engineers split the work, and why the billable hour as a business model is being upended.

在 YouTube 上观看完整对话 →Watch the full conversation on YouTube →

计费小时发生了什么变化？What changed for the billable hour?

雅各布第一次读莱娅的 YC 申请时，只给了一条反馈：人们已经尝试开发法律科技产品无数次了，但从来没有成功过。如果律师按小时收费，效率提高就意味着收入减少。他们为什么要买一个会让自己收入减少的工具呢？When Jacob first read Leya's YC application, he had one piece of feedback: people had tried building legal tech a million times, and it never worked. If lawyers bill by the hour, getting faster means earning less. Why would they buy a tool that does that?

他说，变化在于企业面临的压力。在美国，像小型审查这样的常规工作，已经开始改按固定项目费用收费。有些律所甚至免费提供这类服务，以留住客户。收入则转向并购、诉讼，以及与客户相处所花的时间。What changed, he says, is the pressure on firms. In the US, routine work like small reviews is already shifting to flat project fees. Some firms do it for free to hold on to the client. The money moves to M&A, litigation, and time spent with clients.

我们问，这是否意味着律所将不再需要律师助理和初级律师。Jacob表示，Legora看到的情况并非如此。如今，一名优秀的初级律师比两年前更有价值，因为他们能完成的工作多得多，工程领域也是如此。We asked whether that means firms will stop needing paralegals and junior lawyers. Jacob says that's not what Legora is seeing. A good junior is worth more today than two years ago because they get so much more done, just as in engineering.

跳转到计费模型讨论 →Jump to the billing model discussion →

无法访问客户数据的评估Evals without access to customer data

大多数团队通过阅读追踪记录来改进智能体。Legora 却一条也不能读。用户上传的几乎所有内容都属于机密，其中很大一部分属于律所的客户，而非律所本身。Most teams improve their agents by reading traces. Legora can't read any. Almost everything users upload is confidential, and much of it belongs to the firm's clients, not the firm.

因此，评测团队必须重现客户所做的各类工作。三名工程师和三名律师会构建强化学习环境，模拟客户的工作。其中一个例子是一个包含 10,000 个文件的数据室，智能体要在其中开展尽职调查，并生成一份能够识别出恰当风险点的报告。团队会按业务领域和任务将结果绘制成热力图。新模型往往在某些类型的法律工作上表现更好，在另一些类型上则更差。So the evals team has to recreate the kinds of work customers do. Three engineers and three lawyers build RL environments to simulate the work their customers do. One example is a 10,000-file data room where the agent runs due diligence and has to produce a report that catches the right red flags. The team plots the results on a heat map by practice area and task. A new model often gets better at some kinds of legal work and worse at others.

团队的人才库中约有 100 名法律工程师：他们是曾在大型律所工作三四年的律师，通常还拥有计算机科学学位。在生产环境中，Legora 会将故障归类为 10 到 20 种已知故障模式，并在功能开关控制下运行成本低廉的基于 LLM 的检查。The team draws on a pool of about 100 legal engineers: lawyers who spent three or four years in big law, often with a CS degree. In production, Legora classifies failures into 10 to 20 known failure modes and runs cheap LLM-based checks behind feature flags.

这些限制也决定了 Legora 能使用哪些基础设施。所有服务都运行在 Azure 上，因为 Legora 创立时，律师事务所不接受 AWS 或 GCP，而微软已通过 365 成为获批供应商。Those restrictions also determine which infrastructure Legora can use. Everything runs on Azure because when Legora started, law firms wouldn't accept AWS or GCP, and Microsoft was already an approved vendor through 365.

跳转到不含追踪记录的评估 →Jump to evals without traces →

更智能的模型仍然发挥重要作用的场景Where smarter models still matter

Legora 长期以来一直抵制加入模型选择器，后来才妥协。Jacob 认为，律师不应该还得判断一份合同属于 Sonnet-high 还是 Opus-mid 级别的任务。Legora 的职责是，在质量、成本、延迟、治理要求和区域可用性之间，找到当前最佳的平衡。Legora resisted a model picker for a long time before giving in. Jacob's view is that a lawyer shouldn't have to decide whether a contract is a Sonnet-high or an Opus-mid task. It's Legora's job to find the best available tradeoff between quality, cost, latency, governance, and regional availability.

他估计，模型目前已经能非常出色地处理约 60% 的法律工作。对于这些任务，如今的问题在于成本和延迟。对百万 token 级合同进行复杂分析和编辑，仍然需要更强的智能能力。He estimates that models already handle about 60% of legal work really well. For those tasks it's a cost and latency problem now. Complex analysis and editing across million-token contracts still need more intelligence.

Legora 已开始针对法律研究等狭窄任务，对开放权重模型进行后训练，但没有训练自有前沿模型的计划。对于起草工作，Jacob 认为 Fable 和 Opus 过于冗长，而 Astra 更简洁。为解决幻觉问题，智能体写下的所有内容都会附上引用，并在源 PDF 中高亮显示对应的边界框，即使该内容位于文档深处、需要深入数层才能找到。Legora has started post-training open-weight models for narrow tasks like legal research, with no plans to train its own frontier model. For drafting, Jacob finds Fable and Opus too verbose. Astra is more concise. To address hallucinations, everything the agent writes gets a citation that highlights a bounding box in the source PDF, multiple layers down.

跳转至模型讨论 →Jump to the model discussion →

少数基础模块，众多产品A few building blocks, many products

入职六个月时，Jacob给团队看了一张300名斯巴达勇士的幻灯片，并宣布Legora将永远只维持20名工程师。如今团队已有100名工程师，而他说这仍然感觉太少了。他现在的工作，是打造一个系统，让团队只需交付少数几项功能，客户却觉得收获颇丰。Six months into the job, Jacob showed the team a slide of the 300 Spartans and declared that Legora would stay at 20 engineers forever. It now has 100, and he says that still feels too small. His job now is to build a system where the team can ship just a few things and customers feel like they're getting a lot.

团队围绕几个基础构件构建：用于存放文件的项目、智能体、表格审查和列表。列表是一种灵活的 Notion 风格数据库，支持触发器和自动化。智能体可以创建和读取列表，而列表发生更改时也可以启动智能体。Intake 是一个 Slack 机器人，可以将请求转化为列表中的工单。The team builds around a few primitives: projects that hold files, the agent, Tabular Review, and lists. Lists are a flexible Notion-style database with triggers and automations. Agents can create and read lists, and changes to a list can start an agent. Intake is a Slack bot that turns a request into a ticket on a list.

表格审查是业内竞相模仿的模式，连名字都照搬了。每一行是一份文档，每一列是一个问题。导入 10,000 份雇佣合同，询问合同终止期限、知识产权转让条款和遣散费条款。它最多支持 100 万个单元格，但一开始只处理 10 行，这样你就能先调整提示词，再运行剩余内容。Tabular Review is the one the industry copied, down to the name. Each row is a document, each column is a question. Drop in 10,000 employment contracts and ask for the termination period, the IP assignment clause, and the severance terms. It supports up to a million cells, but starts with just 10 rows so you can fix your prompts before running the rest.

法律工程师无需编写代码，就能基于这些基础构件配置新产品。甚至 Wexler 的事实情报产品也运行在同一套构件之上。Legal engineers configure new products on these primitives without writing code. Even Wexler's fact intelligence product runs on the same building blocks.

跳转到原语讨论 →Jump to the primitives discussion →

法律工程师优先，其次是前沿部署工程师Legal engineers first, forward deployed engineers second

大约九个月前，Legora 在纽约聘用了首位前沿部署工程师：Di Qi。他曾两次创办 YC 初创公司，并在 Meta 从事搜索相关工作。法律工程师会像高级用户一样配置产品，但不编写代码。前沿部署工程师则会编写代码并部署到生产环境，为律所的 CRM 或文档管理系统构建集成，并将他们的经验反馈到核心产品中。他们仍然不会接触客户数据。Legora hired its first forward deployed engineer about nine months ago in New York: Di Qi, a two-time YC founder who worked on search at Meta. Legal engineers configure the product the way a power user would and don't write code. Forward deployed engineers write code into production, build integrations into a firm's CRM or document management system, and push what they learn back into the core product. They still don't touch client data.

当一个规模较大、参与度高的客户签约后，Legora 会安排法律工程师与其合作；如果项目需要，还会加入一名 FDE。Jacob 预计，一年内会组建一支轮换的 FDE 团队，每次在客户那里驻场几个月，为其打造所需的解决方案，然后再转到下一个项目。When a big, engaged client signs, Legora assigns legal engineers to work with them and adds an FDE if the project needs one. Within a year, Jacob expects to have a rotating group of FDEs who sit with a client for a couple of months, build what they need, and move on.

跳转到前沿部署讨论 →Jump to the forward deployed discussion →

这会放在哪里Where this goes

雅各布认为，两年后，大型企业会通过 Claude Cowork 之类的工具使用 Legora，而律师事务所则会直接在 Legora 中工作。他不认为这些实验室会像对待编程公司那样封杀 Legora。对于一笔 4 亿美元的并购交易，他认为人们仍会想要使用 Legora。In two years, Jacob thinks big enterprises will use Legora through tools like Claude Cowork, while law firms work inside Legora directly. He doesn't expect the labs to cut Legora off the way they've done to coding companies. For a $400 million M&A deal, he thinks people will still want Legora.

跳转至预测结果 →Jump to the predictions →

节目说明Show notes

• Legora——面向律师的协作式 AI 平台，Jacob 在该公司担任 CTO。

  Legora — the collaborative AI platform for lawyers, where Jacob is CTO.

• Jacob Lauritzen — Jacob 在 LinkedIn 上。

  Jacob Lauritzen — Jacob on LinkedIn.

• Legora宣布完成E轮融资，融资额为5.5亿美元，估值达88亿美元，ARR超过2亿美元。

  Legora's Series E announcement — $550 million at an $8.8 billion valuation, with ARR above $200 million.

• “智能体需要的不只是聊天”——Jacob 在 2026 年欧洲 AI 工程师大会上的演讲。

  "Agents need more than a chat" — Jacob's talk at AI Engineer Europe 2026.

• “将 AI 接入数十亿份法律文档”——Jacob 与 turbopuffer 的 Simon Eskildsen 畅谈 AI 工程。

  "Connect AI to Billions of Legal Documents" — Jacob's AI Engineer talk with Simon Eskildsen of turbopuffer.

• Legora 如何在 Modal Sandboxes 上并行运行 2,000 个智能体评测——支撑 Legora 评测任务的基础设施。

  How Legora runs 2,000 parallel agent evals on Modal Sandboxes — the infrastructure behind Legora's eval runs.

• Legora 收购 Qura——这家斯德哥尔摩法律研究初创公司于 4 月被 Legora 收购。

  Legora acquires Qura — the Stockholm legal research startup Legora bought in April.

• Legora 正在收购 Wexler——这家伦敦初创公司专注于从大量诉讼文件中提取并核实事实。

  Legora is acquiring Wexler — the London startup that extracts and verifies facts across large litigation document sets.

• Claude for Legal——Anthropic 的法律连接器和插件，为“各大实验室是否会进军法律领域”这一问题提供了背景。

  Claude for Legal — Anthropic's legal connectors and plugins, the backdrop to whether the labs will come for legal.

• AIE 欧洲复盘 + Agent Labs 核心观点——Swyx 谈他在开场时提到的 agent lab 操作手册。

  AIE Europe Debrief + Agent Labs Thesis — Swyx on the agent lab playbook he brings up at the start.

时间戳Timestamps

• 0:00 冷开场 • 0:40 认识 Legora 首席技术官 Jacob Lauritzen • 1:05 从斯德哥尔摩地下室里的弱势新秀起步 • 1:54 Legora 如何从面向瑞典法律的 ChatGPT 成长为法律智能体 • 4:12 从助手转向智能体 • 5:23 Legora 使用哪些模型，以及为什么加入模型选择器 • 7:48 现在的模型对律师来说够用了吗？ • 8:20 智能体九个月前还做不到的事 • 9:18 在模拟环境和包含 10,000 个文件的数据室中测试智能体 • 10:07 评测团队与 Legora 的 100 名法律工程师 • 11:03 评测曾经看似不值得投入，以及一次发布如何将引文准确率从 70% 提高到 99% • 12:18 五到十年后，法律实践可能会是什么样子 • 14:18 律所合伙制模式能否延续？ • 16:40 辅助律师和初级律师将何去何从？ • 18:00 计时收费制，以及早期法律科技为何未能奏效 • 19:39 现在钱都花到哪里去了 • 22:44 Legora 如何用自家产品进行收购尽职调查 • 23:59 瑞典律所与美国律所有何不同 • 26:12 无法查看追踪记录时，如何衡量产品指标 • 27:41 对追踪记录进行转述，并将故障归纳为 10 或 20 种类型 • 28:42 为什么数据隐私促使 Legora 把所有服务都运行在 Azure 上 • 30:04 开放权重模型与模型套利 • 30:55 转录模型，以及 Legora 为什么不用 ElevenLabs • 33:33 智能体、Tabular Review、项目和 Word 加载项 • 34:24 律师如何使用 Tabular Review • 35:51 为什么电子表格行之有效，以及从前 10 行能学到什么 • 37:53 Slack 中的法律需求接入机器人如何运作 • 39:11 智能体如何创建列表，而列表又如何启动智能体 • 40:15 基于这些相同组件构建产品并收购 Wexler • 42:37 Legora 的前线部署工程师负责什么 • 43:52 法律工程师和前线部署工程师如何帮助大型客户完成入门 • 46:43 未来一年前线部署团队的工作重点 • 47:24 Legora 为什么一直不愿开发 API • 49:26 智能体能成为 API 吗？ • 50:12 “300 名斯巴达人”幻灯片，以及将工程师团队维持在 20 人的计划 • 51:32 Legora 主动押注、却没人要求它去做的事 • 53:05 针对法律研究对开放权重模型进行后训练 • 54:05 搜索、检索与晚期交互 • 54:51 优秀的法律智能体工具框架需要具备什么，以及为什么起草取决于搜索 • 56:25 哪个模型写出的草稿最好？ • 57:03 幻觉与 Legora 的引文系统 • 57:41 律师会核查引文吗？ • 58:19 百万 token 的合同该用 MapReduce，还是一个大型上下文窗口？ • 1:00:16 获取匿名化案例，以及为数据收购一家律所的想法 • 1:01:53 Jacob 为什么建议不要默认使用嵌入 • 1:02:44 律师会在 Legora 中工作，还是通过 Claude 工作？ • 1:03:26 模型实验室会切断 Legora 的服务吗？ • 1:04:24 法学院与面向计算机科学专业学生的奖学金 • 1:05:28 在纽约、斯德哥尔摩和伦敦招聘 • 1:06:09 为什么大多数工作最终都会变成法律工作• 0:00 Cold open
• 0:40 Meet Jacob Lauritzen, CTO of Legora
• 1:05 Starting out as underdogs in a Stockholm basement
• 1:54 How Legora grew from ChatGPT with Swedish law into a legal agent
• 4:12 The switch from assistant to agent
• 5:23 Which models Legora uses and why it added a model picker
• 7:48 Are today’s models good enough for lawyers?
• 8:20 What agents couldn’t do nine months ago
• 9:18 Testing the agent in simulated worlds and a 10,000-file data room
• 10:07 The evals team and Legora’s 100 legal engineers
• 11:03 When evals didn’t seem worth it and one release took citations from 70% to 99%
• 12:18 What legal practice could look like in five to ten years
• 14:18 Will the law firm partnership model survive?
• 16:40 What happens to paralegals and junior lawyers?
• 18:00 The billable hour and why earlier legal tech fell short
• 19:39 Where the money goes now
• 22:44 How Legora used its own product for acquisition diligence
• 23:59 How Swedish and American law firms differ
• 26:12 Product metrics when you can’t read the traces
• 27:41 Paraphrasing traces and narrowing failures down to 10 or 20 types
• 28:42 Why data privacy led Legora to run everything on Azure
• 30:04 Open-weight models and model arbitrage
• 30:55 Transcription models and why Legora doesn’t use ElevenLabs
• 33:33 The agent, Tabular Review, projects, and the Word add-in
• 34:24 How a lawyer uses Tabular Review
• 35:51 Why a spreadsheet works and what you learn from the first 10 rows
• 37:53 How the legal intake bot works in Slack
• 39:11 How agents create lists and lists start agents
• 40:15 Building products on those same pieces and acquiring Wexler
• 42:37 What forward-deployed engineers do at Legora
• 43:52 How legal engineers and forward-deployed engineers onboard a big client
• 46:43 What’s next for the forward-deployed team over the coming year
• 47:24 Why Legora resisted building APIs
• 49:26 Can the agent be the API?
• 50:12 The 300 Spartans slide and the plan to stay at 20 engineers
• 51:32 The bets nobody asked Legora to make
• 53:05 Post-training open-weight models for legal research
• 54:05 Search, retrieval, and late interaction
• 54:51 What makes a good legal agent harness and why drafting depends on search
• 56:25 Which model writes the best draft?
• 57:03 Hallucinations and Legora’s citation system
• 57:41 Do lawyers check the citations?
• 58:19 MapReduce or one big context window for a million-token contract?
• 1:00:16 Getting anonymized examples and the idea of buying a law firm for its data
• 1:01:53 Why Jacob advises against default embeddings
• 1:02:44 Will lawyers work in Legora or through Claude?
• 1:03:26 Will the model labs cut Legora off?
• 1:04:24 Law schools and scholarships for computer science majors
• 1:05:28 Hiring in New York, Stockholm, and London
• 1:06:09 Why most work eventually becomes legal work

文字稿Transcript

根据修订后的编辑内容自动转录；姓名和措辞可能有误。Automatically transcribed from the revised edit; names and wording may contain errors.

0:00 人们一直在问我们 Mythos 之类的东西，尽管我觉得现实中，很多法律工作并不需要 Mythos 级别的智能。律师助理，没了。没错。初级律师，没了。我不需要他们。其实我们看到的并不是这样。我们遇到过一个问题，比如它引用资料不太准确，于是我们就想，下一次模型发布就能解决。表现就从大概 70% 提升到了大概 99%。我们什么都没做，结果就好了。至少，如果你要做一笔 4 亿美元的并购交易，你肯定会想用 Legora。我们这家公司在往后的整个历史中都只会有 20 名工程师，然后我们要打败所有规模比我们大得多的公司。我们现在有 100 名工程师，但感觉人手还是太少了。不过，我正在尝试的一件事是：我们的产品有多少部分可以直接由智能体来完成，而不是依赖其他所有东西。 好，我们现在在录音室，特邀嘉宾是 Legora 的首席技术官 Jacob Lauritzen。我念对了吗？Lauritzen？——没问题。——好。——我不是西班牙人。——这名字怎么……——这是丹麦语。——没错。它有什么特别的来历吗？——其实……没有。就像“某某之子”的意思。——“某某之子”。就像你有 son，然后……——对。所以就是 Lauritz 之子。那是我的名字。——不错。 我们之前在各种场合碰过面。你在 AIE 做过几次演讲，而且很显然，Legora 是法律科技史上发展最快的公司之一。——没错。——你在录制前聊天时说了一件我觉得很有意思的事。你说你们的起源非常欧洲化。——嗯。——尽管显然你们正在纽约和旧金山扩张。你们的文化有什么不同之处？——我觉得从根本上说，我们做事可能一直都很谦逊，因为从一开始我们就是处于弱势的一方。没人相信斯德哥尔摩地下室里的一家小初创公司，能跟美国那家遥遥领先的庞然大物竞争。所以我觉得我们一直都是挑战者，也一直很谦逊。我们总得为自己的生存而奋斗。我觉得随着我们不断壮大，这也是我们不得不学习的一件事……随着我们进入美国、在美国待的时间越来越多，我们得学着更大胆一点，告诉全世界我们有多厉害，因为……我觉得这对我们当中的一些人来说，并不是自然而然就能做到的。0:00 People were asking us about Mythos and everything, even though I think realistically a lot of legal work does not require Mythos level intelligence. Paralegals, gone. Yep. Junior lawyers, gone. I don't need them. That's actually not what we're seeing, and we had one problem where, like, it wouldn't cite properly, and so we were, like, next model release solved. It went from, like, 70% to, like, 99. It's just, like, we did nothing and, like, great. At least you're doing a, you know, $400,000,000 M and A transaction. You're gonna want Legora. We're just gonna be 20 engineers for the rest of the company's history, and, like, we're gonna beat everyone who's much bigger. We're a 100 engineers now, it feels like we're way too small. But it's something that I'm trying with is, like, how much of our product can just be the agent versus all the other stuff. Okay. We're here in the studio with special guests, Jacob Lauritzen, CTO of Legora. Did I get it right? Lauritzen? That's fine. Yeah. That's okay. I'm not Spanish. How... It's Danish. Yeah. Exactly. Does it have a special history? It just... No. Is like son son of. Son of. It's like you have son and... Yeah. So son of Lauritz. That's my name. Nice. We've crossed paths here and there. You you you spoke at AIE a few times, and, obviously, Legora is one of the fastest growing legal tech in history. Yep. And you said something at the start in a pre show that I thought was really interesting. You said that your origin is very European Mhmm. Even though, obviously, you're expanding in New York and SF. What feels different about your culture? I think we are, just at the core, probably very humble about what we're doing because we've always been the underdog from the very, beginning. No one believed the small startup in a basement in Stockholm was gonna, you know, outcompete this huge behemoth that was, like, way ahead of us in the US. So I think we've always been underdogs, so we've always been very humble. We've always sort of had to fight for our existence. I think it's something we've had to learn as we've also become more and more... As we've entered the US and been in the US a lot more is to be a little bit more bold and, like, tell the world how great we are, because it's... It doesn't come naturally to, I think, some of us.

1:30 我想先聊聊你们的产品，因为我觉得很多这类法律产品都挺保密的，你知道吧？嗯。你们一般不太会展示产品。对，所以你们能不能聊聊产品是什么样的？还有，你们是怎么一步步做到现在这样的？比如说，产品经历过哪些迭代，才发展到今天这个样子？我觉得这会很有意思。 当然。最初一切都是从一个内置了瑞典法律搜索工具的 ChatGPT 开始的。好。那几乎就是全部了。我们就是从这里起步的，先做法律研究。这基本上是个巨大的市场：律师要花很多时间去查找类似的过往案例、法律先例，弄清楚在这种情况下，以前的人都是怎么处理的。这非常重要，所以它一直是我们产品的一部分。基本上，我们最初做的是一个智能体，你可以向它提问，它会查询法律条文。开端非常简单。之后，产品不断扩展，智能体能做的事情也越来越多。现在它能很好地起草各种文书、编辑 Word 文件、整理和移动文件、进行非常复杂的分析，我们也给它增加了许多其他功能。 我们的一个功能叫“表格视图”，现在这个名称已经在业内被采用了。我们很高兴看到 Claude for Legal 在描述他们开发的功能时，也用了我们的术语。 这个我们得聊聊，得聊聊。它本质上就是一种批量提取和批量审阅大量协议的方式，在尽职调查、融资之类的场景中都用得上。1:30 I kinda wanna start by going over your guys' product, because I feel like with a lot of these, like, legal products, you know, they're kinda like under wraps, like, you know Yeah. You're not supposed to like show it. Yeah, if you could just talk about, like, what does the product look like, and maybe, like, how did you guys get there? Like like, what have been the product iterations that got you to where you are today? I think that would be really interesting. Yeah. Absolutely. The way everything started was ChatGPT with Swedish laws Okay. Search tool in it. That was literally the entire everything. And so that's where we started. We started with legal research, which is like a huge market, basically, in the world where lawyers spend a lot of time figuring out what are previous cases that look like this, what's the legal precedent, like, is this, you know, what have previous people done in these situations? That's extremely important. And so that has always been part of our product. Basically, we started from, like, an an agent, you could ask questions and it would query laws. That was a very simple beginning. And then from that, it's expanded. And so it's expanded in, like, what can the agent do? It it will now, you know, draft a lot of stuff really well. It will edit Word files. It will, you know, move your files away, organize everything, run really complex analyses, and we've added just a bunch more features to it. So one of our features is called tabular view, which is now a name that's being used in industry, and we're very happy to see Claude for Legal using our terminology for what they're building. We gotta talk about that. We gotta talk about that. Which is essentially like a way to just bulk extract and bulk review many, many, many agreements, which you do in due diligence and when you raise money or whatever.

3:00 所以这只是其中一部分。我们帮助律师更高效、更好地完成工作，而且效率和效果都提升了很多。除了帮助他们完成工作，我们现在也比甚至一年前更能帮助他们统筹工作。我们有不同的功能，比如一个叫 Lists 的东西，对我们来说有点像 Linear。还有 Intake，可以用来和客户协作；还有 Portal，也是一种协作方式，你懂的，律师和律师、客户和律师、律师和代理人，或许还有代理人和代理人之间的协作。我们在这方面投入了很多时间，因为说实话，十年前，甚至可能直到今天，和律师合作的方式都非常令人头疼，而且有很大的空间可以做得更好。说起来，我觉得主要就是这两件事：帮助完成工作，以及统筹工作。你们是什么时候开始做或者开发智能体的？嗯，我们都……基本上经历了从助手到智能体的转变。对，我们是这么称呼它的，对吧？我想智能体……不知道，应该是一年前吧。对，所以给大家一个大致的时间线，方便了解 YC 2023。是的，我知道你是 2024 届的 YC 创业者。3:00 So that's part of it. It's like we help lawyers just do their work much more efficiently, much much better. And then outside of just doing the work, we're also now helping orchestrate the work a lot more than we were just even a year ago. So we have different sort of we have a a thing called lists, which is sort of like linear for us. And we have intake, which is a way to collaborate with clients, and we have portal, which is also sort of a way to collaborate, you know, lawyer to lawyer, client to lawyer, lawyer to agent, agent to agent, maybe. We're we're spending a lot of time there because there's just... Frankly, the way that you work with a lawyer ten years ago or maybe even today is is really painful, and there's loads of room to do that in a much better way, to be honest. And so those are, like, I think the two things. It's it's doing the work and it's also, like, orchestrating the work. When did you guys start doing or building the agent? So, I mean, we've all... We basically had this switch from the assistant to the agent Yeah. We call it. Right? I guess that the agent... I don't know. I wanna say a year ago, probably. Yeah. So this is a rough timeline for people who are understanding YC 2023. Yes. I know you are you are YC founder as '24.

4:30 YC 24。对。我是 YC ’23 届的。对。Legora 是 ’24 届的。好。然后……所以，确实有……那时候有点早，还在……智能体真正火起来之前。没错。然后，显然，使用模型彻底改变了一切。是啊。我的意思是，你知道，今天我们先预览了一下产品。看起来……嗯，算是预览吧。我们做了个内部演示。之后会把视频发到 YouTube 上。它基本上就像是给律师用的 ChatGPT。没错。对吧？这么说非常准确。不过侧边栏里有不少很棒的定制工具，比如表格视图、资料收集流程——我特别喜欢这个——还有其他一堆东西。没错。完全就是这样。它就像一位同事，或者说 ChatGPT，只不过特别特别适合法律工作。还有……对。这个工作流程既适用于企业内部律师，也适用于……比如，要查找法规的人。对。诸如此类。我们还可以多聊聊这些产品。好。你们底层用的是什么模型？能聊聊吗？我的意思是，我们会根据不同任务使用好几种不同的模型。对吧？就是这样。然后律师是看不到这些的，对吧？你们……或者说智能体？嗯，这方面我们一直在反复讨论。是啊，我想大家都一样。在智能体里，主模型有一个选择器，但有些工具会用不同的模型，还有表格视图之类的。我们会尽量……我觉得，最有效的做法是把问题拆分开，然后随着任务的拆分来分配模型。不过目前，智能体里确实有一个模型选择器。嗯。老实说，我不确定它会不会一直保留。所以……对。4:30 YC twenty four. Yeah. I was in YC '23. Yeah. Legora was '24. Okay. And then... So so there is... This is, like, a little bit pre a... Pre, like, agents really taking off. Exactly. And then, like, using models obviously changed everything. Yes. I mean, you know, in one in one sense today, you know, we had a preview of the products. It does look... Well, preview. We had a private thing. Well, we will put up a video on on the YouTube. It's mostly like ChatGPT for lawyers. Yep. Right? Like, very, like, very fair to say that, but then the sidebar has a lot of nice custom tools like the tabular view, like the intake, which I really liked, and a bunch of other Exactly. That's exactly what it is. It's it's coworker or ChatGPT, just like really, really good for legal work. And the... Yeah. The the the workflow for, like, either, like, in house corporate lawyers, like, people, like, looking for a coding regulations Yeah. All these things. And we can talk about more of the the products there. Yeah. What models are you guys using underneath? Like, can you talk about that? I mean, we're using a bunch of different ones depending on tasks. Right? That's And then that's not available to the lawyer to see. You guys Or the agent? Yeah. Well, we've been back and forth on this. Yeah. Just like everyone else, I think. In the agent, there's a model selector for the primary model, but there are certain tools that we use different models in, and there's, like, tabular view, and, like, we we try... I think we found it most effective to subdivide the problem and and and route on, like, as we subdivide the task. But we do right now have a a picker in the agent. Mhmm. I'm not sure if that's gonna stay there always, to be honest. So... Yeah.

6:00 是啊，我的意思是，我想先补充一点背景，因为我是 Cognition 的顾问。没错。我加入的时候……你知道，我对智能体实验室的构想有一部分是：理想情况下，智能体实验室应该是一个特定领域的 GPT 封装。对吧？我们是这个领域的专家。我们会关注你，为你提供服务，比前沿实验室做得好得多，还会根据我们的判断，为你提供前沿实验室最好的模型。是啊。所以我们的工作就是替你做决定，因为你不是……你的工作不是……没错。我们会始终让你处于帕累托前沿上。是啊，并确保它一直表现出色，你不必深入了解所有细节。因为，像我刚才说的那些……没错，就拿这件事来说吧：如果你去找律师，让他们根据某个先例起草一份合同，你觉得这是 Sonnet High 还是 Opus Mid 级别的任务？其实他们根本不知道。只要把模型给我就行。确保它能用就好。我们的时薪是一千美元。没错，只要把它做好就行。所以，是的，我一开始就是这么想的。我当时想，哦，对，这就是 Devin 的不同之处。而现在，模型选择器里有五个选项，包括 Fable 和 Opus。没错。Astra。对。人们想尝试那些新潮的模型。没错。你们……情况也一样吗？我觉得是。我觉得可能……法律领域肯定远没有编程领域那么明显吧？工程领域。对。不过，确实有人一上来就只想要最好的那个模型，而且你知道，也有人问我们 Mythos 之类的模型。对，对。尽管我觉得，现实中很多法律工作并不需要 Mythos 级别的智能。是的。还有……6:00 Yeah. I mean, so I bring a little bit of context to this because I'm an adviser for Cognition. Yep. And when I joined... You know, part of my my thesis for what an agent lab is, an agent lab is, you know, ideally a domain specific GPT wrapper. Yep. Right? Like, we we we run the... We are the domain experts. We will pay attention to you. We'll we'll serve your needs way, way more than the Frontier Labs, and we'll give you the best of the frontier labs according to our judgment. Yeah. So our job is to decide for you because you're not... Your job is not to Exactly. We'll always keep you on the Pareto frontier Yes. And make sure that it's it's great all the time, and you don't have to dive into all the details. Because, like, of the things that I've... Exactly to this point, if you go to a lawyer and you ask them, oh, you're gonna draft this contract based on this precedent, do you think that's a sonnet high or an opus mid task? Actually, they they won't know. Just give me the I've this model. Just, like, make sure it works. Our hourly rate is thousand dollars an hour. Exactly. Just like Make it work. So, yes, I originally came in with that. Was like, oh, yep. This is this is what Devin does differently. And now we have five things in the model picker, including Fable and Opus Yep. Astra. Yes. And people wanna experiment with the fancy new model. Yep. Do you... Is that the same? I I think so. I think it's prob... It's it's definitely way less in legal than it is in Coding? Engineering. Yeah. But there are definitely people that would just want table one right away, and, you know, people were asking us about Mythos and everything Yeah. Yeah. Even though I think, realistically, a lot of legal work does not require Mythos level intelligence. Yeah. There is

7:30 没错，完全是这样。我觉得基本上就是，你可以这么想：好，我们给你这个选项。你要这个，对吧？没错。但这样一来，你消耗额度的速度会比他们快 10 倍。是啊。我倒是很希望……我们为要不要引入这个功能争论了很久，最后还是有点妥协了。我们想，好吧，如果你想要这种程度的控制权……是啊，你懂的，那我们就给你。是啊。所以你觉得这些模型已经好到满足律师的需求了吗？它们不需要再进步了吗？这……我觉得这关乎能力的分布。也就是说，现在大概有 60% 的工作……是啊，我们现在已经能做得很好了。接下来主要就是成本和延迟的问题。但剩下的工作，尤其是非常复杂的分析，我们还是需要更强的智能。我觉得工作范围也会扩大。就像，我们现在做的是这些工作，但随着智能提升，我们与劳德代尔堡相关的工作范围大概也会扩大。是啊，我想是这样。刚开始的时候，比如说你们刚开始转向智能体……嗯，或者说开始构建智能体时，你们用的是什么模型？我猜应该是 Opus 4.6 之类的？应该是……对，Sonnet 4.6、Opus 4.6，大概没错。那时候智能体有哪些方面做得不好，而现在已经好得多了？虽然也才过去大概九个月。子智能体委派就是其中之一。现在我们已经把它做得很好了，比如用于扇出式检索查询。现在对我们来说效果非常好，以前则很难实现。我们基本上会追踪……我们有一个专门做评测的团队，会追踪模型在不同执业领域（也就是法律领域中的不同细分方向）以及各类任务上的表现。我们基本上会做一张很大的热力图，然后观察它们的变化。通常模型在大多数方面都会进步，但实际上，很多模型发布版本都会有一些方面变好、另一些方面出现退步的情况。明白了。而且这种情况发生的频率比你想象的要高得多。7:30 So exactly right. Like, I think, basically, you're like, okay. We'll give you the option. You you ask for this Exactly. But you're gonna run down your credits, like, 10 x faster than them. Yeah. I would have loved... We we fought introducing it for a very long time, and then in the end, we we just kinda caved in. We're like, okay. If you want this level of control Yes. You know, then we'll give it to you. Yeah. So you think the models are already, like, as good as lawyers need? Like, there's no need for them to get better? It's a... I think it's a power distribution. So, like, there's probably, like, 60% of the work Yeah. We can do really well right now, and you know, from now on, it's like a cost and latency thing. But then there's the rest of the the thing where, especially for, like, very complex analysis, we we still need more intelligence. And then I think there's also just the work expanse kind of. It's like that's the work that we have right now, but also the work with Fort Lauderdale, thus probably expands with intelligence Yeah. I imagine. Early on, like, when you guys first started, I guess, transitioning to the agents Mhmm. Or building the agents, like, what were... Like, what models were you guys using? I guess it would have been like Opus four six or something? That would have been... Yeah. Sonnet four six, Opus four six is probably right. What were the agents, like, not good at at that point? That they're much better at now even though it's only a little, like, nine months later? Sub agent delegation was one of them that we've gotten working really well now for, like, fan out type retrieval queries. Works extremely well for us now. That was difficult before. We basically track... So we have a team that's just doing evals, and so we track all of the model's performance in different practice areas, which is like subgroups of of legal, and different tasks across all of those. So we basically have, like, a big heat map, and we're like, and then we can see them move, and it's like, usually, they improve most stuff, but there's actually a lot of model releases where they improve some and and progress in others. Okay. And that actually happens a lot more than you'd think.

9:00 比如说，具体有哪些任务？你熟悉你们是怎么进行各种评测的吗？你们会关注哪些不同的子模块？嗯。比如说，会有一些不同的任务框架。我们基本上有不同的场景，本质上就是 RL 环境。比如说，律师的工作几乎总是围绕项目展开的，所以场景可以是：这里有一个包含一万份文件的数据室，你要做尽职调查，最后的 GRIP 报告应该是这样的：里面要包含所有这些风险点，不能包含一些奇怪的内容，而且要简洁，等等。我们的系统大致就是这样构建的。任务也各不相同，其中一部分是起草报告：草稿写得好吗？该写的内容都包含了吗？措辞准确吗？分析可靠吗？还有一部分是找出风险点，也就是前面的分析工作，所以我们可以把它再细分一下，这就是我们进行评测的方式。挺有意思的。等等，你们的评测团队是什么样的？你们会聘请律师来做这些吗？会的，团队是混合构成的。我们有工程师，也有律师，他们都在同一个团队里。现在大概有三名工程师和三名律师，此外还有一批我们称为“法律工程师”的人，他们通常在大型律师事务所工作过三四年，也许有计算机科学学位，而且对科技很感兴趣，然后他们就加入我们……因为他们想帮助改变这个行业。9:00 Like, what are some of those tasks? Like, are you familiar with how you guys are running a lot of the evals? What are those different sub segments, I guess, that you look at? Yeah. So, like, there's different just like framing stuff. So we have different, basically, like, worlds, like RL environments, essentially, which is like, here's a... Because because lawyers work in projects always, more or less, so it's like, here's 10,000 file data room, you're doing your due diligence, in the end, a GRIP report looks like this, like, it contains all these red flags and, like, it doesn't contain whatever weird stuff and, like, we want it to be concise, etcetera, etcetera. That's kind of how we've built it. So there's different tasks, which is, part of it is drafting a report. Like, does the draft look good? Does it contain all the right stuff? Is the wording sound? Is the analysis sound? There's also the part of finding the red flags, like the analysis before that, so we can kinda like subdivide it a little bit, And that's how we run the evals. Interesting. Wait. What does your evals team look like? Like, do you hire lawyers to, like, do that? So, yes, it's a mix. We have engineers, and then we have lawyers, and there's one team. So there's, like, three engineers and, like, three lawyers right now, and then we have a pool of what we call legal engineers, which is people that have worked in big law for like three or four years. Typically, maybe they have a CS degree and like they're very tech forward, and then they come to us and... Because they wanna like help change the industry.

10:30 他们是在大型律所做 IT 的人吗？还是不是。他们是在大型律所当律师的人。他们是在大型律所当律师的人，而且有计算机科学学位。是的。我的天。这是个维恩图……对，没错。这类人是我们最难招的，但他们也能极大地提升我们的效能，因为我们可以让他们参与评测相关工作和产品开发。你想想，我们有一群 100 人的专家，每天都在用这个产品，也每天都去跟客户交流；而且他们亲身经历过、一直都在经历这些。我们只要让他们参与评测，帮我们打磨产品，这就是非常非常大的优势。挺好奇的，你们是怎么组建这个团队的？评测团队是在某个阶段从助手代理团队发展出来的。我觉得直到不久之前，有段时间说实话，做评测确实不太值得，因为所有模型都变得更好了。所以问题就是，我们要不要花很多时间去做特别复杂的评测，还是干脆去构建……因为改进测试框架、顺势相信一切都会变好，能带来很多收益，而后来一切确实也都变好了。嗯。不过，是啊，每隔两三个月就会有一个不同的模型。对。没错。我的意思是，我们把它过一遍。对吧？没错。我的意思是，我们遇到过一个问题，比如它没法正确标注引用，所以内容很难追踪；结果下一个模型版本就解决了，准确率从 70% 提高到了 99% 左右。我们什么都没做，事情就变好了。是啊。也许真正需要做的工作，是确保所有数据都保持私密、可以打标签、可以排序，再添加新功能，比如表格视图。你是指……也就是除了 ChatGPT 之外？对。不过显然，你们现在也在做评测了。我猜你们对开源模型的使用也在增加。这个我们可以聊聊。是啊，我也不确定自己想表达什么，除了……你们的工作到底是什么？也许一开始的工作并不是做评测，因为那时候主要是把模型的能力开放给律师使用。10:30 Are they IT in big law? Or No. They're lawyers in big law. They're lawyers in big law. With CS degrees. Yes. My god. It's a Venn diagram that... Yeah, exactly. That's like our hardest person to hire for, but they're also such a force multiplier because we can pull on them for the eval stuff and for the product development, which is like you have a pool of a 100 experts that use the product every day and go and talk to clients every day, and also, like, they live this and have lived this, and we just pull on them to do eval stuff, and just like help us shape the product, and it's such a such an advantage. And curious, like, did you guys start that team? Evals grew out of the assistant agent team at some point. I think there was a point up until recently where it was like, honestly, it's just not really worth it, because all models just got better. So it was like, do we want to spend a lot of our time doing really complex evals, or do we just want to build... Because there was so much to be gained from building a better harness and sort of leaning into everything's gonna improve, and everything kind of just improved. Mhmm. But, yeah, like, every two, three months, you get a different model. Yeah. Exactly. Mean, we pass it through. Right? Exactly. I mean, we had one problem where, like, it wouldn't cite properly, and so it would be difficult to follow, and it was like, next model release solved, it went from, 70% to, like, 99. It's just like, we did nothing and, like, great. Yeah. Maybe the the the real work is, you know, keeping all the data private, making it taggable, sortable, adding new features like the the tabular view. Is that the, you know You mean over and above ChatGPT? Yes. But obviously, now you're also doing evals. I assume your your use of open models is also increasing. We can talk about that. So, yeah, I I don't know what I'm saying here apart from, like, you know, what is the job? The job is maybe not... In the early days, it was not so much to do do evals because it was mostly exposed model capability to lawyers.

12:00 是的。其中很大一部分就是弄清楚，随着这些变化，法律行业五年、十年后会是什么样子。比如，编程领域可能领先一两年，我们多少可以参考一下，但关键是，基本构件会是什么？人们会如何工作？如果不再是发两百封邮件、来回传五个版本的 Word 文件，人们要如何协作？那会是什么样子？而且这甚至不一定纯粹是 AI 的事，只是要打造好得多的产品体验。没错。还要确保做到审计日志、数据留存，以及受监管行业里那些非常重要的事情。然后还有 AI 这一部分，就是让工作更快、更好，并让主动式智能体在收到新事项时就能作出响应，这样你就不必事事都亲自启动。我觉得，很多工作就是要弄清楚，我们认为五年后这一切会是什么样子。如果模型能力持续提升，价格不断下降，我们能做些什么？你登录后，看到的会不会是一个指挥中心，里面列出智能体最近处理的五件需要你响应的事，让你去审查、核对并改进系统？也许我们正朝这个方向发展，或者说，我们一定会朝这个方向发展。对，对。我本来想问，你能不能再详细谈谈这个？就是，嗯，几年后法律行业会是什么样子？12:00 Yep. The huge part of it is just figuring out what does the legal profession look like in five years, ten years with all of this stuff. Like, coding is maybe a year or two ahead, that we can kinda look to that a little bit, but, like, what are the primitives? How do people work? How do you collaborate if it's not a 200 thread email with, like, v five word files back and forth? Like, what does that look like? And that's, like, not even necessarily purely AI. That's just, like, build a much better product experience Yeah. And make sure you do it with, like, audit logs and data retention and all of those sort of really important things in regulated industry. And then there's the the AI part, which is, like, make the work faster and better and proactive agents that react to stuff when it comes in so you don't have to, you know, kick off everything. So I think a lot of the work is just figuring out what do we think it's gonna look like in five years. If, you know, model capabilities keep increasing, if price keeps coming down, what could we do? Can you log in and it's like command center where it's just like, you need to react to these five things that the agents have been doing lately and, like, review and check and improve the system? Maybe that's where we're going or it is where we're going. Yeah. Yeah. I was gonna ask, like, could you go into that in a little more depth? Like, yeah, what what is the, like, law profession gonna look like in a few years?

13:30 我得先说明一下，我也不知道。我的意思是，说真的，谁知道呢？但我觉得，我们可以多看看编程领域。我认为所有律师的效率都会比以前高得多。他们能以更低的成本，完成更多质量更高的工作。我觉得整个行业本身、工作本身都会有很大改善。会变得更……我觉得律师会把更多时间花在他们当初选择当律师就是为了做的事情上，比如判断什么样的风险状况才合适、什么样的风险可以承受，还有……发挥他们的判断力和战略判断力，以及和客户合作。他们会更多地做这些事，而少做那些……翻查大量文件寻找关键证据，或者处理那些繁琐又特别烦人的杂活——很多律师一直都在做这些。你觉得合伙人模式会怎么样？你觉得它会保留下来吗？还是说会发生什么变化？ 我上周刚好认真想了想这个问题。对。就是合伙人模式……我不想讲太多细节，但合伙人模式之所以存在，某种程度上是因为这些公司的资本只有人力资本。它不像我们这样的公司，也不像软件工程公司、制造企业之类的——这些企业拥有实物资产，或者某种会随着时间积累、并为股东带来红利的无形资产。但如果你在专业服务行业工作，比如咨询公司或者其他类似行业，你所有的价值都在员工身上，而人力资本并不是你能真正拥有的。你知道，你不能……奴隶制早就废除了。人不是可以拥有的，但可以租用。你可以付钱雇他们。所以合伙制本质上是一种让人力资本成为公司所有者的架构，这样他们就会留下来。对。这会改变吗？我也说不准……倒也不是非改不可，但我觉得我们真正想做的，是让律师——实际上是让律所——能够建立某种无形资本、某种持久的资本。比如说，你可以建立一个能够学习你们工作方式的系统。13:30 I should preface this by saying I don't know. I mean, who knows, really? But I think we can look to coding a lot. I think all lawyers are gonna be way more productive than they were before. They're gonna be able to do a lot more stuff at a higher quality for lower cost. I think the profession itself, the work itself, is gonna improve a lot. It's gonna be a lot more... I think lawyers are gonna spend more of their time doing the stuff that they became lawyers to do, you know, figuring out what's the right risk profile, what's manageable risk, what's like the... All of their judgment, strategic judgment, and like working with clients, they're gonna do that a lot more and much less of the, like, trolling through big documents to like find the smoking gun or like do all this menial super annoying work that a lot of lawyers have been doing. What do you think is gonna happen in the partnership model? Like, do you think that stays or, like, what happens? I gave this... I actually gave this a lot of thought just, like, last week. Yeah. Just the the partnership model... I don't wanna get into too much of the details, but, like, the partnership model kind of exists because the only type of capital that these companies have is human capital. So, like, it's not like, you know, us or like a software engineering firm or like a manufacturing or whatever. You you have physical assets or like some sort of intangible assets that that you accrue over time and and they pay dividends to the shareholders. But if you work in an industry like professional services, and it's the same for, like, consultancies and and whatever else, all of your value is in your people, and you can't really own human capital. You know, you you can't... You know, slavery is abolished. Can't own people, but you can rent them. You can pay them money. And so partnerships essentially is a structure where you make the human capital the owners of the company so that they stay. Yeah. Will that change? Like, I don't... It doesn't have to, but I think what we really want to do with the lawyers, we actually want lawyers to be able... Law firms to be able to build some sort of intangible capital, some durable capital, and, like, you can build a system that learns of how you work.

15:00 现在很多公司会想，嗯，律所原本有一套做事方法，我觉得我们可以把它编纂成体系，建立一个系统，不管它是能自我改进的知识图谱还是什么，也可以把先例文件和文档纳入其中。这样就有了一套能产出非常优质成果的工作方法，而这也可以成为他们的知识产权。也就是说，把人的专业知识和判断力编纂进一个系统里。我觉得这……我现在完全是在漫无边际地说了，但这确实是我们想帮助他们实现的事情。那么，这会改变合伙制模式吗？因为激励机制不一样。在合伙制模式下，你只需要招揽更多客户，而你不想把招揽客户的时间挪去编纂自己的信息，基本上就是这样。嗯，因为你得到的回报是“做多少，得多少”，或者说“靠自己拿下业务”之类的。这一点需要稍作改变。我们得想出一种激励机制，让合伙人……有动力为公司编纂这些东西、打造资产。但现在还没有这样的动力，因为他们一旦离开合伙关系，就无法从自己打造的资产中获得任何分红。15:00 A lot of firms right now think about, like, well, there's the ex law firm way of doing things, and we can codify that, I think, and we can build a system, and whether that's like a self improving knowledge graph or whatever it is, and like precedent files and documents, where there's like a way of working that produces a really good output, and that can be their IP. So it's like a way to codify the human expertise and judgment into a system. That I think is... Now I'm just rambling completely, but like, that's something that we want to help empower. And does that change the partnership model then? Because incentives are different. In the partnership model, you just gotta bring more clients, and you don't wanna take away time from bringing in clients to codify your information, basically. Mhmm. Because, like, you get rewarded from, you know, eat what you kill or make this luck step, whatever. That needs to change a little. We need to figure out an incentive structure where partnerships... Partners are incentivized to codify things for... To build an asset for the company, but right now, there aren't because as soon as they leave the partnership, they don't get any dividends from the assets that they built.

16:30 是啊。所以这就有点……对。我觉得整个商业模式……我们再看看吧。是啊，是啊。现在一切都还很不确定。是啊。我另一个实际的看法……你知道，在合伙人层面，那算是行业的顶端。嗯哼。其实我的看法更偏向底层。对吧？对。那边发生什么了？律师助理没了。对。对吧？比如初级律师没了。我不需要他们。就……其实我们看到的不是这种情况。所以，是啊，情况是……我觉得是……我知道。你会不会说我们在工程领域也看到了这种情况？是啊。所以初级工程师现在确实有点完了。你得从……从大学出来时就已经完全成型，否则你就……混不下去。我觉得我不是……我这么说有点夸张，是为了搞笑。不过，是啊，我认识很多律师。他们一开始都是律师助理，然后去读法学院，经历这些。对吧？而现在他们不……不再被需要了。他们真的就是不需要了。对。对。是啊。合伙人制会变成什么样，比如金字塔结构，这挺有意思的。我觉得一般来说，律师还是需要……需要下一代。至少对工程来说，我的想法是，今天一个优秀的工程师，价值比两年前高得多，因为他们能产出那么多东西，而且质量还很好。我觉得法律行业也是一样，优秀的人确实能做很多非常出色的工作。所以我觉得更像是，真正优秀的律师，或者初级律师，价值高得多。是啊。一直都是这样。对吧？是啊。似乎现在更明显了……不过，是啊，两极分化确实更加严重了。16:30 Yeah. So there's like a... Yep. I think the the the whole business model is is... We'll we'll see. Yeah. Yeah. It's all it's all in question right now. Yeah. My my actual other take... You know, so on the partnership level, that's like the top end of the Mhmm. The industry. Actually, my take is more on the bottom end. Right? Yep. What happened there? Paralegal's gone. Yep. Right? Like, junior lawyer's gone. I don't need them. Like, it's That's actually not what we're seeing. So, yeah, it is... I think it's... I know. Would you would you say that we're seeing that in engineering? Yeah. So junior engineers are really kind of cooked. You kinda have to come out of... Drop out of university fully baked or else you're You're not gonna make it. I feel like I'm not... I'm being a bit dramatic for for a comedic purpose. But, like, yeah, I know a lot of lawyers. They started as paralegal. They go to law school, all these things. Right? And now they don't... They're not needed. They're they're seriously just not needed. Yep. Yep. Yeah. It is... It's interesting what happens to the partnership model, like the pyramid. I think generally, a lot of lawyers, you still need your... You need the next generation. And at least for engineering, the way I think about it is you have a good engineer today, they're worth a lot more even than they were two years ago because they can produce so much more stuff, like, they're good. And I think it's the same thing in legal, where they can actually do a lot of really good work if they're good. So I think it's more that really good lawyers or junior lawyers are worth a lot more. Yeah. It's always been the case. Right? Yeah. Seems like even more But it's, yeah, it's polarizing even more so.

18:00 嗯。我其实挺好奇这个的。就整个法律行业基本上都是按小时收费的。嗯哼。对。所以你知道，如果现在每个项目都能做得快很多，对。那是不是会便宜很多？比如说——其实是这样的……有个很有趣的故事。我当时在看 Max 的 YC 申请，还在帮他申请进 YC。当时它叫 Leya，是一家 YC 公司。这是我当时唯一想问的事。我就想，而且我很好奇，Leya 是丹麦语里的某个词吗？不是。瑞典语？瑞典语？不是。它也没什么含义。有个……好吧，好吧。早些时候有家西班牙报纸报道我们 Leya，我觉得是吧，里面有个特别好笑的故事。他们解释说，这个缩写的意思是“法律高效尤达助手”。为什么是尤达？是啊。我也不知道。所以我就想，为什么……你看，产生幻觉的不只有 LLM。不。我觉得记者也会产生幻觉。不过，不，它确实没什么含义。他写了申请。他写了……对。我给他的反馈是：你知道，已经有一百万人试过这个了。法律科技一直都没成气候，因为律师没有动力去……他们为什么要提高效率？他们赚的钱反而会更少。这根本说不通。但我觉得律师确实想采用这类技术，原因是他们明白形势有多严峻。他们知道商业模式和一切都面临压力，而且竞争压力太大了，他们必须改变。我觉得计费模式本身也会改变。我们已经……我们已经看到这种情况了，尤其是在美国，很多服务实际上是按项目收费，而不是按小时收费，尤其是那些处于较低端的业务，比如审查之类的。不管是什么，其中一些说实话是免费做的，还有一些就是按固定价格收费。那么……在哪里呢？18:00 Yeah. I'm actually curious about that. So, like, the whole, like, law business is basically, like, charging per hour. Mhmm. So Yep. You know, if each project now can be done Yeah. A lot faster is it going to be, like, way cheaper? Like This was actually... So fun story. I was I was reading Max's YC application and, like, helping him getting into YC. Leya, it was called back then, is YC. And this was my one thing. I was like And and curious, Leya is a Danish thing for something? No. Swedish. Swedish? No. Also, it means nothing. There's this Okay. Okay. There's this really funny story in... I think it was a Spanish newspaper that wrote about us in the early days, Leya, and then they they explained that the abbreviation was legally efficient Yoda assistant. Why Yoda? Yeah. I don't know. So I was like, why why is it See, hallucinations are not just from LLMs. No. I think journalists hallucinate. But, no, it doesn't mean anything. So he wrote the application. He he wrote... Yeah. And my feedback to him was, you know, people have tried this a million times. Legal tech has never been a thing because, like, there's no incentive for lawyers to to... Why would they be more efficient? They earn less money. Like, it doesn't make any sense. But I think the reason that lawyers do wanna adopt this is that they understand the gravity of the situation. They know that, like, the business model and everything is under pressure, and it's like they need to change because the competitive pressure is so high. The billing model itself, I think, is also gonna change. We've... We're already seeing that, particularly in the US, that a bunch of stuff is actually being priced on, like, a project fee and not on an hours fee, particularly these, like, sort of lower end stuff on the spectrum, like, whatever it is, like, reviews and whatever. Some of it, honestly, is being done for free, and some of it is just being priced on a flat rate. So where is

19:30 那时赚到的钱呢？是……因为在战略工作方面。是啊。我觉得你们谈的很多都是并购，特别是并购，好像那才是所有钱都在的地方，这是个很容易理解的例子。哦，好吧。好吧。诉讼也是。重要的工作仍然在那里，而且价值很高。所以他们会……律师会把更多时间花在这上面。客户关系也非常重要。像 CRM 那样？不是。就是多花些时间陪客户。很多律师事务所会——或者说过去一直都是这样——基本上会免费做一些事，只为了维持和客户的关系。对，对。所以现在他们可以继续这么做，但不用因此亏那么多钱。我还以为……我还以为我得付律师钱，就为了让他当我的朋友。像是……对。这不就是聘用律师要付的那笔预付款吗？对。对。那……你刚才告诉我可以免费做？就……对。你只要找到合适的律师就行。没事。不错。或者 Latent Space 得发展壮大很多才行。我也不知道。是啊。是啊。我们在努力嘛。你懂的？对。所以……好。所以你是说，显然，战略性工作以及所有这些高附加值工作仍然存在，但我们的法律费用并没有下降。不过你也说了，律师助理不受影响，或者说，他们只是转变了工作方式。这也是我们在 Cognition 讨论过、担心过的事情。没错。大家会担心这种说法，比如“我们会取代你的初级工程师”，因为没人喜欢这种说法。但事实上，确实有一部分情况会这样。还有一些初级工程师会拥抱这项技术，变得更高效，这正是你刚才说的那种情况。所以我想问的是，这种工作或职位有没有一个新的称呼，正在得到客户的认可？19:30 the money being made then? Is it... Because In the strategic work. Yeah. I feel like a lot of what you guys talk about is M and A specifically, like, that's where all the money is a very easy example. Oh, okay. Okay. Litigation is also. The big work still is there, and it's very valuable. And so they'll they'll spend... Lawyers will spend more of their time there. Client relationships, also super important. Like a CRM? No. Just like spending more time with the client. Like, a lot of law firms will, or historically have basically done stuff for free just to have their relationship with the client. Yeah. Yeah. And so now they can do that, but they don't have to lose a lot of money on doing it. I thought I thought I just... I have to pay the lawyer money just to be my friend. Like, in this Yeah. Isn't that the retainer thing? Yeah. Yeah. The... That's You you told me I can do it for free? Like Yeah. You just have to find the right lawyers. Yeah. It's alright. Cool. Or Latent Space has to grow much bigger. I don't know. Yeah. Yeah. We're trying. You know? Yeah. So... Okay. So so so you're saying, you know, obviously, the the strategic work and all these high value add is still there, but our legal fees are not going down. But you also said paralegals are not affected or, like, they're they're just shifting. And and this is something that we also did with... Worry about with Cognition. Like Yep. People worry about the phrasing of, like, we will replace your junior engineer because nobody likes that kind of statement. But in actual fact, yes, like, some some part of it does happen. Other junior engineers embrace it and they become more productive, which is exactly like you just said. So I guess, like, is is there a new term for this role or this this job that is taking... That is having traction with your customers?

21:00 你是说那个超级拥抱 AI 的初级法律从业者？我还没找到什么。也许我们应该造一个词。那不算合法吧，合法——那是我们的……而且现在也成了一个行业术语。对。对。CSJD。没错。对。对。对。对。也许就是……还有些其他人最适合做这类事情。我觉得“肉身代理”这个贬义词，说的就是你坐在机器前，输入别人让你输入的东西，然后再把机器告诉你的内容复述出来。没错。不过，我确实觉得，一个很优秀的肉身代理，就像……你很了解智能体，也很了解 Legora。没错。你还知道该怎么做，比如说，这是我的工具包，而每个行业都有自己的工具包。没错。我可以用我的工具来做你需要我做的事。不过我想这样的话，你就不太算代理了，对吧？你实际上是在做有意义的工作。对。就是我不再……你知道，我就拿表格审查来说吧，因为那……这事儿经常发生。我有几百、几千、几百万、几千……还有几千份。我们一次好像最多能处理一百万个单元格。在一个……真厉害。没错。我们应该聊聊那个表格和窗口化。至少，你知道，我有……我在处理一个大案子，有一百万份文件。我把它们输进去。通常我会把一对……把一队律师助理关在房间里，一直干到凌晨四点。可这根本没必要，对吧？因为现在他们可以把信息提取出来，你也可以重新提示所有内容。嗯。所以，是啊，我现在不需要律师助理来做这件事了。不过，一个很懂 AI、善用 AI 的律师助理可以去做别的事。21:00 You mean the the Super AI poll junior legal? Nothing I found yet. Maybe we should coin one. Isn't it legal legal That's our... And and also now an industry thing. Yeah. Yeah. CSJDs. Exactly. Yeah. Yeah. Yeah. Yeah. And maybe that's the... There are those other people that are best positioned to do this kind of stuff. I think the the the derogatory term of meat proxy, where you just, like, sit in front of the machine and type in something that someone else tells you to type, and then you repeat what the machine tells you to say. Yep. But, like, I do think a very good meat proxy is just like... Well, you know agents very well, and you you know Legora very well Yep. And you know how to, like like, this is my toolkit, and, you know, every industry has a toolkit Yep. And I can use my tools to do the thing that you need me to do. But I I guess then you're not really a proxy. Right? You're actually just doing meaningful work. Yeah. It's it's just like I no longer... Like, you know, well, I'll talk about tabular review because that... That's that's a thing that happens a lot. I have hundreds, thousands, millions, thou... And thousands. We do up to a million cells, I think. In a in a single... That's impressive. Yep. We should talk about the the the table and the windowing. Yeah. Least, you know, I I have, like... I'm working on a big case. I have a million documents. I type it in. Usually, I'll throw a pair... Throw, like, a team of paralegals locked in the room. There, so, like, 4AM. Like, that that doesn't need to happen. Right? Because now they can extract it, and you can go and re prompt everything. Mhmm. So, like, yeah, I I think now I don't need a paralegal for that, but, like, a good, like, super AI pilled paralegal would do something else.

22:30 你知道吗？是的。对。没错。不管是什么，他们都能做更多事情，或者把更多时间花在以前没时间做的事情上。也可能真的去睡觉，谁知道呢？不。我有个问题。你们收购了一家公司，Qura。是的。是这么念的吗？不是。你们用自己的产品来做并购？是的。我们一直这么做。那效果怎么样？能不能稍微聊聊……可以。没问题。效果很好。我们经常这么做。我们的法务团队……我的意思是，接收流程是和我们的法务团队一起搭建的。因为现在我们有一个 Slack 工作流，大家都把东西往里扔，然后就会说：“这真烦人。”对吧？而且效率低，所以我们就想，太好了，这种事应该用 Legora 来做。是啊。能不能稍微聊聊这个？比如说，这个流程大概是什么样的？可以。当然。我们自己会做很多尽职调查，因为可以用表格审查和其他工具来做。不过我们还是会请律师参与。嗯哼。比如评估某些风险之类的。所以 Qura 这个案子很简单，因为一切都在瑞典。不过我们还在加拿大和澳大利亚各做过一笔交易，澳大利亚那边就涉及薪资问题。好。我们要不要请人从更具战略性的角度来看看？所以在这方面，我们还是会请律师来做战略层面的工作，他们可以成为我们的讨论伙伴，帮我们想清楚怎么以合适的方式安排这笔交易。但很多更琐碎的事情，我们其实可以自己做。比如说，很多早期的问题都是我们自己发现的，也有过发现某个问题像是个红旗，然后我们就踩了刹车的情况。能不能稍微聊聊其中一个？不，我想还是算了。好吧。你刚才说，早期的客户都是瑞典的？嗯哼。比如瑞典的律所？对。你们觉得和瑞典律所合作，跟和美国律所合作有什么不同吗？嗯哼。类似这种？而且你们合作的主要都是大型、大型律所，对吧？22:30 You know? Yes. Yeah. Exactly. Whatever it is. They can do more stuff or, like, spend more time on the things that they, you know, didn't have time for before. Or actually sleep, who knows? No. I have a question. So you guys bought a company, Qura. Yes. Is that how you say it? No. You guys used your own product to do the M and A? Yes. We do that all the time. So how did that work? Like, can you talk little bit about Yeah. No. That works great. We're doing that a lot. Our legal team is is... I mean, intake has been built with our legal team. Because right now, we have a Slack workflow where everyone throws their stuff in, and they're like, this is annoying. Yeah. Right? And inefficient, so it's like, great. We should use Legora for that. Yeah. Can you talk a little bit about that? Like, you know, what that process looked like? Yeah. Absolutely. So it's it's... We do a lot of the due diligence ourselves because we can do it with tabular review and other tools. We do still have counsel coming in Mhmm. To, like, assess certain risk or, like, you know, there's and so Qura was simple because it was all in Sweden, but, you know, we've done one in Canada and one in Australia, and so there's, like, payroll in Australia. Okay. Do we want someone to look at this, like, a bit more strategically? So there, we will still use lawyers for the strategic part, where they can actually be sparring partners and help us a lot on, like, how to figure out how to structure this the right way. But a lot of the more menial stuff, we could actually just do ourselves. So, like, a lot of the early things we would find, and we have had cases where we found something, that's been, like, a red flag and then we've we've pulled the brakes. Can you talk a little bit about any of these? No, I think so. Alright. Early on, you said you... Your early customers were Swedish? Mhmm. Like, Swedish law firms? Yep. Have you seen a difference between working with, like, Swedish law versus, like, American law firms? Mhmm. That sort of thing? And you guys are mostly working with, like, big, big law firms. Right?

24:00 嗯，我们也……规模较小的律所也有。哦，我们确实最早是从特别特别大的律所开始的，但有 50 名律师的律所也能从中获得很多价值。是啊。那么，你有没有发现和比如说瑞典的律所合作……是啊……跟美国的律所相比，或者甚至，嗯……我不知道，亚洲的律所？是啊，是啊。我觉得欧洲的律所行动起来稍微慢一些。美国的律所则很积极，行动特别快。我觉得这在美国公司的议程上也排得很靠前，比欧洲公司更靠前。我觉得这是一件具有战略意义、极其重要的事。他们会觉得，我们得在这方面投入很多。总体来说，美国公司，无论是企业还是律所，都更愿意定制开发东西。这是他们都想做的事。比如说：“嘿，我们想和你们一起打造一个很大的东西，用你们的产品，并且按任何我们想要的方式来设计。”而欧洲人则更倾向于：“我们直接用现成的就行。”他们想自己开发的东西，比如说，是什么？接入他们自己的……有些客户会自己开发 CRM、自己的文档管理系统，还有各种其他系统，他们想把这些系统接进来；然后他们还想设置触发机制，比如说，你在 Salesforce 里做了一次审查，系统就会因为某种原因把它发给某个人，那个人处理完之后，再传到 Legora，然后再传回来。还有……这算是比较简单的那一端。有些客户的想法则非常大胆，比如想和我们一起打造一个全新的产品，而我嘛，你懂的，我们其实做不了这种事。是啊，是啊。但可以肯定的是，美国公司比欧洲公司更想自己开发东西。24:00 Well, we're also... Smaller law firms also. Oh, We we definitely started really, really big law firms, but 50 lawyers also and and get a lot of value. Yeah. So have you seen differences between working with, like, the Swedish ones Yeah. Versus American ones, maybe, like, even Mhmm. I don't know, Asian ones? Yeah. Yeah. I think the European ones have been a bit slower to move. The American ones are very forward, move really quickly. I think it's also very high on their agenda of the American companies, more so than the European ones, I think. It's like a strategic, super important thing. We need to just, like, do a lot here. The American companies generally, both corporates and law firms, also have a much higher appetite to, like, custom build stuff. That's like a thing that they all want to do. Like, hey, we wanna build this, like, huge thing with you guys and, like, structure it in whatever way, whereas Europeans are much more like, we'll just take the off the shelf and What's an example of something they would wanna build themselves? Integrations into their own... Like, some of these customers build their own CRM and their own document management system and their own all sorts of stuff, and they wanna build that, and then they wanna build triggers that, like, you do a review in in in Salesforce and it goes to this one person for whatever reason, and they do something, and then it goes into Legora, and then it goes back in. And there is... So that's like the simple end, and there are some that are extremely ambitious, whereas, like, a whole new product that they want to build with us, and I, you know, we can't really do that. Yeah. Yeah. But but that's definitely a thing where, like, American companies really want to build their own more than European ones do.

25:30 欧洲那些客户就会说，嘿，你知道，他们会给你一些产品建议什么的……对，当然。我的意思是，我们和他们所有人合作得都很好。是啊。他们会说，嘿，这个很糟糕，这个很棒，我们想要更多这样的功能，然后我们就会做出一个能实现这些功能的产品。好吧。所以他们不觉得这会对他们的业务构成生死攸关的威胁？我是说……而美国那些客户会这么觉得，还是……我不知道这是不是他们想要定制开发，而不是用非定制产品的原因。我觉得他们都知道，这对他们公司的成功以及企业的成功来说极其重要、至关关键。所以他们……这件事对双方都极其重要。我觉得只是实现方式不同。好吧。还有个问题，我觉得你可能也想提一下，就是产品指标。比如说，你怎么知道我们实际上做出了有价值的东西？对。对。你怎么看这个问题？我猜这也和评估有关。所以我们有一些非常看重的常规产品指标：用户是不是在我们的产品上花了很多时间？他们似乎从中获得了很多价值吗？他们会续订吗？他们会经常回来使用吗？作为企业软件，我们的数据非常好，比如 DAU/MAU 的比率，还有用户每月在产品上花费的小时数，都非常高。然后我们也会做一些……我们也会进行在线评估，大致就是想弄清楚，我们是否解决了这个问题。我们会用一些成本极低的 LLM 评估方式，设法建立一个在线指标；这样在推出新功能时，也可以用这个指标。我们可以按功能标记来比较：好，效果是提升了，还是灾难性地下滑了？其实有一点可能挺有意思，但对我们的业务来说也很棘手：我们基本上没法查看任何调用轨迹，也看不到任何数据，因为所有内容都高度机密。这让人工作起来非常沮丧，因为你基本上不知道用户在做什么。就好比，这是一个智能体，它能做很多了不起的事，可我完全不知道用户实际上在拿它做什么。25:30 And the European ones are like, hey, like, you know, they give you product suggestions or... Yeah. For sure. I mean, we have great collaboration with all of them. Yeah. So they'll they'll say, hey, this sucks, this is great, we want more of this, and we'll just build a a product that does that. Okay. So they don't see, like, an existential threat here in their business? I mean... Whereas the American ones do, or... I don't know if that's the reason that they want a custom build versus not custom I think they both know that this is, like, extremely important and key to their firm's success and their corporate success. So they... It's both extremely important. I think it's just a different way to achieve it. Okay. Another question I think maybe you you wanna bring this up is, like, product metrics. Like, how do you know that, like, you know, we're actually doing something good here? Yeah. Yeah. How do you think about that? I'm assuming that's also related to evals. So we have the regular product metrics that we care a lot about. Are people spending a lot of time in our product? Do they seem to get a lot of value out of it? Do they renew it? Do they come back a lot? We have really good numbers for an enterprise software of, like, our DAU over MAU and our, like, you know, hours spent per month in the product are extremely high. And then we do some sort of... We do online evals also where it's, like, try to figure out generally, like, do we solve the problem here? Just with, like, super cheap LMS address, try to just, like, have an online metric, and then we can use that also when we do rollouts of new things. We can just like compare feature flag wise, okay, does it did it improve or like did it catastrophically decline? Because one thing actually that might be interesting, that's difficult for our business is we can't look at any traces basically. We can't look at any data because it's all super confidential. And that's, like, extremely frustrating to work with because you you basically don't know what people are doing. So it's like, here's an agent, and it can do so many amazing things, and I have no idea what people are actually doing.

27:00 我本来以为你们会和律所签某种协议之类的，他们就会说：“哦，我倒希望如此。”是啊，我也希望。没有。不过问题是，他们掌握的很多数据甚至都不属于他们，对吧？是他们客户的。哦。而且这些数据高度机密，所以我们不能看。所以基本上，我们有一个评测团队，试着估算我们认为大家都在做什么，然后把这些内容纳入评测集。哇。所以这就是业务中一个特别烦人的棘手问题。你们能做释义改写吗？比如说，ChatGPT 有个类似 /feedback 的功能，能直接完整地设定它的整个轨迹。嗯。不过显然，你们是做不到这一点的。嗯。但你们可以用大语言模型来改写，比如把这个真实案例——对——做成一个足够相似的合成案例，把所有名字、所有细节都改掉，但整体上仍然——对。是啊。我们不能……这里面……我的意思是，这总有个尺度，对吧？如果你把内容一路改写到只剩下三个值，然后从中选一个，那显然没问题。如果你只是把名字删掉，那可能就不太行。是啊。我们通常会选择最安全的做法。是因为我们会尽可能严格地限制自己，因为你知道，我们必须确保完全没有风险。对。不过有些情况下，我们知道可能的失败模式大概就这十种或二十种，所以我们可以把它归类到其中一种，这就是——对——一个有用的信号。我们也可以把它归类到技术……任务类型，以及实践领域里，然后我们就能有个大致概念。这是我们一直在努力想办法用更聪明的方式解决的问题。27:00 I would have assumed you guys, like, signed some sort sort of agreement with the law firm, and they're like, oh, I wish. Yeah. I wish. No. But the thing is, like, a lot of the data that they have is not even even theirs. Right? It's their clients. Oh. And it's super confidential, so we can't look at it. And so, basically, we have an evals team that's, like, trying to approximate what we think everyone is doing and, like, build that into eval sets. Wow. So that's just, like, a super annoying kink of the business. Can you do paraphrase? So, for example, ChatGPT has, like, a slash feedback thing Mhmm. That directly fully sets its full trajectory. Mhmm. But, obviously, that would... That is not possible for you guys. Mhmm. But you can use an LLM to paraphrase, like, from this real thing Yep. Make a synthetic thing that is, like, close enough, but change all the names, change all the details, but, like, still Yep. Yeah. We can't... There's... I mean, there's always a a range, right, of, like, if you if you paraphrase all the way down to, like, three values and you pick one of them, then obviously, that's fine. If you if you just remove the names, then that's maybe maybe not fine. Yeah. It's like, we are generally pick the the most safe way. Yeah. Because, like, we we restrict ourselves as most possible because we, you know, there should be no risk at at all. Yeah. But there are things where it's like, okay, we know that the failure modes are probably these 10 or 20, so we can classify it into one of those, and that's Yeah. A helpful signal. We can classify it into the tech... Types of tasks and the practice areas also, and then, like, we have an idea. It's something that we're always trying to to figure out how to do in a smart way.

28:30 另外，关于隐私方面，我想，就算是两年前，要让律所接受这件事肯定也难得离谱。是啊。不过，嗯，他们怎么看待把数据发送给 Anthropic 这件事？是啊，是啊，是啊。这肯定是个大问题。我们的所有基础设施都在 Azure 上。是啊，主要就是因为这个。我们刚开始的时候，律所都说：“AWS？不行。”就……是啊。绝对不行。不信任 AWS。是啊，是啊。现在情况不一样了。不。不过……当时还有亚马逊、谷歌。我们的数据不能交给他们，所以我们……基本上，我们不得不把所有东西都托管在微软上，因为他们已经在用 Microsoft 365 了。哦，所以他们现在接受微软了？没错。因为他们已经针对所有文档、所有 Word 文件之类的东西，做过各种安全审查等等。所以在他们看来，微软是获批的供应商，我们就不得不把所有东西都托管在 Azure 上。这几年情况变了很多。现在大家都明白了：好吧，我们得上云，得做这些事情。当时也有更多人在谈本地部署。这种说法现在也完全消失了。大家明白了，你就该直接上云，而且它是安全的，甚至可能比你现在的做法更安全。是啊，变化很大。不过我……而且这才三年，所以这其实……是啊，是啊，变化太大了。这真是……太好了。我的意思是，真替你们高兴，不过，是啊。我是说，我没想到，那个……法律行业居然就这么忽视了这件事十年左右。是啊，是啊。我觉得我们需要 AI。28:30 Also, like, on the privacy front, I guess, I'm assuming even two years ago, it must have been, like, insanely difficult to get the law firms to be okay with this. Yes. But, like Yep. How do they think about their data being sent over to Anthropic or Yeah. Yeah. Yeah. It must be, like, a huge thing. All of our infrastructure is on Azure. Yeah. Primarily for this reason, because, like, when we started, law firms were like, AWS? No. Like Yeah. No way. Don't trust AWS. Yeah. Yeah. It's different now. No. But it... There was, like, Amazon, Google. We can't trust them within our data, so we... Basically, we had to host everything on Microsoft because they already have three sixty five. Oh, so they're okay with Microsoft now? Exactly. Because they've already done, like, whatever security review, etcetera, for all of their documents and all of their Word and whatever. So, like, Microsoft is an approved vendor in their mind, so we had to host everything on Azure, and that's changed over the past few years. So, like, now people understand, okay, we need to get on cloud, we need to do all this kind of stuff. There was also a lot more people talking about on prem back then. That's also disappeared completely. People understand that, like, you should just do it in the cloud, and it's and it's safe, and it's probably safer than what you're doing. Yeah. So that's changed a lot. But I was And by the it's only been three years, so, like, it's actually like Yeah. It's yeah. Yeah. It's a huge shift. It's it's it's it's great. Mean, I good for you guys, but, yeah. I mean, I don't expect the the, like, you know, the legal firm... Legal industry has somehow ignored this for the last, like, ten years. Yeah. Yeah. I think I think we needed the AI

30:00 给大家踢一脚，让所有人明白我们得稍微改变一下做法。好。好。另外，现在开源模型也越来越厉害了。嗯。所以，你们是怎么考虑这个问题的？嗯。我们现在的想法是，正如你说的，带着客户一起沿着帕累托前沿往前走。嗯。也就是梳理所有模型的分布图，了解它们各自擅长什么、延迟和成本如何、治理情况如何、在哪些地区可用，然后看看怎样才能综合优化这一切。有些开放权重模型很适合某些任务，我们就会在那些场景下使用它们。有些模型我们会做后训练，还有些场景就直接用前沿模型。嗯。你把这叫作模型套利。对。你们各有偏爱的模型。你之前做过一些研究，年初的时候你坚定地看好 Anthropic。对。Grok 进步很大。对。能不能举个例子，说说哪个模型整体排名不算高，但你想特别提一下，因为它好得出乎意料？嗯，我觉得它不是整体表现好。Gemini，有一款早期的 Gemini，在某些类型的转录任务上表现特别好。嗯。还有说话人归属识别。所以我们用它来做转录。你们为什么要转录？Granola？我们没有……我们可能会开发一个会议工具。嗯。不过这只是语音转语音、语音转文本、文本转语音，就像其他带这类功能的应用一样。我们也会转录文件。如果你上传听证会记录之类的文件，哦，我们就会把它们转成文本，这样我们就能据此进行所有分析了。不用 ElevenLabs？因为数据处理协议方面的原因，不用 ElevenLabs。我是说，具体来说，就是数据处理协议方面的原因。我们的数据处理协议里列出的供应商很少。嗯。30:00 kicking the butt for everyone to understand that we need to change our ways a little bit. Okay. Okay. So, also, then, on, like, open source models now are getting really good. Mhmm. So, yeah, how are you thinking about that with the Yeah. We're just thinking about, to to your point, just bring our clients along on, like, the Pareto frontier. Yeah. And, like, the of the map of all the models and what they're good at and their latency and their cost and their governance and and and their regional availability, how do we just optimize across all of that? So, some open weights models are great for certain tasks, then we'll use them there. Some will post train, and and some will just use the frontier models. Yeah. You call you call it model arbitrage. Yeah. You have different favorites. You you did some research on this, where you were firmly Anthropic at the start of the year. Yeah. Grok is improving a lot. Yeah. You wanna give an example of a model that wasn't ranked highly overall, but, like, you wanna shout it out as surprisingly good? Yeah. I think not overall good. Gemini, one of the early Geminis is, yeah, extremely good for certain types of transcription Yeah. And speaker attribution. So we use that for trans... Transcription actually. You have a Yeah. Why are you guys transcribing? Granola? We don't have... We actually probably will build, like, a meeting thing Yeah. But this is just, like, talk to speech, speech to text, text to speech, just like in in any any other app that has this. And we do actually also transcribe files. If you upload hearings or whatever Oh. We'll we'll then, like, transcribe them, put them into text, and so we can do all of our analysis on it. No ElevenLabs. No ElevenLabs for DPA reasons. Same thing as, like, the data processing agreement, I should say. So, like, our data processing agreement has very few vendors on it Yeah.

31:30 因为我们只需要把它放在首位。没错。我们得严格控制任何数据的去向。所以我觉得，嗯，这也是 Gemini 和 OpenAI 这类服务的一个好处：你来找我们，我们有你需要的大部分东西，这其实很有价值。即使它不一定是最先进的，只要对所有任务来说基本够用就行。对。那还有什么好在意的？对。就像，嗯。没错。其实我还应该提一下微软的转录模型。他们可能想要 MAI-Transcribe？是吗？好吧。它很棒。比旧版 Gemini 好。对，对，对。他们确实做出了一些不错的结果。我也不太清楚……甚至都说不上来……我觉得通常它们在 Arena 评分里都排在前二、前三、前四，这就够好了。是啊，没错。而且你还有整套治理机制，规模扩展也很容易，那大概就够好了。是啊。我想说的是，现在我们的……语音转文本模型基本已经够好了，好到你会觉得，哦，偶尔有一两个词识别错了，那也……不行？不行？当然不行。我在这方面做过很多工作。尤其是在法律领域，有些……我们碰到过这么一种情况，我记得……我忘了是不是在澳大利亚，有个称呼是用来指法庭里某个角色的，而模型总是把它转录成“避孕套”。哦。对。所以有些词你肯定不希望它识别成那样。是啊。不过……通常你会在它之上再加一层 LLM。对。然后提取一些关键词，再把这些关键词交给转录模型，让它重新转录一遍。嗯。这大概就是我目前的工作流程。对。不过，如果你真的在任何大规模场景下用过它，就会发现，只要领域不是日常对话，就会出现大量、非常多的转录错误。31:30 Because we just need to keep it first. Exactly. We need to keep it super tight on where any data goes. And so I think, no, that that is one of the benefits of, like, a Gemini and OpenAI where they're like, you you come to us, we have most of the things that you need, and that is actually very valuable. Even if it's like, it doesn't have to be, like, state of the art. As long as it's, like, mostly good enough for all tasks Yep. Then who cares? Yep. Like Yeah. No. Absolutely. I should actually shout out also Microsoft's transcription model. They may want MAI-Transcribe? Yeah. Okay. It's great. Okay. It's better than the old Gemini. Yeah. Yeah. Yeah. They they they did some good results. I don't know... Don't even know, like... I think, like, usually, all of them are, like, top two, top three, top four in terms of, like, the the Arena scores, and that's good enough. Yeah. Exactly. And it's like, you have all the governance and it's super scalable, it's like, then that's probably good enough. Yeah. I was gonna say, like, at this point, our... Pretty much the speech to text model is, like, good enough to where you're like, oh, one word here or there, if it's off, that's No? No? No. For sure. Like, I've done this a lot. And particularly in legal, there's like some... We we had this thing where I think it's... I forget if it's like in Australia where there's like a certain name for like a a person in court, and they would consistently prescribe it as condom. Oh. And so there... Yeah. There's there's certain things that you don't want there. Yeah. But... So usually, you run an LLM layer on top of it Yeah. And then you you extract some keywords, and then you run the transcription model again with the keywords. Mhmm. Like, this is kind of my my workflow so far. Yeah. But, yeah, if you... If if you've just done this on any large scale, you will pick up tons and tons of transcription errors on any domain that is not, like, average day to day conversation.

33:00 是的，当然。如果里面有任何术语，那就会很难。是啊。我觉得 Whisper 和 ElevenLabs 现在还是会把 Claude Code 转录成 Cloud Code。对，就……我其实也遇到过这种情况。很常见。是啊，所以这就是你能看出来的地方，还有很多可以改进。没错，当然。我确实想稍微聊聊智能体产品这方面。我们已经介绍了一些比较高层面的东西，对吧？我们已经提到了 Job Tabular Review。嗯哼。还提到了 Jobs Intake。对。你们还有其他一些东西，我们甚至还没谈到。不过我想先讲讲两大重点，就这么说吧。对。说它们是两大重点公平吗？智能体和表格审查？还有……还有项目。大概是吧。对，对。还有我们的 Word 插件。好吧，这也是一项……因为很多人都在 Word 里工作，所以我们的 Word 插件非常重要……还有 Outlook 插件也是。对。所以我想聊聊大家必须了解的那些重点功能。没错。表格审查基本上就是把任务分发到很多文档上。你可以针对所有文档提出同一个问题。没错。在表格里。对，然后来回处理。然后可以……对。一百万很惊人。还有什么其他工程方面的考量或性能问题吗？比如……这个有什么难点？在聊这个之前，你能先讲讲律师的用户体验是什么样的吗？然后，比如说……好吧。其实，我想用一下表格视图？对，对。比如说，他们用表格审查时会做些什么？对。然后我们再聊这个。33:00 Yeah. Absolutely. If there's any sort of jargon in there Yeah. It's gonna be tough. Yeah. Like, I think Whisper and ElevenLabs still transcribe Claude Code as Cloud Code. Yeah. And, like, it's Yeah. I've I've actually had that. Very common. Yeah. So so, like, that's that's how you know it. There's, like, a lot more to improve. Yeah. Absolutely. I do want to cover a little bit on the the agent product side. So we've covered a few of, like, the of the top level stuff. Right? We've already named job tabular review. Mhmm. We named jobs intake. Yep. And you've you've got some other stuff that we haven't even touched on yet. But I just wanna cover the the big two, let's call it, like Yep. Is it... That is is that fair to say that they are the big two? Agent and tabular review? And our And projects. Probably Yeah. Yeah. And our Word add in Okay. Is also something... Because that... A lot of people work work inside of Word, and so our Word add in is is a huge... And and Outlook add in also. Yeah. So I just wanna cover, like, the the big ones that everyone must know. Yep. So tabular review is basically like fan out of, like, a lot across each document. You can ask the same question across all all documents Exactly. On the table Yeah. And then work back and forth. And then can Yep. Million is impressive. What's... Any other, like, engineering considerations or performance things? Like, what... What's hard about this? Before we do that, can you just, like, walk through, like, what is a user experience for a lawyer? And then, like Okay. Actually, wanna use tabular view? Yeah. Yeah. Like, so, like, what are they doing, like, with tabular review? Yeah. And then we can go into that.

34:30 当然。对，对。所以他们……通常……我的意思是，一个很好的例子就是尽职调查，对吧？比如说，我们需要审查所有劳动合同、所有供应合同，或者其他所有文件。比如劳动合同，你把 10,000 份劳动合同放进去，然后你……每个单元格是一份劳动合同吗？每一行是一份劳动合同。然后，你针对各个单元格提问，基本上就是这样。比如，第一个问题……第一……不，抱歉，是每一列。第一列可能会问：终止通知期是多久？然后系统就会把每份合同的结果都填出来，比如都是三十天，或者实际的期限。下一个问题可能是：有没有知识产权转让条款？再下一个可能是：如果我们解雇他们，需要支付多少？或者他们的薪酬是多少？然后系统就会把整张表都填好。你也可以直接让代理来做。比如你把 10,000 个文件直接丢给代理，然后说“对这些文件做尽职调查”，它就会创建一个表格化审查，逐一处理所有文件，等它完成后再分析结果，然后标出一些风险点。好。比如，CEO 的薪酬高得离谱，或者有某种奇怪的奖金结构。好，那显然我们得看看这个问题。代理还可以据此起草一份备忘录发给客户；或者它只负责审查，然后由律师，也就是用户，来查看，比如说：“好，我们来看看这里的风险点。”还有协作功能，其他人可以在上面评论，等等。那你们为什么选择电子表格这种格式？对。它就是……好用得多。它……用起来特别方便……在里面操作很方便。你可以很快概览情况，也很容易进行审查。我们生成的所有内容也都带有指向源文件的引用，所以你基本上可以非常轻松地进行核查。如果你只是想确认一切都没问题，也可以标记单元格，比如标记哪些已经审查过，哪些存在风险，然后你还可以发表评论、进行协作。这比我们尝试过的任何其他形式都方便得多。对。34:30 Sure. Yeah. Yeah. So they... It's typically... I mean, a good example is just a due diligence, right, where it's like, we need to review all employment contracts, all supply contracts, all of whatever. And so employment contracts, you put 10,000 employment contracts in there, and then you Like, each cell is one employment contract? Each row is one employment contract. And then for cells, you ask questions, basically. So, like, the first question... The first... No, sorry, per column. So the first column might be, what's the termination period? And then it's gonna, like, fill out for everyone. It's thirty days or whatever it is. And the next question is, like, is there an IP assignment clause or whatever. Next one is, like, how much, you know, how much do we have to pay them if we fire them, or whatever it is, like, much are they getting paid? And then it's gonna fill out the entire thing. You can also just ask the agent to do this. Like, if you just drop 10,000 files into the agent and say, like, do a due diligence on this, it will spin up a tabular review and run all of them and wait for it to be done, and then analyze the results, and it can then highlight, like, red flags. Okay. This... The CEO is getting paid an outrageous amount or there's, like, this weird bonus structure. Okay. Then, like, obviously, we need to look at that, and then the agent from that can draft a memo to send to the client, or it can just do the review and then have, you know, the lawyer, the user, just be like, okay, let's let's look at this, here are the red flags. There's also collaborative slide, you can comment on it for other people to look at, etcetera. And why do you choose that, like, spreadsheet format? Yeah. It's just a lot better to... It's it's super easy to work with... Work in. It's really easy to get an overview super quickly of what's going on and to review it. So all of the stuff that we produce also has citations back to source documents, and so you can basically review it super easily. If you wanna just make sure that everything's right, you can also, like, highlight cells, these are reviewed or whatever, these are the ones that are flagged and you can comment and collaborate. That's much easier than any other form factor that we've tried. Yeah.

36:00 我得说，电子表格这种形态在某些领域确实很有吸引力，但不知怎么的，始终没真正普及开来。嗯嗯。我……我觉得这挺奇怪的。比如 Clay，在 GTM 领域也是这样，你知道，它因为电子表格而非常出名。没错。你们……我不太确定还有没有别家在做大型的电子表格 AI 产品。没有。没有。之前有个 V7 之类的，试着推出过……还有 Sheets，还有……就连在金融领域，我也想不出什么特别知名的、专门做电子表格的公司。这真的挺……是啊。没有，我同意。我觉得这真是个很棒的……如果你能扇出处理，做完全相同的事情，而且……提取信息，那电子表格真是很棒的形态，能让你了解发生了什么。你们觉得用户必须导出到 Excel，还是说大家很乐意直接在你们的电子表格里工作？大家很乐意待在里面。你们用 AG Grid，还是自己做的？我们用 AG Grid。AG Grid。看起来它是领先的那一个。我是说，他们做这个有……我不知道，十五年了。是啊。是啊。他们知道怎么……是啊。所以……好吧。是啊。我……我很喜欢这个……电子表格的角度。我是说，对我来说，工程方面唯一另一个有意思的事情就是扇出。没错。工程师也许……很擅长理解其中的权衡，比如，好吧，在对一百万份文档进行扇出处理之前，也许先拿两份试试。36:00 I will say that the spreadsheet form factor has has a lot of pull in some areas, but somehow hasn't really spread. Mhmm. I'd... I find I find it very weird. Like, Clay is another one on the GTM side that, you know, very known for the spreadsheets. Yep. You guys... I don't really know if anyone else does large spreadsheet AI thing. No. No. There there was, like, V7 that, like, tried to launch... There's, like, sheets, there's... Even in finance domain, I don't really have, like, a a top name that, like, does spreadsheets. That's just really well... Yeah. No. I I agree. I think it's, like, a great... If you if you have fan out and you do the exact same thing and you... You're extracting stuff, it's it's such a great form factor to, like, understand what's going on. Do you find, like, you must export to Excel or, like, people are happy to live in your spreadsheet? People are happy to live there. Do you use AG Grid or your own We do use AG Grid. AG Grid. That seems to be like the leading one. I mean, they've done this for, I don't know, fifteen years. Yeah. Yeah. And they know how to... Yeah. So so... Okay. Yeah. Mean, I'm I'm happy on this... That spreadsheet angle. I mean, I think the the only other interesting thing for me engineering wise is the fan outs. Yep. Engineers maybe can... Are well equipped to understand the trade offs of, like, well, okay, before you use fan out on a million documents, maybe test it on, like, two.

37:30 嗯哼。对。然后……对。对。你放开手脚之前，这样能行吗？对。他们会这么做吗？会的。我们的表格视图实际上会先运行前 10 行。对。先检查一下。对吧？对。这说得通吗？它显示的是我们预想的内容吗？对。这……真的很不错。哦，而且对使用者来说，搭建这个东西时，能先针对这 10 行快速迭代，直到……得到想要的结果，然后就可以放开手脚了，这真的很不错。对。我很乐意接着看下一个东西：受理流程，也就是你的 Slack 机器人。对。对。大家都有 Slack 机器人。你的有什么不同？我们的机器人只是专门面向法律场景的。它实际上并不是……也可能……我们……可能也会做一个只作为智能体接口的版本，但这个机器人实际上是专门用来启动智能体，然后由智能体把任务添加到我们的清单里。我们有一个清单，类似 Linear，嗯……这是最接近的类比。它会把任务加到一个专门的受理清单里，然后律师和智能体就可以进入 Legora，说：“好，这些就是大家发给我们律所，或者大型企业法务团队的事项。”我们可以为这些事项启动智能体、协作处理，然后把它们标记为已完成，结果就会发回 Slack。所以它更像是……它不只是启动智能体，而更像一个……我不知道，类似服务入口的东西，比如：“嘿，这是工单，我想让你审核这个。”好。然后事情就……处理完了。结果回来了。不需要任何人再做修改，法务团队只要待在 Legora 里，管理智能体和其他一切就行。我觉得这种东西的数据结构很有意思，因为它基本上就是一个收件箱，对吧？里面有各种各样的请求，这些请求可能会转化成一项任务，或者变成别的什么东西……嗯，我不知道。把这些都和其他功能集成起来很难吗？还是说你们根本不担心这个？37:30 Mhmm. Yep. And... Yeah. Yeah. Does that work before you go nuts? Yep. Do they do that? Yeah. They will. Our tabular view actually will run the first 10 rows. Yeah. Just check. Right? Yeah. Does this make sense? Did it did it show us what we thought it was gonna show us? Yeah. That's a... That's really good. Oh, but also for people, when people build this, it's just really good for them to iterate quickly on the 10 rows until, like, they have a... They're getting the results that they want, and then they can go ham. Yeah. I am happy to move on to the next thing, intake, which is your Slackbot. Yep. Yep. Everyone has a Slackbot. What's different about yours? So ours is just legal specific, and what what happens with ours is it's not actually... It might also... We... We'll probably also build one that's just an interface to the agent, but it's actually specifically to basically kick off an agent that then adds a task to our list. So we have this list thing, which is like linear, for for... You know, that's the closest equivalent, and so it will add it to a specific intake list, and then lawyers and agents can go into Legora and just be like, okay, these are the things that, you know, people are sending to us as our law firm or as our legal team in a big corporate. We can kick off agents for them, we can collaborate on them, and then we can just like mark them as done, and then it gets sent back in Slack. So it's more of a... It's not just like kicking off an agent, it's more of a like, I don't know, storefront thing where it's like, hey, here's the ticket, I want this thing reviewed. Okay. Now it's not... Here it is. It's back. No one needs to change anything, and the legal team just sits in Legora and, like, manages the agents and everything there. I think the the data structure for this kind of thing is really interesting because it's, like, basically the inbox, right, for all these kinds of queries that can potentially translate into a thing, into something else, into And, something like, I I don't know. Is it hard to integrate all these with all the other features, or do you just not worry about it?

39:00 这……不，不是。我觉得我们有一些非常好用的基础组件。比如说，列表就是一个很棒的基础组件，因为它本质上……非常灵活。也有点像 Notion 数据库。对，像 Notion 数据库。基本上，你可以按照自己的需求随意配置它。当然，我们也投入了很多精力，让它极具可扩展性、运行得非常好。你可以在上面设置触发器和自动化流程，它就像一种很棒的黏合剂，能把所有东西串联起来。因为这样一来，只要列表里发生了某件事，你就可以据此启动一次智能体运行；然后智能体会把结果传回来，填入某个字段，或者添加新文档之类的。所以，这就像你可能会用 Linear 或某种任务管理工具来协调一群编程智能体一样，我们也可以用列表来做类似的事。智能体可以创建列表，列表也可以创建智能体，智能体还能读取列表，于是整个智能体集群就活起来了。是啊。人们会对此感到兴奋，还是会觉得不知所措？你知道，我觉得……是啊。这基本上就是面向终端用户的编程。你……你知道，你可以让人们搭建各种复杂的自动化流程。有些人会对此特别着迷，然后变成……你知道，要是能有……是啊，我们有意见领袖。是这样值得吗？会不会太复杂？即使我们实际上不把它开放给终端用户，我们也需要这种灵活性。假设这些功能完全只在内部使用，那么这对我们来说就是一种非常快捷的方式，让我们的法律工程师利用它快速构建大量产品。比如，你可以构建定制的基金管理仪表板。基本上，你可以用列表和自己定制的生成式 UI 来搭建它，然后就得到一个新产品，有些人会非常喜欢。我们还可以建立一个由大量专业产品组成的市场，并迅速扩大规模，因为我们的法律工程师都可以直接构建新产品。比如他们跟客户沟通时，客户说：“哦，你们能给我做一个专属的吗？”没问题，我们马上给你定制一个。所以，它有点像 Legora 内置了一个 Lovable。如果我们不想把它开放出去，它也是我们构建自定义功能的方式。我们还在英国收购了另一家公司，叫 Wexler，他们做的是事实智能，主要面向诉讼场景。比如说，你为某起诉讼上传一百万份文档之类的，我们会从中提取事实，包括命名实体、日期等；然后，等你收集齐所有事实后，就可以把它们放在一起比对，看看哪些事实相互印证，哪些存在……怎么说呢，矛盾。39:00 It's... No, it's not. I think we have some primitives that work really well. So, like, lists is a great primitive because it's essentially... It's extremely flexible. Also like Notion database. Yeah. A Notion database. It's like a Notion database, and so you can basically configure it for whatever you want, and we, of course, invest a lot into making this super scalable and work really well, and you can have triggers on them, automations on them, and it's like a great glue to glue everything together, because then you can kick off an agent run based on a list if something happens, and then that comes back and populates a field or whatever, adds new documents or whatever it is, and so it's like just like you might coordinate a bunch coding agents with linear or some sort of task management tool, we can do it similarly with lists. And so agents can create lists, and lists can create agents, and agents can read lists, and so it's like, the agent swarm comes alive. Yeah. Do people get excited or overwhelmed by that? You know, like, I I feel like Yeah. This is basically end user programming. You're... You know, you can let people set up crazy automations. Some people get super obsessed by it, and they become, like, like, you know, it'd be cool to have, like, Yeah. We have influencers. Yeah. Is that worth it? Too overcomplicated? We need the the flexibility, even if we don't actually expose it to the end user. Let's say this was just all internal, then it's a really fast way for us to utilize people our legal engineers to basically build a lot of products extremely fast. So you can build custom fund management dashboards. You can basically build that on lists with your own generative UI, then you basically have a new product that some people are gonna love, and we can have a marketplace of just a bunch of specialized products that we can scale super quickly, because all of our legal engineers can just build new ones. If they're out with a client, oh, you want your own? Easy. We'll just fix one for you, so it's like a Lovable inside Legora a little bit, and if we don't wanna expose it, that's also how we built our own custom feature. So we we made another acquisition in the UK called Wexler, and they're doing fact intelligence, which is for litigation mostly, which is like you upload a million documents for a lawsuit or whatever, and then we extract facts from those with like named entities and dates and whatever, and then when you have all the facts, you can basically hold them up against each other and see which ones are corroborated and which ones are, what do call it, disagreement.

40:30 他们，你知道，这非常强大，因为这样你还可以转录法庭听证会，然后你就可以说，哦，他说了这件事，但实际上这边的邮件却说他当时不在场。如果你能实时看到这一切，那真的他妈的太酷了。是啊。而且我们也可以基于这个基础构件继续构建。所以，基本上，我们可以围绕这一个基础构件打造很多很酷的东西。这就是我们看待它的方式。我们必须……随着公司规模扩大，我们必须解决很多很多问题，所以我一直在想，我们要怎样构建一个核心基础构件，让它能够扩展到许多不同的终端产品，并创造出高度可定制的体验。好。所以你们有……你们有哪些核心基础构件？我们给大家重新梳理一下。这个是一个核心的。对。哦，它是一个核心的可配置构件。对。项目。对。基本上就是文件容器，也就是文件系统。对。当然还有代理，你可以启动它，并以各种不同的配置来运行它。40:30 They, you know, and that's extremely powerful, because then you can also transcribe a court hearing, and you can like, oh, he said this thing, but actually, the email over here says that he was not there. And it's like, if you see it live, it's like so fucking cool Yeah. And that we can also build on this primitive. And so, like, we can build a lot of really cool things on this one primitive, basically. That's how we think about it. It's like we have to be... As we scale the company, we just have to solve so many things, and like, so so I think a lot about how do we build a core primitive that we can scale to to many different end products and create very customizable experiences. Okay. So you have... What what core primitives do you have? Let's just restate for for people. This a core one. Yeah. Oh, it's core configurable one. Yeah. Projects Yeah. Basically containers of files, file systems. Yeah. The agent, of course, you can kick that off and run that in very different configurations.

42:00 顺便问一下，你们给这个智能体起名字了吗？没有。我们没有。没有。好吧。是啊。没有。你是说，比如我们叫它 Leya，还是说给它起了个名字？没有。是啊。是啊。Leya 这个名字挺不错的。是啊。是啊。以前人们会把 Leya 当作一个有名字的实体来谈论。是啊。不过后来我们就不这么做了。好吧。也许我会把这个名字再用起来。可能会让人困惑。我也不知道。所以，列表、表格智能体审查、表格审查。我们正在考虑的，是这实际上算是一种……实例化。你想把它细分到多具体都行，对吧？不过那些可能是最主要的几项。是啊。是啊。其实我想聊聊 FDE，因为这听起来有点像……是啊。Max 也稍微聊过这个。就是，好的，当你引入一个新的大型客户时，你想尽快让他们获得价值，这样他们就会继续留在……没错。继续用 Legora。对。是啊。这方面是怎么发展的？你们是什么时候开始组建前沿部署团队的？我们聘用了第一位前沿部署工程师。那也是我们第一次去纽约，大概是九个月前？他也是 YC 创始人。我其实不太确定。他们算法律工程师吗？不算。好吧。是真正的工程师。不过，我们确实……我会思考的一件事是，究竟该如何让这件事规模化？我认为一个非常核心的方面是：我们如何赋能法律工程师？如何赋能我们的高级用户和法律工程师，让他们能够以强大的方式定制产品？当然，我们仍然会有前沿部署工程师来提供帮助，比如完成各种定制集成、界定范围、弄清楚哪些事情可行，并把所有这些反馈到核心产品开发中。所以我大致是这么考虑的：这有点像第一道、第二道防线。法律工程师应该能够定制很多东西，打造出色的体验；而在他们做不到、或者我们需要对核心产品进行调整、遇到客户特定情况，或是其他问题时，就由前沿部署工程师来处理。他们也会稍晚一些加入，有时他们只会做产品，有时他们会和……一起工作，也会和……一起。42:00 Do you have a name for the agent, by the way? No. We don't. No. Okay. Yeah. No. You mean, like, we called it, like, Leya or it's like we gave it a name? No. Yeah. Yeah. Leya would be cool. Yeah. Yeah. People used to talk about Leya as, like, an a named entity. Yeah. But, yeah, we we stopped that. Okay. Maybe I'll bring it back. It might be confusing. I don't know. So list, Tabular agent review Tabular review. Which, like, we're considering is that actually an instantiation of... Like, you can take this as far down as you want. Right? But those are probably the big the big ones. Yeah. Yeah. So I actually wanted to talk a little bit about, like, FDEs because this sounds kinda like Yeah. That, and I think Max has talked about that a little bit. That, like, okay, when you onboard a new, like, big customer, you wanna get them value, like, ASAP so that, you know, they stay on Exactly. Stay on Legora. Yep. Yeah. How has that evolved? Like, when did you guys start that forward deployed team? So we hired our first forward deployed engineer. That was our first time in New York, also YC founder, nine months ago maybe? I'm not actually sure. They considered legal engineers or no? No. Okay. True engineers. But we do... It it is something that I think about is, like, how do we exactly enable that at scale? And I think a a really core thing is, like, how can we empower our legal engineers? Like, how can we empower our power users and our legal engineers to customize the product in a powerful way? And then, of course, we we will still have forward deployed engineers to, like, help out, you know, do all the custom integrations and the the the scoping and figuring out what's possible and feed all of that back into core product development. So that's kind of how I think about it. It's like almost a first line, second line defense, whereas, like, legal engineers should be able to customize a lot of stuff and build a great experience. And then where they can't or we need to make core product changes or customer situations or whatever, have forward deployed engineers who also then cycle in a little bit behind, like, sometimes they'll just work on the product, and sometimes they'll work Can with

43:30 你能不能讲讲，比如说一个例子？比如，好吧，你们刚接到一个新客户，第一步是什么？你们会不会说，好，我们有一位前线部署工程师为你们服务，他会去开发……对。为你们开发某种产品？你们是怎么做的？ 这在不同客户之间差异很大，对吧？对。但如果是大客户，而且他们非常投入，也有很多想做的……他们有各种各样的设想，想要实现；或者他们已经……通常，他们已经试着开发过一些东西，但失败了。他们会说，这是我们……这是我们想解决的问题。然后我们基本上会给他们安排法务工程师，也可能安排一位前线部署工程师，具体要看他们的需求是什么。之后我们就会说：嘿，这些人会过来，帮你们取得成功。他们会和你们所有律师一起工作，教他们怎么使用产品，也会了解你们目前的各种工作流程，然后把这些流程整合进 Legora。有些工作只需要法务工程师来做，有些则需要前线部署工程师。 我想问一下，两者具体怎么区分？你能稍微讲讲吗？ 法务工程师会像高级用户那样配置产品。他们不写代码。对。对。而前线部署工程师基本上会编写代码，并部署到生产环境中。我想这就是……对。前线部署工程师通常有什么样的背景？在……他说，YC 创始人。对。创始人、普通工程师，43:30 you walk through, like, an example? Like, okay, so you just got you just got this new client, like, what's step number one? Like, are you guys like, okay, hey, we have a forward deployed engineer for you guys, they're gonna go build Yeah. Some kind of product for you, like, how does So that it's very different client to client, right? Yeah. But if it's a big client and they're super engaged and they like want... They have all these dreams of things that they would want to do, or they already have... Typically, they have already tried to build something and it's failed, and they're like, this is what we... This is a problem that we'd like to solve, then we'll we'll basically staff them with legal engineers and a forward deployed engineer potentially, depending on what, you know, their thing is, and then we'll just be like, hey, these guys will come, they'll make you successful. They will sit with all your lawyers and teach them how to use it and, like, figure out all the workflows that you have right now and and build them into Legora. And some of that will just be a legal engineer, and some of that will require a forward deployed engineer. What is that separation, I guess? If you could, like, talk about that a little bit. Legal engineers configure the product as a power user would. They don't write code. Yeah. Yeah. And a forward deployed engineer writes code into production, basically. I guess that's the the Yeah. What kind of background do the forward deployed engineers normally have at the He said, YC founder. Yeah. Founders, regular engineers,

45:00 以前。对，对。D 是你们的……对，没错。D 是我们第一个。对。两家 YC 公司，还在 Meta 做过搜索。对，她很棒。好吧，这标准可真高。她很优秀。对，对。她是我们在美国落地的第一个人。对。那么这些人都在为客户构建什么样的工作流或功能？你们有没有一些……比如说，有些用例是所有律所都会说的：“这是我们的 FDE 入驻后要做的第一件事。”情况很不一样。好吧。有一些大家都需要的核心工作流，我们可以直接在 Legora 里搭建，内容可能包括“我们如何处理这类保密协议”之类的。然后你只要启用就行了；说到底，这更像是为 Elise 构建技能。大多数 FDE 做的则是另一类事情：有些功能我们现在还做不了，比如某些基础组件还需要更灵活一些，或者我们需要接入一个特别复杂的系统。然后他们就会去现场，基于我们的 API 构建集成，可能还会扩展我们的 API 之类的，让它能正常运行。那法律层面的安排是怎样的？因为你说你们自己甚至不能参与处理案件。对，对。那么他们怎么参与客户项目？你是说律所怎么参与客户项目吗？怎么……抱歉，我是说你们的 FDE 怎么参与客户项目？他们还是不会接触数据，只会搭建系统。好吧。但有时候他们也会像员工一样办理入职手续。比如，可能会给他们配一台笔记本电脑之类的。去律所吗？对，对，因为他们需要这样。挺有意思。这种情况不太常见，因为通常他们只要搭建好集成，它就能正常运行了。不过，如果有需要，他们也可以这么做。45:00 old Yeah. Yeah. D was your Yeah. Exactly. D was our first. Yeah. Two YC companies and worked on Search at Meta. Yeah. She's awesome. Okay, that's a high bar. She's great. Yeah. Yeah. First first boots on the ground in the US. Yeah. So what are some examples of, I don't know, workflows or things that these people are building for the clients? Or do you guys have, like, oh, these are like things. You know, there's a couple use cases that, like, all the law firms are like, you know, this is the first thing that our FDEs are gonna do when they when they go in. It's very different. Okay. Some core workflows everyone have, and we can just build in Legora, and that could be anything from, like, this is how we handle these types of NDAs or whatever. And then it's just like, you activate it, and it's more like skill building, really, for Elise. And most of the FDEs, that's more like, there's something that we can't do yet, like we need some more flexibility in some of the primitives or whatever it might be, or we just have this really complex system that we need to integrate into, and then they'll go out there and they'll build the integration against our APIs and maybe expand our APIs or whatever to make that work. And how does the, like, legal structure work there? Because you said you guys yourselves can't even work on deals. Yes. Yeah. So, like, how do they work on client projects? You mean the law firms, how do they work on client projects? How are Sorry. How how do your FDEs work on client projects So they don't touch the data still. So they'll just build the system. Okay. But they will also be onboarded like an employee sometimes. So, like, they might get a laptop and everything. Like, to the law? Yeah. Yeah. Because they need to. Interesting. That doesn't happen very often because, like, usually, they just build an integration and, like, kinda works. But they, yeah, they can if they need to.

46:30 我猜你们只有在客户规模达到一定程度时，才会向客户派驻 FDE？是的。是的。好。这两种情况都说得通。因为我们的 FDE 也可以参与核心产品的开发，对吧？所以这就像是……需要不断做优先级排序。有没有什么规则？比如说，是不是每个……你知道，Palantir 很有名的一点是，工程师四天在客户现场工作，一天回来为产品做贡献。你们有类似的安排吗？ 没有，我们没有。说实话……坦白跟你们说，这方面很多事情还处于早期阶段。那你觉得未来会是什么样？比如接下来一年？我确实觉得会更……可能不会完全像 Palantir，但我觉得我们会有一支规模更大的 FDE 团队，持续轮转。也许他们会在一个客户那里待两个月左右，把某个项目做出来，然后再去下一个客户那里，诸如此类。是的。不过我确实设想过，我们带着法律工程师和前沿部署工程师一起进场，专门负责具体项目。 是的。API 业务其实挺有意思的，就像你说的，因为你知道，那也是另一种选择：你们可以成为 Legora 平台的 CTO 和维护者。没错。对吧？我们会把 Legora 网页应用里的所有功能都提供为 API，然后你们可以随意使用。你们这是在做出承诺吗？还是说…… 这不是我们现在承诺的事情。是的。我们确实有可以供你们开发的 API，但并没有覆盖产品的全部功能。这就挺值得琢磨的：这是不是你们的策略？或者……当然，有些人会很喜欢，因为他们不用学习新工具，可以直接用现有的工具，对吧？然后他们只需要……比如安装 Legora，就能使用 Word 文档相关的功能。除此之外，我确实觉得，这样一来你们就得自己维护它了。46:30 And you only onboard FDEs into a customer, I'm assuming, if they're a certain size? Yeah. Yeah. Okay. And these two make sense. Because our FDEs can also work on core product. Right? So it's like... It's a constant prioritization thing. Is there a rule, like, I guess, does every... Like, know, so Palantir famously had, like, you know, four days you spend on-site, one day you come back and contribute back to products. Anything like that? No. We we don't. I am... To be fully honest with you guys, it's still it's still, you know, it's still early days in a lot of this. I mean, where do you see it headed? Maybe over the next year? I do think that it's gonna be more... Maybe not exactly like Palantir, but, like, we... I think we will have a larger, in constant circulation group of FDEs that maybe sit with a client for, like, two months or whatever to build out this one thing, and then they go to the next one, etcetera, etcetera. Yeah. But I I do imagine us rolling in with legal engineers and forward deployed engineers to, like, work on specific projects. Yeah. The API business is actually interesting, like like you said, because, know, that's another rule that you can have, where you will CTO maintainer of the Legora platform. Yep. Right? Like, we will offer everything that we have in the Legora web app Yep. As an API, and then you guys can use whatever. Is is that is that a promise that you make? Or That's not something we're promising now. Yeah. We do have APIs that you can build against, but not the entire surface of the product. That's the curious thing, like, is that the strategy? Or... And obviously, you know, some people will love it because I don't have to learn a new tool, I can just use the existing tool I have. Right? And then I just, like Yeah. Install Legora, for example, with the the Word document stuff. Other than that, you know, I I do think, like, then you have to maintain it.

48:00 没错。那就是……这一直是我的想法。这就是为什么我们一直尽量不这么做。我们以前没有 API。我的意思是，我们大概六个月前才开始做 API。我们很长一段时间都在抗拒，主要是因为，嗯，我就是不想……我们的产品迭代太快了，我不想因为必须维护旧版 API，而被那些已经改过的数据模型拖慢。我的意思是，现在我们规模更大了，情况也不一样了，所以现在我们可以承担这笔成本。但当我们只有 20 名工程师时，我们的迭代速度实在太快了，根本做不到。我得强调这一点，对吧？总的来说，所有公司都有平台团队和产品团队是件好事，因为产品团队你已经有一个了，只要平台和产品没有绑得太死，就很容易再组建第二个。对，对，对。你知道，一个很简单的例子就是 Web 应用和移动应用，对吧？或者 Web 应用、移动应用、RTP，然后还有其他各种集成。这样一来，你的 FDE 也不过是整个这个平台的又一个客户端而已。对。我觉得，甚至可以用这种方式来看待智能体。你可以有一个运行智能体的框架团队，或者类似的团队，然后你还可以围绕它构建产品，比如加入技能、自动化，以及其他各种东西。对。这就是亚马逊云服务的著名原则。嗯。贝索斯当时的意思是，所有东西都要通过 API 提供。不再允许后门访问。你只能向其他团队提供 API，而且必须是和向客户提供的同一套 API。嗯。如果还没有，那就自己建。对。然后客户也能用。对。我觉得这方面现在也有些变化了。现在很多 API……我们正在试着探索的一件事是：智能体能不能充当 API？我知道这听起来很奇怪，但对其中一些事情来说，他们真正想做的只是触发某些操作。对于其中一部分，你完全可以……你只要……我不是说我们正在这么做，但我一直在琢磨的是：能不能就这样——好，你触发智能体，让它去做这件事？显然，这比直接有一个随时可用的 API 效率低得多，你知道的……但我一直在琢磨的是，我们的产品里有多少部分可以直接由智能体来实现，而不是靠其他所有那些东西？48:00 Exactly. That's that's... So that's been my thing. That's why we've tried not to do it. We haven't we haven't had APIs. I mean, we started our APIs, like, six months ago or something. We fought it for a very long time, mostly because, yeah, I just don't wanna... We we move so quickly in in the product that I don't wanna be, like, slowed down by having to maintain old API versions for, like, a data model that we've changed. I mean, now we're bigger and things are different, so now it's it's a cost that we can take on, But when we're 20 engineers, it's just like we we move so quickly that we can't do this. I have to point with this. Right? Like, it's just generally good for all companies to have a platform team and a product team because the product team, you have one, but then you can easily spin up a second one as long as the platform is not super tied to product. Yeah. Yeah. Yeah. You know, like, a very simple one is web app mobile app. Right? Or, like, web app mobile app RTP and and then, like, you know, whatever, like Yeah. Other other integrations. Then your FDEs just are one more client to the overall platform anyway. Yeah. And I think, like, even the agent, you can think about it this way. There's, a harness team or, like, a... That runs the agent, and then you can even... You can build a product around it with, like, skills and automations and whatever else. Yeah. This this is, like, very famously the Amazon web services mandate. Mhmm. Like, Bezos was like A of things in the API. No more backdoor access. You only offer other teams your APIs, the same APIs you offer customers. Mhmm. If you don't have it, you build it. Yep. And then the customers also get it. Yep. I think I think that's also changing a little bit. A lot of a lot of APIs now... Like, one thing that we're toying with is, can the agent be the API? I know it sounds weird, but, like, for some of this stuff, what they actually want to do is they want to trigger certain things to happen. For some part of this, it could... You could just be... I'm not saying that we're doing this, but it's like a thing that I'm toying with in my head is, could that just be like, okay, you trigger the agent and ask it to do this thing? Like, obviously, it's way less efficient than if you just had like an API that you always had and it's like, you know, but it's something that I'm toying with is, like, how much of our product can just be the agent versus all the other stuff?

49:30 嗯，你知道，无限制的执行，以及成本和延迟。这就是……这就是……这就是全部的问题吗？就是……对。然后我觉得还有确定性的问题，比如说，一百次里可能有一次，或者两次，它会做出一些我们不希望它做的离谱事。我们稍后会谈到测试框架和模型微调。不过我觉得这里还有一个产品层面的问题，就是克制。嗯，也就是哪些事情不做。没错。因为这些你都可以做，显然这就涉及优先级问题。比如说，你们是不是更倾向于拒绝大多数需求，只做少数几件事？还是说，不，我们其实想做更多，只需要再招一百个人？很长一段时间里，我都非常坚定地支持前一种做法。干脆什么都拒绝，比如让工程团队保持……我想我以前说过，在我加入 Legora 半年后的某次演示里，我放出了一张三百斯巴达勇士的幻灯片，就是电影里的那群人，然后我说，我们公司未来一直只会有二十名工程师，而且我们会打败那些规模大得多的公司。现在我们有一百名工程师了，但感觉人手还是远远不够。我本来想那么做，但我觉得在现实世界里，竞争非常激烈，大家想要的东西太多了。如果我做个清教徒，坚持说不，我们必须保持二十个人，只能做……之类的，那我们就会输，大家也会转投别家。所以我们必须做更多东西。因此我觉得我的工作就是想办法建立一套体系，让我们能做很多东西，但实际上只……就是让客户觉得我们做了很多东西，但实际上我们只做了少数几件。这就是我现在对自己工作的理解。49:30 Well, you know, unbounded execution and cost and latency. That's the... That's... Is that the only whole bag? That's... Yeah. And then I think deterministic, like, maybe in, you know, a 100 out of or in two out of a 100 times, it's gonna do something whack that we didn't want it to do. We were gonna go into, like, harness and and model fine tuning in a bit. But I think there's a there's a product question about restraint Mhmm. About what you don't do Yep. Because you could do all that, and obviously, there's a prioritization issue. Like, are are you at... More on a point of let's say no to most things and then only do a short number of stuff? Or no. Actually, we wanna do more things, and we just need to hire, like, 100 people? I was I was very strongly the first thing for a very long time. Let's just say no to everything, like, keep the eng team... I I think I've said this before, but, like, one of our presentations that, like, after I've been at Legora for half a year, I I pulled up this slide of, you know, the 300 Spartans and and and Yeah. For the movie, and I was like, we're just gonna be 20 engineers for the rest of the company's history, and, like, we're gonna beat everyone who's much bigger. We're a 100 engineers now, and it feels like we're way too small. I wanted to do that, but I think in the real world, we're in, like, fierce competition, people want so much stuff, it's like, I could be a puritan and just be like, no, we have to stay 20 people and we can only build, you know, whatever, but then we just lose, and, like, people would go elsewhere. So we have to build more, and so I think my job is to figure out how do I build a system where we can build a lot, but actually only... Like, where we can... The customer feels like we've built so much stuff, but we've actually only built a few things. That's that's how I think about my job right now.

51:00 是啊。我是说，基础构件这个方向是对的，你懂吧？是啊。没错。我觉得这会……真的很强大。你可以测试新产品，看看有没有人买账，然后你就会有点……你会想，算了，我们还是专注于核心业务吧，或者……我们还没做过……我是说，我们确实做过一些战略性押注。是啊。但我们没有尝试过那些自己还不太有把握的东西，主要是因为我们已经有……足够多的需求了。是啊。就像，我们想要做的这些东西已经够多了，所以……对我们来说，尝试说“嘿，也许这个东西能行”其实没什么意义，因为有那么多人都在强烈要求我们做其他东西，所以我们基本上就做那些。有时候我们会和客户意见不合，然后说：“其实，我觉得那边这个东西从战略上来说更有意义。”然后我们还是会试着把它做出来。比如从战略上来说对你们有意义？是啊，没错。我们觉得哪些方向有潜力？其中一个其实是智能体。就是说，实际上没有人会要求：“我希望我用的助手能有更多自主权，不用我开口就能替我做事。”从来没有人这么提过。但我们就是觉得，这肯定是行业的发展方向。它能让你做很多很酷的事情。协作也是一个方向，我们觉得行业会朝着能在你工作的地方直接协作的方向发展，也就是在 Legora 里面，而不是通过邮件和短信。所以，是的，这样的方向还有好几个。51:00 Yeah. I mean, primitives is right, you know? Yeah. Exactly. I think that's gonna be... That's really powerful. Do testing new products, just seeing if, like, hey, there's, like, any pull here, and then you just kinda, like... You're, like, no. Let's just focus on our main thing, or... We haven't made... I mean, we've made certain strategic bets Yeah. But we've not tried anything that we don't have a lot of confidence in already, mostly because we just have, like, amount... Enough demand. Yeah. It's like, we want all of these things, and so it doesn't... It hasn't really made sense for us to try to be like, hey, what... Maybe this thing could work, because, like, there's so many people screaming for the other things, and so we mostly build those, and then sometimes we'll disagree with clients and we'll be like, actually, I think this thing over here will make a lot of sense strategically, and then we'll try to build that anyway. Like, strategically for you? Yeah, exactly. What are the ones that we see? One of those things was the agent, actually. So, like, no one was actually asking for, like, I wanted to be... I want the assistant that I'm using to have more agency and do things without me asking it, like, no one no one ever, but we're just like, this is definitely where the industry is going. This is gonna empower you to do so much cool stuff. Collaboration was another one where we're like, we think the industry is going in a place where you can collaborate inside of where you're working, which is Legora and not in, like, emails and text messages. So, yeah, there there are there are few of those.

52:30 然后怎么说……比如，一旦你们说，好吧，没错，我们希望你们用智能体，我猜在律所内部，你知道，他们要真正学会怎么做，还需要花不少功夫。对。所以你们就会说，嘿，FDE 来这边教他们怎么做。对。法律工程师大多……哦，法律方面的。别……对，有道理。不过有些人会把他们称为驻场法律工程师，如果你愿意的话，但他们本质上就是做这个的。然后他们会过去，没错，帮助他们围绕智能体的能力来构建工作流程。你们之前提到过，我记得是 Max 说的，你们早期并没有微调自己的模型。现在还是这样吗？对。所以我们之前没有微调过任何模型，不过我觉得就在最近几个月，开放权重模型已经变得非常强大了。因此，对于某些任务，如果我们认为前沿模型在成本和延迟方面的表现，与我们认为经过后训练的模型能达到的表现之间存在巨大差距，我们就会进行微调。法律研究就是一个很好的例子。这就像是一个扇出检索问题，你可以通过这些查询，非常快速、非常便宜地构建出一个效果很好的模型，而且延迟很低。具体来说，你们是对它们进行后训练，让它们在哪些方面表现更好？非常具体的……对，有明确边界的。没错。好。对。我们不是想训练自己的 Fable……对，让它成为主智能体。我觉得那不是……我们大概赢不了 Anthropic，你知道的。还有很多，比如说，检索研究或者 RAG 研究……嗯……要做。对。52:30 And then how... Like, once you say, okay, yeah, we want you guys to use agents, I'm assuming, like, internally at the law firm, you know, it's gonna take a lot for them to actually learn how to do that. Yep. And so that's where you kinda, like, say, hey, FDEs go in there and teach them what they Yeah. Legal engineers mostly Oh, legal. Don't be. Yeah. That makes sense. But they're like, some people call them forward deployed legal engineers if you want, but that's essentially what they are. Then they go in, yeah, and help and help them structure their workflows around the capabilities of the agent. You guys said previously, I think Max said this, but that early on specifically, you guys weren't fine tuning your own models. Is that still the case? Yeah. So we haven't fine tuned any models up until I think now in the past few months, open weights models have become extremely capable, and so we will do it for certain tasks where we think there's a huge delta between what a frontier model can do at the cost and latency versus what we think a post trained model can do. Legal research is a great example. It's like you have to make... It's a it's a fan out retrieval problem, and you can make a really effective model through those queries super quickly and super cheaply at low latency. So specifically, you're post training them to be better at? Very specific Yeah. Bounded Exactly. Okay. Yep. We're not trying to, like, train our own Fable Yeah. To, like, be the main agent. I I don't think that's like a... We won't win that, you know, versus Anthropic, most likely. There's still a lot of, let's say, and retrieval research or RAG research Mhmm. To do Yeah.

54:00 也就是说，可能没那么光鲜。有时候甚至都不是在训练模型，就是做架构和流水线。你们和 Turbopuffer 做了个案例研究，不过，你知道，还有……你们会做 late interaction 吗？会。绝对会。你们会做那个。对。其实我们不做那个，不过……你说得完全对，这确实是……团队里有个人叫 Song，他特别在意搜索，基本上老是来找我，说：“哦，我们应该做这些事。现在有这些新的嵌入模型，我们可以做这个 late……还有 late interaction。”他还联系了 Turbopuffer 的团队，让他们开发 multivector，这样我们就能做这类事情了。不过，确实还有很多容易实现的改进，通常来说，甚至在模型方面之前、甚至是在 harness 里，说实话也是如此。要是你只是想给客户提供更好的结果，我们有很多事情可以做，而且在训练模型之前就能立刻见效。对。那我想问，怎样才算是适合法律行业的优秀 harness？我猜你们是在做自己的定制 harness。没错。对。我们处理文件的方式很重要。我们进行分析的方式很重要。你做法律研究的方式也都融入其中了。比如，遇到一个问题，或者发现了什么，它就知道该怎么做法律研究，弄清楚答案是什么、有哪些先例。这些都非常重要。当然，围绕它还有各种治理方面的东西。对。看起来搜索可能是你们定制得最多、最重要的工具？因为如果你找不到正确的信息……对。那你就……对。完蛋了。完蛋了。这……对，我觉得是……你们要搜索很多文档。还有分析。对，对。搜索、分析，可能还有起草。好吧。因为，像是起草文书……起草很长的合同，其实也没那么简单。54:00 That is, like, maybe less glamorous. It's not actually even model training sometimes, it's just like architecture and pipelines. You You did a case study with turbopuffer, but, like, you know, there's also... Like, do you do late interaction of Yes. Absolutely. You you do that. Yeah. We actually don't do that, but it's... You're completely right that this is... So there's one guy on the team, Song, who's... He just really cares about search, and he he basically comes to me all the time, and he's like, oh, we should do these things, and, like, there's these new embedding models, and, like, we can do this late... And late interaction. He actually talked to turbopuffer teams to get them to build multivector so that we could do these types of things, But, no, there's loads of low hanging fruit generally, even before models, and even in the harness, honestly, like, if you just wanna create a better output for our clients, there's, like, so much stuff that we can do that has immediate impact before training models. Yeah. Like, what makes a good harness for, like, the law industry, I guess? Like, you guys are making, I'm assuming, like, your own custom harness. Yep. Yeah. Of Yep. Way that we edit files is important. The way that we do an analysis is important. The way you do legal research is all baked into it. So it's like, you might have a problem that comes up or sees something, then it knows how to go and do the legal research to figure out what the answer is, what the precedent is. That's all very important. And then, of course, there's all the governance stuff around it as well. Yeah. It seems like the search is probably, like, would you say the most important tool that you guys are, like, customizing? Because It's if you can't pull in the right information Yeah. And then then you're Yeah. Screwed. You're screwed. It's... Yeah. I think it's it's it's You guys are searching, like, a lot of documents. And analysis. Yeah. Yeah. It's search, it's analysis, it's drafting, probably. Okay. Because, like, drafting word... Drafting huge contracts is actually also not super trivial.

55:30 比如说，哪些方面并非微不足道？因为你得……假设你要从头起草一份庞大的合同，那就得先找先例：以前我们是怎么处理这类事情的？需要这种 A 类条款，还是要找特定的律所？他们以前是怎么做的？对。所以我是这家律所，正准备起草一份很大的……总之，NDA 很简单，但有些合同要复杂得多，你得先找到合适的先例。因为律师不会坐下来就开始敲字。他们会先弄清楚，在类似的交易中我们用过哪些条款，而“类似”可以有很多不同的含义。我想你们肯定知道，在搜索领域，可能是同一行业、同类型的公司，或者其他任何相似之处。所以你得找出合适的人，可能还会有好几份文件，然后你得想清楚该从哪份文件里拿哪一部分，这其实并不简单。所以本质上这也是一个搜索问题。哪个模型最擅长起草？对，我想说……有人昨天告诉我，那个效果简直完美。Fable 和 Opus 肯定都特别能说。我相信你们也知道这一点，它们话很多。所以这不太好，一般来说确实不太好。是啊。我注意到 Astra 写得挺不错。Astra……对，写得挺不错。它会简洁很多吗？你们是不是跑了大概 100 个不同的提示词来……对。确保这一点。我们可以……对，我们可以做一些调整，但我觉得要控制它话多、冗长的问题，对我们来说确实有点难。哇。所以这……我觉得，你知道，我们拭目以待吧。不过 Anthropic 的下一个模型大概会更擅长这方面。Astra 实际上非常简洁。55:30 Like, what about it is not trivial? Because you have to... Let's say you're drafting a huge contract from scratch, then you have to find precedents of, how do we do this previously? Do we need this type of clause A or specific law firm? How did they do it previously? Yeah. So I am the law firm, and I'm about to draft a huge... Whatever. NDA is very simple, but, like, there's... There are contracts that are way more complex, and then you need to find the right precedent first, because, like, lawyers just don't don't, like, sit and, like, start typing stuff. They they figure out what clauses have we used in similar deals, and similar can mean a lot of different things. As I'm sure you know in search, it could be like same industry, same type of companies, same whatever. And so you need to identify the right person, and you might have like multiple documents that you then try to figure out which part should I take from what document, and that's actually nontrivial. So it's essentially also a search problem. Which model's best at drafting? Yeah. So I'll say... I was told that was perfect this yesterday. Definitely Fable and Opus are extremely chatty. I'm sure you guys have known this. They're very verbose. So that's not good. That's generally not good. No. I noticed Astra's been pretty good at writing. Astra's... Yeah. Pretty good at writing. Is it much more concise? You've running, like, a 100 different prompts to, like Yeah. Be sure. We we can... Yeah. We can we can do we can do certain things, but I think the chattiness verbosity is actually a little bit difficult for us to do. Wow. So this is... I think, you know, let's see. But probably the next model from Anthropic is gonna be much better at this. Astra is actually very concise.

57:00 是啊，是啊。我想他们都有一个和你们的 Slack 频道，你会在里面跟他们说这些，对吧？也就是说，这对他们来说不是什么新鲜事。 不，不是。他们知道。 然后另一件我特别在意的事是幻觉。嗯。他们会不会编造东西？ 我们有一套……引用系统，基本上会检查他们写的内容有没有链接。这样用户就能点链接，系统会在 PDF 或其他文件里用一个边框框出相关位置，标明这些信息是从哪里来的。这样我们就能确保它说的内容总能在某个地方找到出处，即使要往下追好几层也是如此。比如说，如果智能体根据一份表格审查写了某些内容，它就可以引用表格审查中的某个单元格或某一行，这样你就能点进去，一层层往下查证，一路验证到最初的来源。 对。你们发现律师会去核实这些引用是否正确吗？ 一开始……有点像是另一项工作，但现在…… 对，没错。情况就是这样。见鬼，比如说它幻觉出指向最高法院的链接。他们也不在乎。这种事会发生。 对，没错。这种事确实会发生。 我觉得这就像编程一样，你知道，有些工程师根本不在乎，就直接把任务交给智能体；还有些人会坐下来逐字逐句地读。情况完全一样。事情实在太多了，一个人怎么可能全都做完呢？对吧？ 是啊。但我觉得他们可能……就像我们在代码审查时知道该看什么一样。 对。 比如系统是什么之类的。我觉得他们也知道，好，这些是重点，这些是需要注意的……比如说，这部分很重要，我来重点核对一下这些内容；其他看起来也合理，所以应该没问题。 明白了。有意思。就像你收到一位初级律师的东西时，你也会这么做，对吧？就是想，我应该核对一下这个；这个大概没问题。 法律推理方面还有其他问题吗？或者说，法律领域里还有哪些前沿问题是现有模型还做得不够好，而你希望下一代模型能够解决的？ 我现在想不出什么特别具体的。我觉得长文档推理仍然很困难。很长的文档，比如有一百万 token 的合同，基本上就是……你有一份巨大的合同。 是啊。 你要在里面各处修改，这对现在的模型来说仍然很难。它还得记得你在前面改过哪些地方，定义也跟着变了，条款顺序也变了，等等。所以，最简单的做法就是一次性把所有内容都输入进去，然后一次性输出结果。57:00 Yeah. Yeah. And I I assume they they all have a Slack channel with you and you're telling them this. Right? Like, Yeah. This is not news to them. No. No. They know this. Then the other thing that I really care about is hallucination. Mhmm. Do they make stuff up? So we have a system of... Citation system where we basically check what they're writing has a link, so the user has a link, they can click and it highlights with a bounding box, like in the PDF or whatever, where where this information was taken from, and we can sort of make sure that what it says always is citable somewhere, even multiple layers down. So, like, if the agent writes something based on a tabular review, you know, it can cite the cell in the tabular review or the row in the tabular review, so you can actually click multiple layers down and find, you know, sort of validate it all the way up. Yeah. Are you finding that lawyers are, like, verifying that these citations are correct or not? In in in in initially, it's kinda like Some other work, but now... Yeah. Exactly. It's the same thing. Shit, like, hallucinated links to Supreme Court. They don't care. That happens. Yeah. Exactly. That happens. Yeah. I I think it's like coding, you know, you have engineers that just don't give a shit and just, like, send agents, and you have people that, like, sit and read every comma. Same description. There's just, like, so much that, like, how can one human do that? Right? Yeah. But I think they probably... It's just like we know in code review what to look for Yeah. Like, what is the system or whatever. I think they also know, okay, this is the the thing, these are the red... Like, this is the thing that's important, let me just, like, double check these things, and, like, these sound reasonable, so it's probably fine. Okay. Interesting. Just like if you if you receive something from a junior lawyer, it's the same thing you do, right? Just like, I should double check this, and, like, this probably is fine. Any other elements of, let's call it, legal reasoning, or what's the the the legal frontier that models are not good enough yet that you want from the next generation of models? Nothing I can think of specifically right now. I think long document reasoning is still difficult. Long, like, million token token contracts is basically You have a huge contract Yeah. And, like, you're making edits across it, it's a difficult thing to do for a model today even. And I remember that you changed all these things up here, and, like, the definitions changed, and that you... The ordering changed, whatever else. So, the simplest would be one shot, you feed everything in, and then one shot output.

58:30 是的。你不是那样做，而是把它拆成若干块，然后编排编辑。对。没错。完全正确。这就是这个行业里一些挺有意思、参与其中也很有趣的事情之一，对吧？我们的 Word 插件就是这样。我们做过 MapReduce，后来又改了，有过一个很大的版本，而且还不止一次。就像我们一直在改。是的，随着模型变化，我们搭建的支架也会跟着变。是的。我最近……我做很多新闻报道，嗯，显然是因为我的工作，而且最近我从 MapReduce 转向了在这些上下文之间来回切换。效果更好。是啊。不过这也……不一样。我的意思是，我们在搜索方面遇到过这个问题：比如你要在一份很长的合同里搜索某些内容，可以用 MapReduce，但这样可能会……如果你要总结什么内容，就可能会丢失一些重要的细微之处。没错，尤其是把各个部分单独总结时，块与块之间的内容就可能被忽略。所以我们一直面临这样的问题：我们可以做两种搜索，也可以做另一种，就是需要在不同步骤之间进行推理的那种。我在想一个特别好的例子。有些文档会用很大篇幅谈论某件事，比如说两页左右，但每个单独的文本块未必和搜索词或他们想弄清楚的问题相关，合在一起却极其相关。所以，是啊，这些都是我们遇到的一些问题……而且这一定很难，58:30 Yes. You don't do that, you break it up into chunks, and then you orchestrate edits. Yeah. Exactly. Exactly. And this is one of those things that are funny, that are fun to be part of in this industry. Right? So our Word add-in, it was like, we've done, you know, we've done MapReduce, we've changed, and there was one big one, and there was one big one in multiple times. It's like, we keep changing Yes. The the scaffolding that we build around as models change. Yes. I I I would say I I I've recently... I do a lot of news reporting Mhmm. Obviously because of what I do, and I've recently shifted from MapReduce to throwing between these context. It's better. Yeah. But it's also... It's different. I mean, we had this problem with search where it's like, you're searching for certain things in a huge contract, and you can map reduce it, but then it might... If you're summarizing any sort of thing, then it might lose some important nuance Yep. In between the chunks, essentially, if you just summarize the individual parts up. So this has been a problem where it's like, we can do two types of search, we can also do the other one where it's like, you need to have reasoning in the different steps there in between. So I'm trying to think of a really good example here. There's some documents where they talk about something at length over, like, let's say, two pages or whatever, and then each individual chunk is not necessarily relevant to the search term or what they're trying to figure out, but together it's extremely relevant. And so, yeah, these are some of the, you know, problems where it's... And it must be hard,

1:00:00 就像你刚才说的，要真正让模型在这方面变得更好，因为你甚至都不能看那些合同。对吧？没错。你们会不会向他们提出，比如说：“嘿，能不能……”会。做一些跟真实合同类似的示例合同？会。所以我们会收集反馈，可以从他们那里收到反馈；如果他们真的很投入，没错。或者他们会直接说：“嘿，这对我们来说特别重要，但现在效果不太好。”然后我们就会说：“好啊，太好了。你们能不能把这些文件和问题之类的内容匿名化一下？然后我们……”我们……如果能重现问题，就能修好。是啊。我的意思是，这件事有一种做法，就是收购一家律所。对。有些人正在考虑这么做。是的。不过这里面有些……我们……有些问题在于，他们的数据到底算他们自己的，还是客户的数据。所以，如果他们的客户都已经不在了，那也许就没问题。我其实也不太清楚。有意思。是啊。不过……没错。这是获取大量数据的一种方式。是啊。像 Google 就在收购 Spiritualize。没错。是啊。收购了……是啊。他们收购了……是啊。是啊。你知道，如果一家公司倒闭了，你还是有 Slack 的。你可以把 Slack 卖掉。我记得有家 YC 公司专门收购倒闭的公司。对，是有这么一家。是啊，是啊。因为他们想要的就是：“哦，我只想要真实的人在做事。”没错。然后就……对，就把东西拿出来。没错。收购律所数据的问题是……嗯，你能拿到所有输入，也就是实际的输入，这很棒。但输出就比较难处理，因为你拿到的只是他们最终的结果，很难判断他们是怎么做到的……基本上你得反向推导他们是怎么得出那个结果的。所以在我们的产品里，如果能拿到真实用户的真实操作轨迹，价值非常高；但如果我们只拿到输出，却不知道它好不好、也不知道它是怎么生成的，那就有点难办了。1:00:00 like you said earlier, to actually make the models better at that, because you can't even look at the contracts. Right? Exactly. Do you guys ever ask them for like, hey, can you Yes. Make example contracts that are similar to the real ones? Yeah. So we have we have feedback, we can receive feedback from them, and if they're really engaged, yeah, exactly. Or they're just like, hey, this is, like, super important to us and it's not working well, and then we're like, okay, great. Can you, like, anonymize the documents and, like, the problem, whatever, and we'll... We... If we can recreate it, then we can fix it. Yeah. I mean, so one version of this is you buy a law firm Yep. Which some people are considering. Yes. But there's some... We... There's some stuff there on whether their data is actually their data or if it's the client's data. So if their clients are all gone, then maybe it's fine. I actually don't know. Interesting. Yeah. But it's... Yes. It's it's a way to get a lot of data. Yeah. People like Google is buying Spiritualize. Exactly. Yeah. Bought Yeah. They bought the... Yeah. Yeah. Like, you know, if you have a failed company, you still have your Slack. You can sell your Slack. I think there's like there's like a YC company that buys your failed company. Yes, there is. Yeah. Yeah. Because, like, it's just like, oh, I just want real people doing work. Exactly. And then, yeah, just so it was like Exactly. Pull stuff out. Exactly. The problem with buying law firm data is do... Well, you get all the inputs, the real inputs, which is great. The outputs are a little bit more difficult, because you just get what their end result was, and it's difficult to figure out if, like, the way... You have to basically, like, reverse engineer how they got there. So in our product, when we get real traces from really users, it's extremely high value, But if we just got this output, and we don't know if it's good or bad or how it was created, it's a little bit difficult.

1:01:30 我们快结束了。最后我还想提醒大家，embedding 方面仍有很多 alpha 可挖。如果你的工作负载大量依赖 RAG，只看 MTEB、直接选排名靠前的模型，那就错过了很多机会；你应该看看 Voyage，或者一些定制的 embedding。你已经深入研究过这个了。对。所以基本上，不要直接用默认的 embedding，哪怕是 OpenAI 的 embedding、Gemini 的 embedding，都一样。要定制自己的 embedding。对吧？是的。你们会支持这一点。是的。我觉得没错。我觉得没错。唯一要确保的是支持多语言。我们服务的语言很多，所以这方面世界是王者。对。没错。Mistral 在这方面有帮助吗？没有。直到……嗯，直到现在，他们刚融了一大笔钱，对吧？所以也许……是啊，拭目以待。你们最好……他们也融了很多钱。你们会成为自己的拥护者。我觉得我们需要更多来自……来自欧洲的模型公司。是的。我想问一下，你对法律行业未来一年左右有什么预测吗？我们之前稍微聊过这个。好，我把问题改成你刚才的第二个问题：两年后，律师会在 Legora 里完成工作，还是会使用 Claude 或 TrejoC 之类的其他工具，然后通过 API 调用 Legora？1:01:30 We're gonna wrap up soon. The last thing I'll point people to is also there's a lot of alpha left on embeddings. You're if you're a RAG heavy work... Workload, you're just looking at MTEB and you're taking, like, top model, you're leaving a lot on the table, you should look at Voyage or some of the custom embeddings. You've gone into this. Yeah. So basically, don't take default embeddings, even OpenAI embeddings, Gemini embeddings, doesn't matter. Customize your embeddings. Yeah. Right? Yes. You you you will support that. Yes. I think that's right. I think that's right. The only thing is you need to make sure that you're multi language. So for us, we we serve so many languages, so that's the world is the king of this. Yeah. Yeah. Exactly. Mistral, helpful there? No. Until... Well, until now, they just raised a ton of money. Right? So maybe Yeah. Let's see. You guys better They also raised a ton of money. You're gonna be your champions. Like I think we need way more model companies out of the... Out of Europe. Yes. I just wanted to ask you, like, what predictions do you have maybe, like, over the next year for, like, the law industry? We we asked a little bit of that earlier. Yeah. I'll modify it to your second question, which is in two years, will lawyers do their work inside Legora, or will they use other tools like Claude or TrejoC and call Legora, like, you know, the API?

1:03:00 是啊。是啊。是啊。是啊。我觉得两种情况都有。真的要看情况。我觉得如果你是一家大型企业，可能会用 Claude Cowork，然后你会……这……我觉得这跟我通过 Claude 使用 Linear 的方式差不多，但我还是会去用 Linear。我觉得企业会这么做。我觉得律所就会直接用 Legora。是啊。因为那样更简单、更好。是啊。这让我有点想到 OpenEvidence，比如在医学领域的应用。你们基本上就是法律领域的 OpenEvidence。没错。是啊。完全正确。在编程领域，情况发展到一个地步，人们……那些实验室会因为你跟它们竞争而切断你的服务，这种事已经发生过两次了。没错。你觉得法律领域也会发生这种事吗？不会。我觉得不会。因为你知道，他们也在推出法律服务。是啊。是啊。是啊。他们确实在做。我不……我的意思是，我们当然经常跟他们交流，但这也……就像，嘿。你们打算……你们的销售代表会说不会。没错。他们说了不算。没错。那不是他们要做的决定，对吧？而且他们往往根本不知道实际发生了什么。我不觉得……我不觉得情况是那样。他们在知识工作领域，或者说在企业的知识工作领域，有那么多东西要覆盖，所以法律也必须是其中一部分，就像数据分析、销售、市场营销和财务也都必须涵盖一样。所以他们必须把这些领域的产品做好，但他们绝不会像我们这样深入。我觉得，如果 Claude 接触一些小客户、小型企业客户，让他们通过 Claude 处理部分法律事务，那很棒，完全没问题。但如果你要做一笔价值 4 亿美元的并购交易，你肯定会想用 Legora。我还有个不太相关的问题，不过我很好奇，你们会去法学院，想办法让他们在课堂上教 Legora 吗？这事是怎么运作的？1:03:00 Yeah. Yeah. Yeah. Yeah. I think both. It really depends. I think if you're an an big enterprise, you might have Claude Cowork, and then you'll... It's... I think it's similar to, like, how I use Linear through Claude, but I still go to Linear. Think I enterprise will do that. I think law firms will just use Legora Yeah. Because that's simpler and better. Yeah. It kinda reminds me of, like, OpenEvidence for, like, medicine. Like, you're you're basically that for law. Yep. Yeah. Exactly. In coding, it's going to a point where people... The labs will cut you off for competing with them, and this has happened twice. Yep. Do you do you think that will happen to law? No. I don't think so. I think Because, like, you know, they are releasing legal services. Yeah. Yeah. Yeah. They are. I don't... I mean, we obviously talk to them a lot, but it's also... It's like, hey. Are you gonna Your sales reps will say no. Exactly. They don't make the call. Exactly. That's not their goal to make. Right? And they don't even know, often what's actually going on. I don't think I don't think that's the case. They have so much stuff to capture in knowledge work or in in sort of enterprises knowledge work generally, that, like, legal has to be part of that, just like data analytics has to, and sales, and marketing, and finance. So, like, they have to do things that are good there, but they're never gonna go as deep as we are. And I think if if Claude takes small clients, small, you know, corporate clients and they do some of their legal via Claude, like, that's that's great, that's fine. But if you're doing a, you know, $400,000,000 M and A transaction, you're gonna want Legora. I had kind of an unrelated question, but I'm curious, are you guys going to law schools and kinda like getting them to like teach Legora like in classes? Like, how does that work?

1:04:30 是啊。这绝对是我们正在……我想我们已经开始做了。其实我不太清楚目前进展到哪一步，但我们确实希望法学院的学生能学会使用 Legora。是啊。你觉得法学院正在大幅调整课程设置吗？还是其实没有？我不知道。说实话，我还真不清楚。有意思。其实我会换一种方式来做。你们应该为计算机科学专业的学生、工程师提供法学院奖学金。我们会承担你的……你学过计算机科学？是的。我们会送你去读法学院？计算机科学背景加上一些软件行业经验。是的。我们会送你去读法学院。是啊。费用我们来承担。或者你们可以办硕士加 JD 的项目？是啊。他们不在乎硕士学位。嗯，我的意思是，因为很多人本科没学计算机科学，所以他们会读个硕士，补补课。是啊，是啊。我不知道。我倒没那么常见到这种情况。是啊，是啊。有人读 JD 加 MBA。这倒是有。是啊。但 JD 加计算机科学，我还没见过。确实有。是啊。是啊。好吧。总之，进展真的非常令人印象深刻。恭喜你们取得这么大的成功。你们还会继续保持这个发展势头吗？当然。是啊。你们现在在旧金山和纽约招聘，对吧？我们在……是啊，工程岗位在纽约、斯德哥尔摩和伦敦招聘。其他所有岗位，我们在各地都招。你们现在纽约有多少人？大概一百人。差不多一百。好。那你们计划在纽约招多少人？纽约？是啊。就大概来说？就大概来说，是的。1:04:30 Yeah. It's it's definitely a program that we are... I think we've started. I actually don't know exactly where we're at, but we we do want people in law school to Yeah. Learn to use Legora. Like, are you seeing that, like, law schools are changing their curriculums a lot or not really? I don't know. I I I wouldn't know, actually. Interesting. I would actually do it a different way. You should do scholarships for for law school for CS majors, for engineers. We will cover your done CS? Yes. We'll we'll put you in law school? CS plus some software experience Yeah. We'll put you in law school. Yeah. We'll cover it. Or you do, like, the MA... Like, master's plus JD programs? Yeah. They don't care about the master's. Yeah. Well, meaning meaning, like, because there's a lot of people who didn't do CS undergrad, so then they do, like, master's just to kinda, like, get get caught up. Yeah. Yeah. I I don't know. I don't see that as as often. Yeah. Yeah. People do JD MBA. That's a thing. Yeah. But JDCS, I haven't seen it. Yeah. Happens. Yeah. Yeah. Yeah. Okay. Look. Overall, you know, super impressive progress. I mean, congrats on all your success. Like, are you still gonna keep up the pace? Like Absolutely. Yeah. You know? And I think you're hiring in SF New York. We're hiring in... Yeah. Engineering, we're hiring in New York, Stockholm, and London. Yeah. And all other roles, we're hiring everywhere. How many people do you have in New York right now? A 100. About a 100. Okay. And you guys are planning on how many in New York? In New York? Yeah. For just, like, generally? Let's just say, yeah, generally and generally.

1:06:00 很多。能有多少就有多少。是啊。这挺激进的。嗯，那个……那儿是公司法的中心，对吧？就……确实是。嗯。美国中部的历史。所以，嗯。我的意思是，我很期待你买下一家律所，然后，呃，你知道，成为……是啊。是啊。这真的特别令人兴奋……我觉得，你知道，每个行业，尤其是顶尖领域，比如医生、律师、程序员，你知道，都得有一两家智能体实验室，彻底主导这个领域。我同意。所以你们的市场很不错。是啊。很好。我也这么觉得。而且随着法律不断变化、变得……让更多人都能接触到，我觉得这个市场只会变得更大。是啊。另外，我觉得……我的意思是，我们已经聊得太久了，不过你们解决的问题真的很……属于前沿的知识工作。你懂吧？没错。它们……好吧，除非你真的是在研究纳维–斯托克斯方程，比如科学领域的纳维–斯托克斯方程。除此之外，大多数工作基本上都会归结到法律上。是啊，到头来全都是法律问题。是啊。招聘、金融，什么都一样，我只要咨询我的律师就行。嗯。我知道。我知道。所以我觉得这是个特别令人兴奋的研究方向，因为现代社会在某种程度上是建立在法律之上的。是啊。是啊。好。行了。谢谢。好。谢谢。多谢。1:06:00 A lot. As many as you can. Yeah. It's it's it's aggressive. Yeah. Mean, that... That's the home of corporate law, right, as far as, like It is. Yeah. Mid America's history. So, like, yeah. I mean, I'm looking forward to you buying a law firm and, like, you know, become... Yeah. Yeah. It's it's super exciting to to... I think, like, you know, every industry, especially, like, the top, like, doctors, lawyers, coders, you know, they all have to have, like, you know, two or agent labs that, like, just completely dominate the space. I agree. So you guys have, like, a great market. Yeah. Good. I think so. And I think it's only gonna get much bigger as, like, law changes and get... Becomes more available to more people. Yeah. Also, I think, like... I mean, we're already, like, super over, but, like, the problems that you work on are very... Like, the frontier knowledge work. You know? Yep. They're... Okay. Unless you're literally, like, doing, like Navier Stokes. Science Navier Stokes and science. Other than that, like, most work is, like, basically converges to legal. Yeah. It's all legal in the end. Yeah. Hiring, finance, everything, I just have to consult my lawyer. Yeah. Know. I know. Which is why I think it's a really exciting thing to work on, because modern society runs in some way on legal. Yeah. Yeah. Cool. Alright. That's it. Thank you. Cool. Thank you. Thanks.

感谢阅读Thanks for reading