Eve
Eve:改造原告侧律所运营的垂直 AI 独角兽
Eve 在原告律师工作流里已经跑出真实牵引力,也有顶级投资人背书;但公开收入和留存仍不透明,独角兽估值很难激进承销。
封面要素
公司概况
Eve 是一家总部位于 San Francisco 的法律 AI 公司,聚焦原告侧律所,而不是 Big Law 或通用法律研究。公开产品和客户证据显示,其平台覆盖获客接待、医疗记录综合、索赔函起草、证据开示支持、研究、沟通,以及一个较新的案件数据层,意图横跨律所工作流。公司的市场叙事很强:Eve 2025 年 1 月完成 $47 million Series A,2025 年 9 月以超过 $1 billion 的估值完成 $103 million Series B,Andreessen Horowitz、Lightspeed、Menlo 和 Spark 都公开背书。客户证据也扎实,既有具名律所引用,也反复声称节省大量时间、提升转化率。核心尽调问题不是 Eve 有没有势头;而是当前付费 ARR、留存和法律风险控制是否足以支撑此前估值,并在现有厂商和相邻法律 AI 供应商追赶时守住长期领导地位。
- 成立时间
- 2020-01-01
- 创始人
- Jay Madheswaran, Matt Noe, David Zeng
- 创立地点
- San Francisco, California, USA
- 总部
- San Francisco, California, USA
- 产品
- 面向原告侧律所的 AI 原生工作流平台,覆盖获客接待、医疗时间线、起草、证据开示、案件沟通和律所情报。
- 客户
- 原告侧人身伤害、群体侵权、劳动就业、工伤赔偿、残障权益及相邻按胜诉收费的律所。
- 商业模式
- 销售驱动的软件订阅,配套工作流部署、上手培训和转型支持;公开价格仍未披露。
- 阶段
- Series B
- 融资情况
- 2025 年 9 月以 $1.1B 投后估值完成 $103M Series B,此前 2025 年 1 月完成 $47M Series A;2026 年材料显示累计融资超过 $160M。
执行摘要
主要优势
- 原告律师原生工作流覆盖面广,从线索接收到诉讼,再到正在形成的律所情报层
- 客户证明强,既有具名律所,也有可量化生产率提升和大规模案件量主张
- 蓝筹投资人组合和快速融资节奏,支撑其在法律 AI 品类里保持领导位置
主要风险
- 没有公开 ARR、留存、利润率或 ACV 披露,独角兽估值缺少干净支撑
- 法律 AI 的幻觉、特权信息、隐私和未经授权执业风险,都要求持续保持人工复核纪律
- 在位案件管理套件和相邻法律 AI 厂商,可能收窄 Eve 当前可见的产品优势
未决问题
- 当前付费 ARR、NRR、毛利率和部署服务强度未披露
- 450+、1,000+、1,200+ 和 1,400+ 等公开口径之间,客户数定义和付费账户构成尚未对齐
- 2025 年融资中的董事会控制、优先权堆叠和任何老股成分都未公开
目录
01公司概览
1.1 身份、产品范围与公开运营足迹
Eve 将自己定位为专为原告律所打造的法律 AI 平台,而不是通用法律研究助手。无论是官网首页、发布材料还是独立报道,公司都把软件描述成覆盖获客接待、评估、诉前、诉讼支持,并越来越延伸到全所运营。到 2026 年 6 月,公开产品话术已从案件工作助手转向「EveOS」——一个面向原告法律业务的 AI 操作系统,承诺提供 24/7 获客接待、AI 生成式起草、分析能力和律所实时数据层。总部披露比产品披露薄得多。独立融资报道一致称公司位于 San Francisco,但官网没有公布街道地址。法律实体层面的身份也不如品牌叙事清晰:面向公众的品牌是 Eve,而隐私材料反复提到 Butler Labs,SiliconANGLE 则称公司为 Butler Labs Inc.。因此,Eve 的商业身份清楚,但准确法律命名惯例比其营销表层更不透明。[CO001, CO002, CO007, CO014, CO015, CO016]
| 指标 | 数值 / 状态 | 日期 | 置信度 | 缺口 / 注意事项 |
|---|---|---|---|---|
| 运营品牌 | 面向原告律所的 Eve / EveOS | 2026 | 高 | 官方和第三方材料中的品牌清晰 |
| 最有支撑的起点 | 创始团队到 2020 年已活跃;随后商业化发布 | 2020-2025 | 中 | 公开材料支持 2020 年起源,但未发布清晰的注册时间线 |
| 总部 | 总部位于 San Francisco | 2025-2026 | 高 | 官方网站未发布街道地址 |
| 法律实体命名 | 品牌为 Eve;隐私材料提到 Butler Labs;SiliconANGLE 称 Butler Labs Inc. | 2025-2026 | 高 | 公开文件未清晰确认 “Eve Legal, Inc.” |
| Series A | $47M,由 Andreessen Horowitz 领投,Lightspeed 和 Menlo 参与 | 2025-01 | 高 | 官方和投资人来源支持强 |
| Series B | $103M,由 Spark 领投,估值 $1B+ | 2025-09 | 高 | 未公开所有权比例或董事会条款 |
| 已披露最低累计融资额 | $150M | 2025-09 | 高 | 仅简单加总已披露的 Series A 和 Series B |
| 2025 年客户数 | 450+ 家律所 | 2025-09 | 高 | 代表 Series B 后披露,而非当前运行日客户数 |
| 2026 年规模主张 | 500+ 至 1,400+ 家律所;每年 / 活跃事项 200,000+ 个案件 | 2026 | 中 | 当前精确客户数随来源和日期变化 |
| 安全姿态 | SOC II Type 2;HIPAA;零留存模型训练主张 | 2026 | 高 | 主张由公司发布,审阅材料中未见监管审计 |
| 员工人数 | 2026 | 低 | 审阅的公开来源未披露员工人数 |
当前客户数和法律实体字段是最重要的公开不确定项;null 表示审阅来源未公开披露该指标。
[CO001, CO007, CO008, CO009, CO010, CO014]Eve 的身份把创始人市场匹配、原告律所工作流覆盖、融资、客户和信任风险串成一个运营论点。
[CO001, CO003, CO004, CO009, CO016, CO025]1.2 创始人、领导班底与治理可见度
创始团队是 Eve 公开资料里最强的一块。公司官网列出 Jay Madheswaran、Matt Noe 和 David Zeng 为核心创始团队,投资人和访谈报道也拼出一条连贯的创始人—市场匹配故事:Jay 曾在 Facebook、Rubrik 和 Lightspeed Venture Partners 工作,Matt 和 David 则被描述为技术功底深的 Rubrik 校友。这一点重要,因为 Eve 把工作流 AI 卖进一个高风险、文件密集的法律垂直领域,产品质量和信任是战略差异点,不是锦上添花。公开来源也支持 2020 年的起点,尽管市场发布显然更晚。治理则远不那么可见。已审阅来源没有披露董事会构成、投票控制、保护性条款,也没有精确披露创始人之外的内部管理层。因此,创始人匹配和产品可信度看起来很强,但机构治理和股权控制仍是尽调问题,而非已验证事实。[CO003, CO004, CO005, CO006, CO007, CO040]
| 人物 | 职务 | 背景 | 创始人市场匹配 / 职能覆盖 | 关键人物依赖 |
|---|---|---|---|---|
| Jay Madheswaran | 联合创始人兼 CEO | 前 Facebook 工程师、Rubrik 运营人员和 Lightspeed 风险投资人 | 结合 AI 深度、企业软件经验和原告法律转型的市场叙事能力 | 高 |
| Matt Noe | 联合创始人兼 CPO | 前 Rubrik 创始工程师和产品负责人 | 负责产品架构、工作流设计,以及在原告律所中的实际部署 | 高 |
| David Zeng | 联合创始人兼工程负责人 | 聚焦 AI/ML 的工程负责人;Lightspeed 称其为早期 Rubrik 工程师 | 负责核心技术执行和平台可靠性 | 高 |
各行只覆盖公开识别的创始团队;更广泛的高管梯队和董事会在审阅来源中披露不清。
[CO003, CO004, CO005, CO006, CO007]1.3 融资历史、投资人组合与当前阶段
Eve 的融资形成过程有官方和第三方证据充分支撑。公司 2025 年 1 月宣布由 Andreessen Horowitz 领投的 $47 million Series A,Lightspeed 和 Menlo Ventures 参投;八个月后又完成由 Spark Capital 领投、Andreessen Horowitz、Lightspeed 和 Menlo 继续参投的 $103 million Series B,估值超过 $1 billion。仅这两轮就意味着 2025 年 1 月以来披露融资至少 $150 million,也把公司稳稳放进后期私营公司地带,而不是早期实验性法律科技。投资人评论也解释了这笔资金要做什么:Spark、Lightspeed 和 Eve 管理层都把公司描述为在打造「AI-native law」新类别,资金用于更深产品开发、转型服务、上手培训和更广案件生命周期覆盖。公开资料仍未回答的是,Series B 后治理如何变化、是否包含老股交易,以及投资人组合和创始人之间的持股如何分布。[CO008, CO009, CO010, CO042]
| 利益相关方 | 角色 | 控制 / 经济重要性 | 尽调问题 |
|---|---|---|---|
| Spark Capital | Series B 领投方 | $103M 轮次的领投人,可能是关键董事会 / 治理影响点 | 确认董事席位、持股比例和任何保护性条款 |
| Andreessen Horowitz | Series A 领投方及 Series B 参与方 | 公开文件中最早被点名的机构领投方;可能影响后续融资策略 | 确认 Series B 后当前持股和信息权 |
| Lightspeed Venture Partners | 种子轮 / Series A / Series B 支持方 | 多轮持续支持,并与 Jay Madheswaran 关系密切 | 厘清 Lightspeed 是否保留董事影响力或观察员权利 |
| Menlo Ventures | Series A / Series B 参与方 | 两轮公开融资中均被点名的重复投资人 | 核验持股水平和后续跟投承诺 |
| 旗舰原告律所客户 | 商业验证群体 | 具名律所提供可信度、工作流反馈和品类契合的公开证明 | 衡量头部 logo 的集中风险和续约依赖 |
| 创始团队 | 运营控制核心 | 公开叙事和产品可信度高度集中在创始人身上 | 要求提供资本表、投票控制和继任细节 |
公开文件点名了投资财团和头部客户,但未披露完整资本表结构、董事会组成或二级交易活动。
[CO008, CO009, CO010, CO011, CO029, CO030]1.4 客户证据、里程碑速度与负面信号
牵引力是 Eve 超越融资新闻而值得关注的最清晰理由。已审阅材料显示,2025 和 2026 年的规模口径迅速上升:Series B 前后超过 450 家律所,2026 年 1 月超过 500 家,3 月超过 800 家,当前官网写 1,200-plus,2026 年 6 月 LawNext 报道写 1,400-plus。这些数字方向上亮眼,但彼此并不完全对齐,因此准确当前数量是尽调项,而非既定事实。即便如此,客户证据异常具体。公开来源点名 Mike Morse Law Firm、James Scott Farrin、Barrett & Farahany、Disparti Law Group、Frontier Law Center、Laurel Employment Law 和 Hershey Law;案例研究则把每周节省 20 多小时律师时间、产能翻倍或两倍增长、部分工作流平均周期四个月、以及 $27.5 million 判决额归因于使用 Eve 的律所。融资后产品里程碑也在延续,包括 2026 年 1 月 AI Workforce 发布和 2026 年 6 月 EveOS 扩展。负面一侧同样重要:LLRX 和一份 2026 年 3 月法院文件把 Eve 关于幻觉防护的营销与培训说法,同一名律师的不准确引用事件联系起来,尽管该律师表示无法判断错误来自 Eve 本身,还是来自他自己的复制粘贴工作流。这不会推翻业务,但说明信任、验证和法律流程控制是实质性尽调风险。[CO011, CO012, CO013, CO018, CO019, CO021]
| 日期 | 事件 | 类型 | 金额 / 估值 / 状态 | 参与方 | 含义 |
|---|---|---|---|---|---|
| 2020 | 创始团队开始搭建 Eve | 创立 | 起点 | 创始团队:Jay Madheswaran、Matt Noe、David Zeng | 公司创立叙事中最有支撑的起始日期 |
| 2025-01 | Series A 公布 | 融资 | $47M | Andreessen Horowitz;Lightspeed;Menlo | 推动 Eve 从早期动能进入规模化商业化和产品扩张 |
| 2025-03 | LawNext 播客阐述「AI-native law firm」论点 | 产品 | 使命阐述 | Jay Madheswaran;LawNext | 展示公司在独角兽轮次前的品类叙事 |
| 2025-09 | Series B 公布 | 融资 | $103M 融资,估值 $1B+ | 投资方:Spark Capital、Andreessen Horowitz、Lightspeed、Menlo | 确认独角兽级估值和更强投资人阵容 |
| 2025-09 | 公开规模披露达到 450+ 家律所和每年 200,000+ 个案件 | 规模 | 客户和使用里程碑 | Eve;PRNewswire;LawNext | 显示私有法律科技中异常快速的品类采用 |
| 2026-01 | Eve 2.0 / AI Workforce 发布 | 产品 | Agents + Auditor + Analyst | Eve;LawNext | 把产品从起草辅助延伸到自主工作流执行 |
| 2026-03-27 | Rushing v. Turner 道歉信提到 Eve 辅助起草卷入幻觉相关事件 | 不利 | 准确性事件 | Ross LeBlanc、Dudley DeBosier 与 Eve | 提出关于核验控制和培训主张的尽调问题 |
| 2026-05-31 | LLRX 记录包括 Eve 在内的法律科技供应商不断演化的幻觉防护营销主张 | 不利 | 独立批评 | Damien Charlotin;LLRX | 显示信任和准确性主张仍是声誉风险类别 |
| 2026-06-11 | EveOS 发布把产品扩展到 Atlas、通信代理、分析和研究 | 产品 | 平台扩张 | Eve;LawNext | 推动 Eve 走向操作系统定位,而非点状工作流工具 |
该时间线覆盖截至运行日,在创立、融资、产品演进、规模和不利信号方面最有支撑的公开里程碑;不包括每一次产品发布或商业事件。
[CO007, CO008, CO009, CO011, CO012, CO024]从创立到融资、AI Workforce、2026 年不利准确性事件,再到 EveOS 发布的关键公开里程碑。
[CO007, CO008, CO009, CO011, CO012, CO026]公开融资、使用和客户指标显示增长很快,但截至运行日期的精确规模仍未解决。
客户数量这一行有意写成区间,因为 2026 年公开数字在不同来源之间并不一致;图表保留方向判断,但不夸大精度。
[CO008, CO009, CO010, CO011, CO012, CO013]1.5 图表
02市场分析
2.1 市场边界、相邻领域与现状替代品
Eve 服务的市场比「法律 AI」更窄,但又比单一人身伤害点工具更宽。Eve 官方页面称产品为原告律所打造,并用于获客接待、医疗概览、索赔函起草、证据开示答复和更广运营。Eve 还明确列出原告侧执业领域,包括人身伤害、工伤赔偿、医疗过失、劳动就业、群体侵权和 Social Security disability。这一点重要,因为原告律所的现状不只是 Westlaw 式研究或通用聊天机器人。Smokeball、CloudLex、CasePeer 和 Paxton 的竞品与工作流资料都把原告工作描述为从获客到和解的操作系统:客户筛选、医疗记录收集、时间线构建、起草、和解追踪、留置权处理和款项分配。换句话说,纳入的支出是按胜诉收费原告业务的案件生命周期工作流人力和软件;排除的支出是大多数仅面向被告的电子取证、通用 BigLaw 知识工具,或从不触达原告运营的宽泛法律研究预算。核心替代集合是人力、传统案件管理软件、外包记录与留置权供应商,以及仍需大量原告专属提示和审查的通用 AI。[CM001, CM002, CM003, CM004, CM006, CM026]
| 细分市场 / 类别 | 纳入支出 / 工作流 | 排除支出 | 买方 / 付款方 | 相关性 |
|---|---|---|---|---|
| 人身伤害原告律所 | 线索受理、医疗记录、时间线、索赔函起草、证据开示、和解跟踪、款项分配 | 辩方研究、仅限辩方的文档审查、未绑定原告运营的通用知识工具 | 由领导层控制的律所运营预算;受理人员、律师助理、律师使用 | 主要服务市场 |
| 群体侵权原告平台 | 索赔人筛选、记录审查、事实表、MDL 协调、和解管理 | 原告运营之外的辩方共同利益或产品辩护工作流 | 诉讼领导层和运营预算 | 核心原告相邻市场,量级会脉冲式爆发 |
| 集体诉讼 / 雇佣法原告律所 | 索赔人受理、沟通、诉状、证据开示、和解通知 / 管理工作流 | 辩方合规调查和按小时计费的辩方研究 | 业务负责人加运营预算 | 可扩展的原告细分市场,工作流逻辑相似 |
| 相邻原告专业领域 | 工伤赔偿、医疗过失、SSDI 及类似风险代理收费或索赔人工作流 | 索赔人密集型运营之外的一般法律业务 | 部门负责人或管理合伙人预算 | 支撑同一工作流引擎向汽车 PI 之外扩展 |
| 现状替代技术栈 | 人工受理团队、律师助理、案件管理套件、记录 / 留置权供应商、通用 AI 工具 | 无运营嵌入的纯研究支出 | 同一律所运营预算,已经在支付人员和点状工具 | 定义真实替代基线 |
纳入和排除支出围绕原告侧工作流执行划定,而不是整个法律软件宇宙;买方 / 付款方标签概括有证据支持的运营模式,而非普遍适用的职位名称。
[CM001, CM002, CM003, CM004, CM006, CM026]证据从美国法律职业整体收窄到人身伤害原告律师核心人群,再落到面向原告的工作流层;这些层的 SAM 和 SOM 仍只有部分公开。
前四层都有来源支撑,但有意混合了劳动力、案件量和复杂度三种视角,因为公开资料没有给出干净的原告工作流 AI SAM。最后一层受证据约束,而不是数字化估算。
[CM013, CM014, CM015, CM016, CM038, CM039]2.2 规模测算视角、分层结构与有边界的 TAM 逻辑
公开证据支持一个有边界的多视角市场判断,而不是单张膨胀 TAM 幻灯片。最宽口径上,ABA 称美国 2025 年有 1.37 million 名律师。Clio 2026 年人身伤害市场综合资料则把这个宽分母缩窄到 135,000-plus 名人身伤害律师,约占执业律师 10%,并有近 400,000 起年度 PI 索赔主要在州法院提交。同一来源引用 2025 年美国人身伤害律师及律所 $61.7 billion 行业收入,经济池很大,尽管这不是软件支出。常规 PI 之外的复杂度也有意义。Lex Machina 称,房屋责任和机动车侵权案件在其 2025 年联邦侵权报告中达到创纪录水平。PFAS AFFF 多区诉讼页面称,单一 MDL 就有 10,000-plus 起关联案件和数万名原告;Duane Morris 则审阅了上一年超过 1,441 份集体诉讼裁决。合在一起,这些视角说明原告律师群体规模可观、碎片化且运营密集,足以支撑垂直工作流软件。公开数据无法清楚揭示的是,各细分市场原告律所的准确数量,或该法律收入池中已经划给 AI 和工作流软件的美元份额。[CM013, CM014, CM015, CM016, CM017, CM038]
| 发布方 | 年份 | 地理范围 | 数值 | 方法 / 视角 | 置信度 | 局限 |
|---|---|---|---|---|---|---|
| American Bar Association | 2025 | 美国 | 1.37M 名律师 | 用全国律师人数锚定广义法律市场分母 | 高 | 并非专指原告侧 |
| Clio | 2026 | 美国 | 135,000+ 名人身伤害律师(约 10%) | 广义律师群体中的原告侧律师子集 | 中 | 供应商汇编口径,不是政府原始表格 |
| Clio | 2026 | 美国 | 每年约 400,000 起 PI 索赔 | 用索赔流测算年度案件量,主要发生在州法院 | 中 | 索赔不等于律所、席位或软件预算 |
| Clio / IBISWorld 引用 | 2025 | 美国 | $61.7B PI 律师行业收入 | 支撑原告律所运营的经济池 | 中 | 收入不等于软件支出 |
| Lex Machina | 2025 | 美国联邦侵权案件案卷 | 场所责任和机动车案件均处在纪录高位 | 衡量原告工作的联邦侵权强度信号 | 中 | 仅限联邦;不是原告总量 |
| U.S. District Court(AFFF PFAS MDL,美国联邦地区法院) | 2026 | 美国联邦 MDL | 10,000+ 起相关案件;数万名原告 | 用单个 MDL 标记集体侵权运营规模 | 高 | 只是一个案件综合体,不代表整个 MDL 市场 |
| Duane Morris | 2025 | 美国 | 审查了 1,441 份集体诉讼裁判 | 集体诉讼复杂度 / 活跃度标记 | 中 | 裁判数量不是原告律所数量 |
本表采用多种有边界的视角,因为公开数据没有直接给出原告工作流 AI 的清晰 SAM 或 SOM。每一行都有来源支撑,但计量单位刻意不同:从人力、索赔、收入到诉讼复杂度标记,分别照亮市场的不同层次。
[CM013, CM014, CM015, CM017, CM038, CM039]近期法律市场调查显示,AI 使用已不再边缘;真正要讨论的是治理质量和原告业务适配度,而不是律师会不会接触 AI。
每一行都是单独的调查点估计,样本框不同。图表只保留一个一致单位——法律受访者在工作或日常流程中使用 AI 的比例——不把不可兼容的调查设计硬做平均。
[CM018, CM023, CM024, CM025, CM048]2.3 买方、用户、付款方与采用动态
原告律所的日常用户就是已经接触案件文件的人:获客接待团队、律师助理、律师、合伙人和运营人员。Eve 自己的页面明确把合伙人、律师、律师助理和运营专业人士放进工作流,其他供应商的原告运营指南则聚焦获客表单、医疗记录、起草、案件管理和和解后工作。但购买和付款不是个别律师的决定。Wolters Kluwer 报告称,超过 90% 的受访者如今在日常工作中使用至少一种 AI 工具;ABA Law Practice 对 8am 报告的摘要也称,法律 AI 采用率同比增长超过一倍。但同一份 8am 摘要称,许多律所仍缺乏正式政策、结构化培训和治理框架,这意味着规模化部署仍由管理合伙人、运营负责人、IT 或其他律所领导层掌握,而不是由单个律师选择工具。原告侧采用由异常直接的 ROI 杠杆拉动:低于 30 秒的获客响应可以提升转化,大多数 PI 案件庭前和解,重复性起草和记录工作会造成重大桌面时间瓶颈。因此,原告律所软件需求更像运营预算,而不是纯研究软件科目。[CM005, CM006, CM018, CM019, CM023, CM024]
| 细分市场 | 买方 | 用户 | 付款方 | 工作流 | 预算负责人 | 采用触发点 |
|---|---|---|---|---|---|---|
| 创始人主导 / 小型 PI 律所 | 管理合伙人 | 进线负责人、律师助理、律师 | 律所运营预算 | 从进线到索赔函的管线 | 管理合伙人 | 下班后线索漏接、桌面待办积压 |
| 区域性 PI 诉讼平台 | 业务负责人 + 运营 | 律师助理、初级律师、合伙人 | 运营 / 软件预算 | 医疗记录、起草、案件推进、和解管线 | COO / 运营负责人 | 不按同等比例增人也要扩规模 |
| 集体侵权原告平台 | 诉讼负责人 | 进线团队、医疗审查人员、原告协调员 | 诉讼运营预算 | 索赔人筛查、记录审查、事实表、MDL 协调 | 执行委员会 + 诉讼运营 | 索赔人数量骤增、中心化工作流承压 |
| 集体诉讼 / 雇佣原告律所 | 业务主席 | 初级律师、律师助理、沟通人员 | 业务组预算 | 索赔人筛查、诉状、证据开示、沟通、和解管理 | 业务负责人 + 运营 | 大量索赔人 / 成员沟通负荷 |
买方和预算负责人标签,是把原告工作流来源与法律 AI 治理来源综合得出的。公开证据支持由领导层治理部署、律师和律师助理使用的模式,但具体头衔会随律所规模变化。
[CM005, CM006, CM023, CM024, CM025, CM032]原告工作流 AI 由案件团队使用,但通常在律所或业务负责人层面评估和治理。
角色标签由工作流角色证据和法律 AI 治理证据综合而来。矩阵是定性的,因为公开资料没有按细分市场披露通用组织架构或供应商预算审批路径。
[CM005, CM006, CM023, CM024, CM025, CM032]原告律所通常从显性的运营痛点切入采用工作流 AI,先试点狭窄用例,再进入有治理背书的推广和续约。
[CM005, CM011, CM012, CM023, CM024, CM025]2.4 采用约束、监管与信任要求
原告侧工作流不同于 BigLaw 或被告侧,不只因为案子不同,也因为商业模式和风险模型不同。Thomson Reuters 指出,大型律所过去通常防御原告按胜诉收费案件,而规模通常较小的原告律所在案件成功时获取上行收益。在按胜诉收费经济学下,更快获客、更干净的案件筛选、更好的医疗综合、更快索赔函产出和更紧密的和解处理,都会直接进入现金转化。监管进一步强化了这种运营特殊性。ABA Formal Opinion 512 称,律师使用生成式 AI 仍负有胜任、保密、沟通、监督、诚实和合理收费义务。Rule 1.5 要求书面约定按胜诉收费机制,并在追回款项结束时出具书面汇款说明。Rule 1.15 和 ABA 信托账户记录规则要求客户资金隔离并维护款项分配记录。NYSBA 2026 年护栏文章称幻觉仍在导致制裁,Thomson Reuters 的伦理指南强调律师仍必须验证引用,并把 AI 当作助手而非自治律师。因此,信任控制、来源可追溯、人类审查和达到和解级别的运营纪律,是原告律所的核心购买条件。[CM007, CM008, CM009, CM010, CM011, CM012]
| 驱动因素 / 约束 | 方向 | 时点 | 证据 | 含义 | 尽调问题 |
|---|---|---|---|---|---|
| 风险代理费经济性奖励更快周转 | 驱动 | 当前 | 现金回收取决于案件筛选、速度和结果 | 工作流自动化能直接给原告律所带来 ROI | 按案件类型对标和解周期和桌面停留时间 |
| 进线响应更快,已签案件转化率更高 | 驱动 | 立即 | Eve 在一个进线案例研究中称转化率提升 40%,响应时间低于 30 秒 | 24/7 进线是楔子产品,不是锦上添花 | 在客户 cohort 中验证转化率增量 |
| 医疗记录和索赔函工作重复且高量 | 驱动 | 立即 | 原告工作流来源集中在记录、时间线、起草和证据开示 | 专用自动化比通用聊天更贴合原告侧重复任务 | 抽样检查错误率和律师复核负担 |
| 以和解为主的案件组合,更利好庭前运营而非纯庭审工具 | 驱动 | 当前 | Clio 称约 95% 的 PI 案件在审判前和解 | 诉前工作流速度比泛化研究广度更重要 | 衡量和解周期压缩和积压减少 |
| 法律 AI 使用已进入主流 | 驱动 | 当前 | 8am、Clio 和 Wolters 的数据中,法律工作流 AI 使用率为 69% 至 90%+ | 相比 2023 式试点市场,教育负担更低 | 按律所规模和业务组合拆分采用情况 |
| 治理和培训缺口拖慢铺开 | 约束 | 当前 | 8am / ABA 称许多律所仍缺正式政策、培训和治理框架 | 全律所部署需要领导层监督和变革管理 | 审查部署 playbook 和管理员归属 |
| 保密、幻觉和核验风险 | 约束 | 当前 | ABA 512、NYSBA 和 Thomson Reuters 伦理指引都强调能力胜任和核验 | 人工复核和来源可追溯是采购硬条件 | 检查引用可追溯性和 QC 工作流 |
| 信托账户和风险代理费合规 | 约束 | 当前 | 规则 1.5、1.15 以及信托记录规则约束汇款和客户资金 | 即便起草质量够好,和解 / 分配错误也可能挡住采用 | 审计和解声明、信托分类账和付款控制 |
各行聚焦会实质改变采用速度、续约韧性或估值相关性的因素。证据混合了直接工作流来源、法律市场调研和有约束力的职业责任规则。
[CM005, CM007, CM008, CM009, CM010, CM011]2.5 图表
03竞争对手
3.1 直接原告 AI 同行与 Eve 的最近重叠
与 Eve 最直接重叠的是 EvenUp 和 Supio,而不是更广法律 AI 领域。EvenUp 已远远超出索赔函,扩展成面向原告的索赔平台,覆盖获客接待、治疗、索赔函、谈判、证据开示、审判、高管分析和沟通智能体。这让它不只是诉前点工具,但其公开叙事仍根植于索赔智能和人身伤害规模。Supio 与 Eve 当前方向重叠更直接。它明确营销获客接待、时间线、索赔函、诉讼起草、跨案件分析和律所情报,同时强调群体侵权和人工验证输出。换句话说,Supio 想拥有的不只是一个工作流,而是围绕原告诉讼的记录、论证和洞察层。Darrow 相关但不同:它在更上游做法律风险暴露发现和事项发起,而不是日复一日运营原告律所。ProPlaintiff 和 LawPro.ai 等较小进入者也说明,医疗时间线、索赔函生成和智能体式案件支持已经在催生更窄挑战者。[CP003, CP004, CP005, CP006, CP008, CP009]
| 竞争对手 | 类别 | 规模 / 融资 | 目标细分市场 | 核心工作流范围 | 战略解读 |
|---|---|---|---|---|---|
| Eve | 参照平台 | 2025 年多轮披露融资合计 $164M;当前官网称服务 1,200+ 家律所 | 覆盖 PI、雇佣、SSDI、工伤赔偿和医疗事故的原告律所 | 进线、医疗概览、起草、研究、证据开示、沟通和律所分析 | 在已审查样本中,Eve 对原告运营系统的野心最宽 |
| EvenUp | 直接原告 AI 同行 | 总融资 $385M;估值 $2B+;声称服务 2,000+ 家 PI 律所 | 高量 PI 和索赔业务 | 进线、治疗、索赔函、谈判、证据开示、审判、沟通、分析 | 原告诉前和索赔运营中规模最强的对手,但相比 EveOS 仍更偏索赔中心 |
| Supio | 直接原告 AI 同行 | 2025 年 Series B 后总融资 $91M;客户名称包括 PI 和集体侵权律所 | 处理复杂、重文档诉讼的 PI 和集体侵权律所 | 进线、时间线、索赔函、案件经济性、诉讼起草、律所情报、集成 | 在医疗记录深度、诉讼准备度和核验叙事上,公开层面最强的对手 |
| Darrow | 相邻的获案 / 情报竞争者 | 行业报道披露融资 $63M;声称覆盖 80+ 个组织和 10K+ 个活跃事项 | 寻找新兴风险敞口的律所、保险公司和合规团队 | 法律风险敞口发现、上游案件来源、监管和诉讼信号情报 | 与获案和组合构建相关,但不是同类原告律所操作系统 |
| ProPlaintiff | 新兴单点竞争者 | 已审查材料未公开融资 | 人身伤害律所 | AI 索赔函、摘要、律师助理、案件经理、医疗时间线、文档生成 | 规模较小,但明确在进线到和解运营这个楔子上进攻 |
| LawPro.ai | 新兴单点竞争者 | 已审查材料未公开融资 | 人身伤害律师和伤害索赔团队 | 可视化时间线、文件审查、案件估值、带引用的答案、法律文书 | 范围窄于 Eve,但当医疗时间线和索赔函准备主导采购标准时值得关注 |
| Litify | 既有工作流套件 | 企业 / 定制报价软件,法律工作流覆盖面广 | 高量 PI 律所以及更广泛的律所和法务部门用户 | 进线、事项管理、分析、沟通、支付、嵌入工作流的 AI | 最接近原告操作系统的既有类比,尤其适合已经想要记录系统的律所 |
| Filevine | 既有工作流套件 | 保留来源未公开量化规模;定价通过演示 / 定制打包推进 | 诉讼、PI 和事项管理重度律所 | AI 医疗时间线、证词情报、验证、文档、日历、支付 | 工作流深度真实存在,即便没有 Eve 式营销语言,也是强劲既有威胁 |
| CloudLex / CasePeer | 原告专用既有套件 | 已审查公开材料中的定价多为定制或演示驱动 | 原告精品律所和区域性 PI 律所 | 进线、事项管理、文档、医疗时间线、和解支持、报告、法律 AI | 前沿模型光环低于 Eve,但嵌入式工作流和原告可信度很强 |
| Harvey / CoCounsel / Lexis+ / Paxton | 通用法律 AI 替代品 | Harvey 估值 $11B;其他产品通过大型法律科技平台或 SaaS 订阅销售 | 覆盖律所和法务团队的广义法律买方 | 研究、起草、文档分析、尽调、助手或 agent 工作流 | 起草和研究层的严肃替代品,但不是原告原生的进线到和解系统 |
本表把原告原生 AI 直接同行、既有记录系统和更广义法律 AI 替代品分开;规模数值只使用公开披露数字。
[CP001, CP003, CP004, CP006, CP008, CP009]| 采购标准 | Eve | EvenUp | Supio | Litify / Filevine / CloudLex | Clio / MyCase / PracticePanther | Harvey / CoCounsel / Lexis+ | Paxton / 单点工具 |
|---|---|---|---|---|---|---|---|
| 原告专精 | 强 | 强 | 强 | 中 | 弱 | 弱 | 中 |
| 全律所操作系统野心 | 强 | 中 | 强 | 强 | 中 | 中 | 弱 |
| 24/7 进线与客户沟通 | 强 | 强 | 中 | 中 | 中 | 弱 | 弱 |
| 医疗时间线 / 记录综合 | 强 | 中 | 强 | 中 | 弱 | 弱 | 中 |
| 证据开示 / 诉讼起草 | 强 | 中 | 强 | 中 | 弱 | 中 | 弱 |
| 记录系统 / 计费 / 日历 | 部分 | 弱 | 部分 | 强 | 强 | 弱 | 弱 |
| 有依据的法律研究 / 引用 | 中 | 弱 | 中 | 弱 | 弱 | 强 | 中 |
| 公开定价透明度 | 弱 | 弱 | 弱 | 弱 | 强 | 弱 | 中 |
单元格是基于已审查产品界面和定价页综合出的、有证据支撑的序位判断,不是经过审计的基准优劣结论。
[CP002, CP003, CP009, CP021, CP022, CP024]将 Eve 和保留替代方案按原告专业化、工作流深度和操作系统野心做序数定位。
坐标轴是分析师基于已审阅公开产品材料综合出的序数评分;不是经审计的市场份额测量。
[CP001, CP003, CP008, CP015, CP021, CP022]3.2 既有工作流套件、现状替代品与预算所有者
Eve 不只是在和新的原告原生 AI 公司竞争。它还在和已经控制原告律所内部获客接待、日程、文件、计费、报告和日常案件流转的软件系统竞争。Litify、Filevine、CloudLex 和 CasePeer 是最重要的例子,因为它们已经贴近事项管理核心。尤其是 Litify 和 CloudLex,公开话术带有操作系统语言:从获客到和解、可配置工作流、分析和嵌入式 AI。Filevine 的公开功能集说明,哪怕品牌听起来没有 Eve 那么「agentic」,现有厂商仍很重要。它已经提供 AI 医疗记录分析、证词智能、验证工作流、日程和支付运营。CloudLex 和 CasePeer 更窄,但原告原生且运营嵌入更深,这意味着它们可以靠切换成本守住预算,而不是靠前沿模型话术。现状还包括 Clio、MyCase、PracticePanther 等更通用的执业套件,以及 Smokeball 这类文档重工具。它们在原告深度上较弱,但价格更清楚,对于还不想引入新 AI-first 操作层的小律所来说可能「足够好」。[CP021, CP022, CP023, CP024, CP025, CP026]
| 平台 | 价格 / 单位 / 合同模式 | 包含能力 | 折扣 / 未知项 | 竞争含义 |
|---|---|---|---|---|
| Eve | 已审查来源没有公开标价;销售主导或演示主导动线 | 原告原生 AI 平台,覆盖进线、起草、证据开示和分析 | 具体席位下限、实施费和按用量计费条款未披露 | 定价不透明或有助于企业打包,但削弱与更简单套件快速对标的能力 |
| EvenUp | 已审查来源没有公开价目卡 | 索赔情报、PLAAS、起草、沟通和全生命周期支持 | 按案件、按席位或按页计价的精确经济模型未公开 | 让 EvenUp 能按价值定价,但也让横向成本尽调更难 |
| Supio | 公开材料承诺固定价格、无平台费、无页数限制;未发布价目表 | Intake、时间线、诉讼文书起草、案件经济性、律所智能 | 未披露公开席位档位、最低承诺或服务成本 | 信息表达比 Eve 或 EvenUp 更买方友好,但采购仍需直接接洽 |
| Litify / Filevine / CloudLex / CasePeer | 已审阅公开材料多为定制报价或演示引导 | 案件管理、Intake、分析、文档,以及越来越多 AI 功能 | 公开比价空间有限,净价可能取决于实施范围 | 有利于既有厂商追加销售和打包,尤其是厂商已掌握核心工作流时 |
| Clio | 起价 $49/user,大型律所采用定制报价 | Intake、CRM、工作流自动化、计费、文档、客户门户、Manage AI | 最终套餐随档位、附加模块和企业谈判而变 | 小型律所若想要看得见的入门价格和广泛功能、又不需要原告专业化,Clio 更有吸引力 |
| MyCase | 按年计费为 $50 至 $130/user/month,按月计费随档位为 $60 至 $150 | Intake、工作流自动化、计费、执业管理、法律 AI 功能 | 附加模块和律所实际组合仍然重要,但定价远比原告原生 AI 同行清晰 | 中小律所若尚未准备更换系统,试点账更容易算 |
| PracticePanther | 按年计费为 $49 至 $89/user/month,按月价格更高、最高 $99 | Intake 表单、自定义工作流、日程、计费、会计、法律执业管理 | 高级财务功能位于更高档位 | 对预算敏感律所最透明,但原告原生差异化最弱 |
本表比较公开包装信号,而非谈判后净价或特定工作负载 ROI。
[CP012, CP027, CP029, CP031, CP046, CP049]| 竞争者类别 | 锁定或多栖驱动因素 | 公开证据 | 对 Eve 的含义 | 尽调问题 |
|---|---|---|---|---|
| Eve 本身 | 可接入现有 CMS,也可独立运行;部署摩擦更低,但硬锁定可能受限 | Eve 称其 AI-ready data 可配合现有 CMS 使用,也可独立使用 | 落地动作强;除非 Eve 成为日常记录系统,否则护城河偏弱 | 询问客户中将 Eve 作为主系统与作为叠加层使用的比例 |
| Supio 及类似叠加层 | 集成优先架构让团队无需替换 CMS 就能叠加 AI | Supio 点名可集成 Westlaw、Litify、MyCase、CasePeer | 买方整合前可能同时运行多层 AI | 要求提供附加率、共存模式,以及面对叠加式部署的赢单 / 输单情况 |
| EvenUp | 专业数据和工作流训练可在 PI 运营中形成流程依赖 | EvenUp 营销 Piai,并覆盖 PI 案件生命周期自动化 | 即便 Eve 扩展到全律所,诉前专业化仍可能保持粘性 | 询问 EvenUp 是从 demand letters 扩展到记录系统,还是继续保持模块化 |
| Litify / Filevine / CloudLex / CasePeer | 现有案件数据、Intake 流程、报表、日历和计费构成典型记录系统转换成本 | 公开产品页面强调 Intake、案件计划、分析、文档和支付 | Eve 面对的最硬既有厂商障碍,尤其是在成熟原告律所 | 衡量迁移成本、实施时间,以及 Eve 是替代还是与各套件共存 |
| Clio / MyCase / PracticePanther | 更低的公开入门价和更广的执业覆盖降低小型律所试点摩擦 | 公开分层定价和通用执业管理广度披露清晰 | 若律所偏好更便宜、够用的软件叠加 AI,Eve 的 SMB 触达会被压住 | 询问何种律所规模和案件复杂度会稳定触发升级到 Eve |
| Harvey / CoCounsel / Lexis+ | 借助高端法律订阅、研究习惯和现有采购渠道分发 | Harvey 规模、CoCounsel 工作流主张、Lexis grounded 起草 / 研究均已公开 | Eve 赢下全工作流之前,这些厂商可先吃掉起草和研究预算 | 询问 Eve 多常被拿来对比通用法律 AI,而非原告专用软件 |
这片市场的转换成本耐久度并不均匀;最硬的锁定来自记录系统,而不是独立 AI copilots。
[CP002, CP010, CP021, CP022, CP024, CP026]3.3 通用法律 AI、相邻进入者与可替代范围
Harvey、CoCounsel、带 Protégé 的 Lexis+ 和 Paxton 之所以重要,是因为它们可以吸收 Eve 工作流的一部分,而不必成为原告操作系统。Harvey 现在销售能端到端执行法律工作的智能体,并拥有规模、融资和律所采用基础;当买方想要一个跨多个执业领域的高端 AI 伙伴时,它可能成为严肃采购替代。CoCounsel 和带 Protégé 的 Lexis+ 带来另一种分发力量:可信法律内容、法律研究工作流、尽调工具和有依据的引用基础设施。这些平台不是围绕原告获客接待、医疗时间线、和解运营或律所专属 PI 分析构建,但仍可以商品化研究、起草和审查。Paxton 则更偏向光谱另一端的 copilot。其公开材料强调起草、文件分析、上下文研究、医疗时间线和计费摘要,因此对于想要 AI 辅助、但不替换现有系统记录的小律所,它可能成为一层工具。含义是,Eve 不能只靠起草或研究质量防守;它需要公开资料中更广法律 AI 尚未展现的原告原生工作流深度和运营杠杆。[CP033, CP034, CP035, CP036, CP037, CP038]
用视觉方式概括哪些竞争者类别能在原告工作流广度上对标 Eve,哪些只能替代狭窄层。
热力图标签概括的是有证据支撑的姿态和工作流适配度,不是实测基准优势。
[CP002, CP009, CP020, CP021, CP022, CP024]3.4 差异化耐久性、切换成本与 Eve 仍可取胜之处
Eve 最好的公开论点不是它把某个原告任务做得比所有人都好,而是它试图连接整条原告运营闭环:常久在线获客接待、案件资格判断、医疗综合、起草、研究、证据开示和律所层面分析,都放在一个原告原生界面里。这比 EvenUp 公开的诉前规模故事更有野心,比 LawPro 或 ProPlaintiff 的点解决方案切口更宽,也比既有案件管理套件更 AI 原生。但耐久性仍有条件。Supio 的话术显示,诉讼深度和人工验证已经成为攻击 Eve 的活跃路线。EvenUp 的融资和客户规模说明,诉前专业化仍可能跑赢更宽平台叙事。Litify、Filevine、CloudLex 和 CasePeer 掌握系统记录关系,制造真实切换摩擦。通用法律 AI 供应商也可以把起草和研究价值压缩进更广订阅。因此,Eve 的护城河大概率取决于它能否在现有厂商和通用 AI 供应商追上可见功能集之前,把原告专业化加工作流广度转化成数据飞轮、日常运营依赖和可衡量 ROI。[CP001, CP002, CP005, CP011, CP014, CP040]
| 护城河主张 | 威胁 | 严重性 | 缓释 / 尽调问题 |
|---|---|---|---|
| 只服务原告的定位让工作流适配和销售话术更锐利 | EvenUp、Supio、CloudLex、CasePeer、LawPro、ProPlaintiff 也直接卖进 PI 工作流 | 中 | 验证 Eve 赢单更多来自广度、品牌,还是可衡量的案件结果 |
| 从 Intake 到 discovery 的广度支撑操作系统叙事 | 起草和研究层可能被 Harvey、CoCounsel、Lexis+、Paxton 商品化 | 高 | 按工作流衡量日活使用,而不只看 logo 数 |
| 原告原生数据结构可能形成可防守飞轮 | EvenUp 营销最大 PI 数据集,Supio 营销已验证输出和案件智能 | 高 | 要求提供专有数据优势、反馈闭环和长期结果改善的证据 |
| 可选 CMS 叠加层降低采用摩擦 | 同样的可选性也会保留多栖使用,削弱锁定 | 高 | 检查 Eve 能否成为主要运营层,而不只是生产力附加工具 |
| 资本和品类势能让 Eve 能参与企业式交易 | EvenUp 披露融资规模更大,Harvey 的资本获取能力远超原告品类 | 中 | 对比同样融资充足的竞品,审阅招聘节奏、服务能力和销售效率 |
| 既有案件管理厂商在 AI 创新上可能更慢 | 但它们已经掌握案件数据、Intake 流程、报表、支付和用户习惯 | 高 | 研究 Eve 面对 Litify、Filevine、CloudLex、CasePeer 的实际替换赢单率 |
| 买方信任可能偏向已验证、原告专用工作流,而非通用 copilots | Supio 直接从幻觉、集成和诉讼级验证角度攻击 Eve | 高 | 审阅质量控制、来源可追溯性,以及复杂诉讼案件的客户推荐 |
| 定价不透明保留灵活性 | 但也让 ROI 更难对标 Clio、MyCase、PracticePanther 的透明套件价格 | 中 | 要求提供当前价格卡、实施费,以及按律所规模划分的回本基准 |
核心耐久度问题在于:既有厂商和通用 AI 抹平可见功能差距之前,Eve 能否把广度和原告专业化变成日常运营依赖。
[CP005, CP011, CP022, CP033, CP040, CP043]对最可能决定 Eve 能否守住相对原告同业、既有厂商和更广义法律 AI 差异化的公开维度做序数评分。
评分是分析师基于已审阅公开证据形成的序数判断,并非公司报告的 KPI。
[CP011, CP033, CP043, CP044, CP045, CP046]3.5 图表
04财务
4.1 收入模式、定价不透明与客户推导出的收入线索
从财务形态看,Eve 更像面向原告法律的直销型 B2B SaaS 平台,而不是自助式法律工具。官网持续销售完整操作系统——EveOS、AI Intake、Agents、起草、证据开示和分析——但从不公布价目表。公司转而把潜在客户导向演示和销售沟通,同时 Head of Legal 岗位明确提到 MSA、Deal Desk 审查、客户协议和合作伙伴合同。这个组合强烈指向合同驱动销售、协商定价,以及收入确认前的实施或法律开销。 最好的公开牵引力信号是客户数量和工作流量,而不是美元收入。官方 Series B 材料称,截至 2025 年 9 月,Eve 已增长到超过 450 家律所、每年超过 200,000 个法律案件、客户追回金额超过 $3.5 billion。到 2026 年,官方招聘信息把规模口径提高到 1,000-plus 家律所,甚至称收入环比翻倍,但公司仍未披露绝对 ARR 或收入。因此只能谨慎做基准建模。相邻法律软件供应商公开收费约为每用户每月 $49 到 $130,足以说明原告律所已经支付有意义的软件预算。但即便把这些基准套到 1,000 家律所,也只能得到很宽的个位数百万到数千万美元年化支出区间。换句话说,Eve 显然拥有可货币化规模,但公开资料仍无法精确钉住实际 ACV、席位数或已确认收入。[CI001, CI002, CI004, CI005, CI007, CI008]
| 收入流 | 机制 | 单位 | 当前价值 / 状态 | 收入质量 | 尽调问题 |
|---|---|---|---|---|---|
| 核心平台订阅 | 覆盖 Intake、起草、discovery 和分析的原告法律操作系统,通过直接合同销售 | 按律所 / 按用户 / 按合同 | 已上线;定价未公开披露 | 若嵌入日常工作流,可能具备经常性和粘性,但标价和计费节奏未公开 | 要求按 cohort 提供 ACV、计费频率、席位数和 logo 留存 |
| AI Intake 模块 | 对线索做资格判断,并自动化前台 Intake 的工作流模块 | 按律所 / 模块附加 | 40+ 家律所 beta 后,2026 年 GA | 若收入与线索捕获绑定,可能提高扩展 ARR,但附加率和定价未披露 | 要求提供模块附加率、转化提升,以及任何按用量计费部分 |
| Agents / EveOS 扩展 | 客户在同一平台内添加起草、审计、分析和其他 AI workers | 按律所 / 工作流 / 用户 | 已上线;公开叙事强调扩展,未公开价格 | 若模块叠加进现有账户,可能成为强 NRR 杠杆 | 要求按模块提供收入结构,以及来自现有账户的扩展 ARR |
| 入职 / 实施层 | 客户故事和招聘暗示其需要高接触部署、法律审阅、客户成功和工作流设计 | 项目 / 打包服务 / 已包含 | 有迹象但未公开定价 | 可帮助采用,但相比纯软件会压低近期毛利率 | 要求按客户分层提供服务收入、实施成本和上线时间 |
| 续约与交叉销售 | 现有律所从单一工作流扩展到更广的律所运营 | 账户扩展 | 案例研究和演进中的产品线提供支撑,但未披露续约指标 | 若工作流粘性强,收入质量可能高;没有留存数据时公开层面仍未证实 | 要求提供 NRR、总留存、流失率和多产品采用率 |
各行区分已上线产品界面、隐含服务和扩展机制;公开来源未披露实际价格卡或收入结构。
[CI007, CI008, CI009, CI010, CI011, CI027]| 销售动作 / 基准 | 公开价格信号 | 披露内容 | 来源 | 含义 |
|---|---|---|---|---|
| Eve 核心平台 | 无标价 | 公开产品页面将买方引导至演示 / 预约通话流程 | Eve 主页、平台页和 Intake 页 | 实际 ACV 可能明显高于 SMB 基准,但公开文件无法显示具体位置 |
| Eve 签约姿态 | 谈判式合同 | Head of Legal 岗位提到 MSAs、客户协议、Deal Desk 和合作伙伴合同 | 2026 年官方招聘信息 | 销售可能涉及法律审阅、采购和摩擦更高的企业式动作 |
| Clio 基准 | 起价 $49 per user monthly | 更高增长、AI 和 PI 专用模块转向演示 / 定制价格 | Clio 官方定价页 | 说明原告法律律所已经为核心工作流软件付费,且高级模块常常脱离公开价目表定价 |
| MyCase 基准 | $50 / $100 / $130 / 用户 / 月 | 已发布 Basic、Pro、Advanced 年付档位 | MyCase 官方定价页 | 可作为律所软件按用户预算的低至中端基准 |
| PracticePanther 基准 | $49 / $69 / $89 / $114 / 用户 / 月 | 已发布年付 Solo、Essential、Business、Business Pro 档位 | PracticePanther 官方定价页 | 为小型和中端律所提供第二条相邻基准区间 |
基准行是相邻法律软件可比对象,并非 Eve 实际成交定价的证据。
[CI008, CI009, CI028, CI029, CI030, CI031]Eve 在公开定价不透明的情况下,如何把直销转化为经常性软件收入。
这是基于公开产品和招聘信息搭出的结构性收入地图;不披露 Eve 的会计政策或实际合同条款。
[CI007, CI008, CI009, CI011, CI027]4.2 GTM 动作、客户 ROI 与单位经济代理指标
Eve 的商业论点在客户故事暴露工作流数学的地方最强。Archuleta 称,该律所每月处理约 1,000 条线索,覆盖全美 50 个州;约一半来电者自愿选择 AI 获客接待;且未增加员工就实现产能翻倍。Laurel 称索赔函起草从两到四小时降到约 15 分钟,其 demand agent 让每周寄出的索赔函从 48 封增至 104 封,并把人工接触时间降到每封三到五分钟。Frontier 称 Eve 把数天的质询书工作压缩到 45 分钟,现在团队能在同一时间内完成五倍工作。这些不是经审计财务报表,但它们是最清楚的公开证据:Eve 卖的是人力杠杆和吞吐量,不只是更漂亮的起草。 更难的问题是,这种工作流杠杆能否转化为有吸引力的软件经济性。上市公司代理指标显示可以:Intapp 2026 年 3 月申报数字显示 SaaS 增长 27%、cloud ARR 增长 31%、cloud NRR 为 123%;DISCO 则显示收入继续增长,但 GAAP 盈利仍为负。这一模式对 Eve 有参考意义。规模化法律软件公司可以复合增长经常性收入,同时比通用 SaaS 更长时间吸收增长投资。但 Eve 没有披露自身毛利率、服务收入占比、CAC、回本期或留存。因此,投资人有足够证据相信模型可能有吸引力,但没有足够证据在缺少 data room 支持时,从客户 ROI 闭环到单位经济。[CI009, CI010, CI018, CI019, CI020, CI021]
| 指标 / 代理 | 公开数值 / 状态 | 置信度 | 重要性 | 尽调问题 |
|---|---|---|---|---|
| 销售动作复杂度 | 通过 MSAs、Deal Desk 和客户协议做直接合同销售 | 中 | 意味着周期比自助式 SaaS 更长、售前成本更高,但可能支撑更大的 ACV | 要求提供 pipeline 阶段、赢单率、平均销售周期和实施时间线 |
| Intake ROI 代理 | 40+ 家律所 beta;Archuleta 的 AI-intake 选择率 50%;未增员情况下产能翻倍 | 中 | 若转化和人力收益成立,前台回本可被衡量 | 要求按 cohort 提供线索到签约案件转化的前后对比,以及工时节省 |
| 起草 ROI 代理 | Laurel 从 2–4 小时缩至 15 分钟;每周 demand letters 从 48 封到 104 封;每封需 3–5 分钟人工介入 | 中 | 显示显著人力杠杆和潜在贡献毛利扩张 | 要求提供 QA 返工率、人工审阅时间和实际节省的律师工时 |
| 大型部署代理 | Laurel 报告称上线 Eve 后拥有 100+ 名员工和 1,500+ 名活跃客户 | 中 | 表明 Eve 能支撑比纯 solo-practice 工具更大、更高价值的客户足迹 | 要求提供前 20 大客户席位数、ACV 和集中度 |
| 成熟法律软件代理 | Intapp:123% cloud NRR;DISCO:347 个客户 >$100k,同时仍为 GAAP 亏损 | 中 | 说明法律软件平台可扩张经常性收入,同时仍承担增长开支 | 要求提供 Eve 毛利率、NRR、>$100k 客户数和利润率桥 |
| 核心指标披露 | 毛利率、CAC、回本期、NRR 和收入结构未公开 | 中 | 没有这些指标,单位经济性只能保持方向性,无法进入可承销状态 | 要求提供董事会 KPI 包,包含软件与服务拆分及 cohort 留存 |
各行把直接观察到的客户 ROI、基于代理的软件经济性,以及仍未披露的指标分开。
[CI009, CI010, CI020, CI021, CI024, CI025]展示从获客到 ROI 的定性路径,也说明为什么 Eve 公开单位经济仍只能用代理指标判断。
公开证据在工作流结果上很丰富,但在 CAC、毛利率和回本期上很稀薄,因此最后一个节点有意保持未解。
[CI009, CI018, CI021, CI024, CI026, CI042]4.3 资本充足性、烧钱代理指标与资本强度
财务上的关键正面是,Eve 不再像一家脆弱的产品前创业公司。已披露的 2025 年 Series A 和 Series B 至少贡献 $150 million,多份 2026 年公司控制来源把总额上调到超过 $160 million 或 $164 million。这一点重要,因为公司仍在积极招聘。Jobera 2026 年 6 月列出 34 个开放岗位,覆盖工程、数据、产品营销、安全、财务、收入运营、客户成功、销售和法律。公开薪资区间从低端约 $82,000 到高级工程领导岗位 $450,000,同时公司还宣传股权、福利、津贴和退休匹配。这些披露没有揭示实际薪酬总额,但显示出一个由风险资本支持的成本基础,与仍在产品、GTM、合规和客户运营上重投入的公司相符。 问题在于,资本充足性仍无法被精确承销。已审阅来源没有公布现金余额、月度烧钱、跑道月数、债务或信贷额度。即便招聘数据也有两面性:它显示公司可以花钱,但不显示 2025 年融资后还剩多少现金。最可辩护的看法是,Series B 实质降低了近期融资压力,且赞助方质量让 Eve 相比多数垂直 AI 创业公司有更好的再融资位置。但公开记录仍不能让投资人判断,Eve 是十二个月后就要再融资,还是在那之后仍资金充裕。[CI001, CI002, CI003, CI004, CI006, CI034]
| 项目 | 公开数值 / 状态 | 置信度 | 重要性 | 尽调问题 |
|---|---|---|---|---|
| 已披露资本基础 | 两轮 2025 年官方融资合计 $150M;2026 年公司材料将总额写作 $160M+ / $164M+ | 高 | 即便剩余现金不明,已融资本仍是最清晰的公开 runway 缓冲 | 核对种子轮历史、secondary 组成,以及当前不受限现金 |
| 投资人质量 | 投资方:Spark、Andreessen Horowitz、Lightspeed、Menlo | 高 | 若公司持续复利,强背书方能改善后续融资可得性 | 确认所有权、董事会权利,以及任何投资人 pro rata 承诺 |
| 手头现金 | 低 | 没有当前现金,无法精确估算 runway | 要求提供最新董事会现金余额和受限现金明细 | |
| 月度 burn | 低 | 招聘数据说明支出强度,但不能给出实际 burn | 要求提供月度净 burn、薪酬、云成本和客户成功支出 | |
| 招聘强度 | 2026 年 6 月有 34 个开放岗位,覆盖技术、GTM、财务和法务职能 | 中 | 显示公司继续投入增长,opex 可能上行 | 要求提供当前在岗人数、招聘计划和离职率 |
| 已列明薪资区间 | 已列岗位基本薪资为 $82k 至 $450k | 中 | 高端招聘区间意味着全负荷人力成本可观 | 要求提供全负荷薪酬预算、SBC 费用和招聘优先级 |
| 债务 / 项目融资 | 低 | 没有公开披露并不证明不存在义务 | 要求提供债务明细、授信额度、供应商融资和任何追索义务 |
null 表示该指标未在已审阅公开来源中披露,并不表示公司没有该项目。
[CI003, CI004, CI005, CI006, CI034, CI035]列出最影响 Eve 财务判断的公开数字边界,同时明确排除未披露的私营公司指标。
每一项都是公开边界或基准;没有一项是 Eve 披露的收入、现金或利润率数字。
[CI015, CI028, CI029, CI030, CI032, CI033]展示近期风险资金如何先流向招聘、产品和客户运营,再转化为耐久的经常性收入。
该图只描述资本流向;不暗示任何已披露的预算拆分、现金余额或债务安排。
[CI003, CI034, CI035, CI037, CI038, CI047]4.4 财务结论与尽调阻断项
公开证据支持一个建设性但不完整的财务结论。Eve 几乎肯定通过直接合同,把经常性软件卖进一个真实、有预算的痛点。客户故事显示,获客接待、起草和诉讼准备中存在实质人力杠杆。赞助方质量强,已披露资本有分量,公司仍像成长阶段企业而不是收缩企业一样招聘。这些事实支持 Eve 正在成为原告律所重要垂直软件平台的判断。 但这仍不同于说该业务可以在 $1 billion-plus 估值上公开承销。最大的缺失项正是决定估值纪律的指标:绝对 ARR 或收入、实际定价、ACV 分布、毛利率、服务收入占比、CAC、回本期、留存、客户集中度、现金、烧钱和债务。负面资料也重要。2026 年 3 月引用事件不能证明系统性产品失败,但它说明法律 AI 供应商可能需要持续投入 QA、审计、人工审查和客户培训。综合来看,Eve 像一个快速增长的合同型 SaaS 平台,拥有可信类别领导力和清晰 ROI 叙事,但财务论点更多依赖增长信号和客户轶事,而不是已披露收入质量。正确的下一步不是讲新故事,而是拿指标包。[CI017, CI031, CI041, CI042, CI043, CI044]
| 缺失指标 | 重要性 | 当前最佳公开代理 | 当前状态 | 精确尽调路径 |
|---|---|---|---|---|
| 绝对 ARR / 按产品线收入 | 没有该指标,估值逻辑无法绑定规模或增长效率 | 客户数量、2X QoQ 收入主张和基准价格区间 | 未公开披露 | 要求提供月度经常性收入、ARR bridge 和按模块收入结构 |
| 实际成交价格 / ACV / 席位组合 | 需要用它把「1,000+ 家律所」转化为实际变现 | Clio、MyCase、PracticePanther 的法律软件价格卡 | 未公开披露 | 要求提供价格手册、平均每客户席位数和 top-decile ACV |
| 毛利率及软件与服务组合 | 需要用它判断收入质量,以及 onboarding 或人工审阅是否拖累利润率 | 客户故事加法律软件可比代理 | 未公开披露 | 要求提供毛利率桥,以及服务附加率 / 服务毛利率 |
| CAC、回本期和销售周期转化 | 决定增长是高效,还是仅靠资本堆出来 | Head of Legal 岗位、客户成功招聘和交易复杂度线索 | 未公开披露 | 要求提供漏斗转化、销售周期、按渠道 CAC 和按分层回本期 |
| 留存、流失和集中度 | 这是理解 $1B+ 估值背后客户基础耐久度的关键 | 2025 年 450+ 家律所、2026 年 1,000+ 家律所暗示扩张,但不说明留存质量 | 未公开披露 | 要求提供 NRR、总留存、流失率、cohort 扩张和头部客户集中度 |
| 现金、burn、runway 和债务 | 这是承销资本充足性和下一轮时点的核心 | 已披露融资加激进招聘和高端薪资区间 | 未公开披露 | 要求提供董事会现金文件、月度 burn、runway 模型和债务明细 |
这些是从叙事尽调推进到财务承销所需的最低缺失项。
[CI017, CI031, CI034, CI035, CI042, CI047]05产品与技术
5.1 工作流深度已从获客延伸到诉讼执行
Eve 的公开产品表面已不只是起草助手。2026 年 6 月官网、2026 年 1 月 Eve 2.0 发布材料和 2026 年 6 月 EveOS 发布报道,都描述了一套工作流栈:从多语言获客接待开始,进入诉前时间线和索赔函起草,再延伸到证据开示、证词审查、交叉询问支持和事项级审计。这种广度重要,因为原告律所不会孤立购买点工具;运营痛点在于获客接待、记录审查、起草、客户沟通和诉讼准备之间的交接。基于这里审阅的证据,Eve 试图占据这些交接点,而不只是一次帮忙处理一份文件。 模块级证据最强的地方是获客接待、医疗概览和起草。Eve 称其获客层 24/7 接听电话和邮件,在通话中签下合格客户,并支持 28 种语言;intake GA 文章还加入 40-plus 家 beta 律所、线索评分、结构化摘要和 CMS 集成。医疗概览页面对法律 AI 营销来说异常具体:它列明时间线、坏事实、ICD codes、损害赔偿台账、回链到记录页、手写记录 OCR 和 15–20 分钟生成时间。客户案例研究再把这些模块连接到真实工作流:Laurel 的 demand agent 从证词自动触发,并把人工接触时间缩短到几分钟;Hershey Law 则描述了从获客到审判支持使用 Eve。这足以得出工作流深度真实存在的结论,尽管多数定量结果仍来自公司控制或赞助来源。[CE001, CE002, CE003, CE004, CE005, CE006]
| 模块 | 主要用户 | 成熟度 / 状态 | 已确认能力 | 差异化 / 转换成本 | 尽调缺口 |
|---|---|---|---|---|---|
| Intake / Jenny | Intake 团队和前台 | 已上线;2025 年 10 月起 GA,并在 EveOS 中扩展 | 24/7 语音和邮件处理、实时签约、多语言 Intake、线索评分、CMS 同步 | 掌握首次接触,并可把结构化数据直接送入下游案件工作流 | 要求提供经审计的转化提升、假阳性 / 假阴性率和来电升级政策 |
| 医疗概览 | 律师助理和诉前团队 | 已上线 | 时间线、不利事实、ICD 编码、损害赔偿台账、页级来源链接、疑难扫描件 OCR | 把原始医疗材料转成可复用的结构化案件上下文 | 索取基准准确率、审阅者改写率和大文件失败日志 |
| 起草智能体 | 律师助理、律师、法律运营 | 已上线 | 按律所风格起草索赔函、诉状、动议和证据开示草稿,并由庭审笔录触发 | 固化律所模板,压缩重复起草瓶颈 | 索取提示词 / 版本控制、修订遥测和律师采纳率 |
| Auditor | 律师和审阅负责人 | 已上线;2026 年增强 | 每晚审阅在办案件,查找缺口、遗漏伤情、缺失文件和下一步动作 | 加上一层持续 QA,而不是一次性提示 | 索取问题标记的精确率 / 召回率和重大误报率 |
| Analyst | 律所管理层和运营团队 | 2026 年 Beta / 早期访问 | 用自然语言查询运营,并通过仪表板查看收入节奏、业绩和案件分布 | 可能把产品使用痕迹转成管理报表和流程调优 | 索取 GA 日期、数据模型覆盖范围和基于角色的访问控制 |
| Atlas / AI 就绪数据 | 全律所 | 2026 年 Beta / 早期访问 | 从 CMS、邮件、诉讼文件、记录和通信中拉取并自更新案件档案 | 做出数据可携性和共享上下文,用得越多越能累积 | 索取血缘模型、冲突解决逻辑和导出工具 |
| 通信智能体 | 案件经理和接案团队 | 2026 年新推出 | 覆盖 31 种语言的外呼跟进、记录请求、客户入职和案件更新 | 把自动化延伸到琐碎但高频的客户 / 服务方沟通 | 索取退出控制、通话录音治理和语言质量审计 |
| 研究 | 律师和起草人员 | 2026 年新推出 | 按司法辖区检索判例,标记被推翻案件,并链接来源段落 | 如果可靠,就把法律研究嵌进工作流,而不是另开一个工具 | 索取权威来源覆盖、漏检率测试和引文核验日志 |
表格把已有直接公开证据的模块,与仍处于 Beta 或早期访问状态的模块分开;多数量化结果主张仍由公司单方控制。
[CE001, CE003, CE004, CE009, CE013, CE014]| 用户任务 | 当前工作流 | Eve 方案 | 已衡量 / 声称的收益 | 限制 |
|---|---|---|---|---|
| 捕获非工作时间线索 | 前台、语音信箱或外包接听服务 | AI intake 接听并筛选来电,实时签约客户,并路由结构化摘要 | 公司称不会漏掉高价值来电;赞助报道称某律所转化率提升 10% 到 35% | 没有公开分母说明线索质量漂移或放弃率 |
| 汇总医疗记录 | 人工跨服务方整理时间线 | Medical Overview 生成时间线、不利事实、ICD 编码和带来源链接的损害赔偿视图 | 公司称生成需 15–20 分钟;客户引述称从几小时或几周降到几分钟 | 没有公开基准说明抽取错误率或审阅者修正负担 |
| 准备索赔函 | 人工基于笔录、表格和记录起草 | 智能体从案件数据自动触发、按律所风格起草索赔函 | Laurel 称人工触达时间为 3–5 分钟,每周可发 48 到 104 封信 | 案例研究数据由公司控制,且未按队列调整 |
| 回应证据开示 | 模板加人工问题识别 | 证据开示指南和产品页面把 Eve 定位为可处理请求、回应、异议和摘要 | 声称周转更快、重复起草更少 | 没有公开对照组与专门诉讼审阅工具比较 |
| 支持庭审准备 | 人工检索庭审笔录并通宵准备 | Hershey 提到实时弹劾支持、limine 动议跟踪和隔夜策略摘要 | 客户证据显示工作流深度不止诉前环节 | 证据来自一份公司案例研究,而非独立法庭观察者 |
| 审阅整个案件清单 | 合伙人审阅和电子表格仪表板 | Auditor 每晚审阅案件,Analyst 用自然语言回答运营问题 | 可能把 QA 和管理闭环压进产品本身 | Analyst 仍处 Beta,公开分析截图有限 |
| 协同现有系统 | 分散的 CMS、邮件、文档和通话工具 | Atlas 和 Clio 同步意在跨系统规范化并迁移数据 | 可能减少来回切换和数据漂移 | 公开 API 与支持集成文档仍很薄 |
已衡量收益混合了客户轶事、赞助报道和公司主张;看起来没有一项经过独立审计。
[CE003, CE004, CE009, CE011, CE013, CE016]一个典型 Eve 工作流从线索捕获开始,经过结构化案件数据、起草、复核和持续监控,而不是孤立提示词。
[CE003, CE004, CE009, CE014, CE016, CE023]5.2 EveOS 增加案件数据层、分析能力和可见技术架构信号
2026 年最清楚的产品变化是 EveOS,它把 Eve 从工作流自动化推向操作系统野心。Lawnext 和 Business Wire 报道一致认为,EveOS 增加了 Atlas,一个自更新案件数据层,会从案件管理系统、邮件、法院文件、医疗记录和客户沟通中拉取数据;当事实来源冲突出现时,系统应将其标记给人工审查,而不是静默解决。同一发布还把 Analyst 产品扩展到 beta,用自然语言报告和解历史、盈利能力、节奏和案件分布;Communication Agents 则跨 31 种语言自动化外呼跟进和记录请求任务。Eve Research 也被定位为智能体内部的原生法院判例层,包括被推翻案例标记和指向来源段落的引用链接。 公开技术信号支持 Eve 不只是在通用模型外套壳,但还不足以证明可防守的模型质量。Built In 列出一个现代应用栈——Django、PostgreSQL、Python、React 和 TypeScript。ML Engineer 岗位明确提到特定领域微调、评估框架、生产部署和产品使用信号,以及与 OpenAI 和 Anthropic 直接合作。工程博客更有信息量:一篇文章称 Eve 内部 WALL-E agent 已经撰写 monorepo 合并请求的 22%;另一篇解释公司在 shard 不稳定和每月超过 $10,000 AWS 成本后,把每案搜索基础设施从 OpenSearch 迁走。这些是可信的架构和开发者信号。它们说明 Eve 底层确实有基础设施和评估工作,但仍未披露公开 benchmark 分数、正常运行历史,或用于客户法律输出的准确检索与验证设计。[CE007, CE008, CE019, CE020, CE021, CE022]
| 层 / 流程 / 组件 | 作用 | 依赖 | 风险 |
|---|---|---|---|
| Web 应用栈 | 运行面向操作员的产品和内部工具 | Built In 列出 Django、PostgreSQL、Python、React 和 TypeScript | 技术栈可见有帮助,但没有公开架构文档展示租户边界或服务拆分 |
| 案件数据层(Atlas) | 把诉讼文件、记录、邮件、CMS 数据和通信规范成实时案件视图 | 需要接入上游系统,并处理 schema 映射和冲突 | 公开材料展示的是意图,不是数据血缘细节或对账 SLA |
| 单案搜索 / 检索 | 为搜索、检索和上下文 grounding 做按案件索引 | OpenSearch 分片不稳定且 AWS 成本高之后迁移到 Turbopuffer | 没有公开披露召回指标、权限模型或索引重建频率 |
| 智能体编排 | 监控案件变化,触发起草、审阅和通信动作 | 工作流规则、案件状态识别和人工批准门 | 如果触发逻辑薄弱,自主触发可能让错误悄悄扩散 |
| 核验层 | 输出离开律所前,提供内嵌来源、来源段落链接和人工批准 | 依赖可靠的引文绑定和 UI 层审阅体验 | 法庭级失败案例说明,真实使用中核验仍可能失效 |
| 集成层 | 从外部系统同步案件详情、联系人、备注和其他记录 | 目前有 Clio 同步;Atlas 暗示更广泛的 CMS / 邮件 / 会计连接 | 公开 API 文档和支持伙伴清单不可得 |
| 模型开发和评测 | 微调、评估并部署法律工作流 AI 系统 | 依赖领域数据、使用信号和与基础模型提供商的协作 | 没有公开基准包覆盖幻觉、延迟或批准通过率 |
| 工程自动化 | 内部 WALL-E 智能体加快软件交付 | Slack 触发后台执行、完整应用启动、浏览器测试和 MR 创建 | 开发者生产率提升不会自动转化为终端用户法律输出更可靠 |
架构行结合了直接技术披露,以及从招聘和工程文章中做出的谨慎推断;缺少公开文档仍是重大问题。
[CE023, CE029, CE034, CE035, CE036, CE037]| 日期 / 阶段 | 功能 / 里程碑 | 状态 | 含义 | 来源 |
|---|---|---|---|---|
| 2025-08 工程文章 | 搜索基础设施向 Turbopuffer 迁移 | 已内部上线 | 说明按案件检索重要到足以推动基础设施迁移和成本优化 | Eve 工程博客 |
| 2025-10 GA | Eve AI Intake | 已上线 | Intake 从 Beta 进入一般可用,成为更广平台的入口 | Eve intake 发布文章 |
| 2026-01 发布 | Eve 2.0:Agents、Auditor、Analyst | 已上线 / 已宣布 | 产品叙事从提示工具转向按角色分工的 AI 劳动力 | Eve 官方发布和 Lawnext 报道 |
| 2026-03 工程文章 | WALL-E 后台编码智能体 | 已内部上线 | 说明公司愿意把智能体工作流用于工程运营,而不只放在客户营销里 | Eve 工程博客 |
| 2026-06 发布 | 带 Atlas、Communication Agents 和 Research 的 EveOS | 已上线 / 已宣布 | 迄今最大的平台扩张;强化操作系统野心 | Lawnext 和 Business Wire / Morningstar |
| 2026-06 Beta / 候补名单 | Analyst Beta 和 Atlas 早期访问 | 尚未完全 GA | 关键模块仍需要采纳证据和生产成熟度证据 | Lawnext 和 Business Wire / Morningstar |
公开路线图可见性来自已发布版本、Beta 标签和工程披露;未发现带日期的前瞻路线图或状态页。
[CE019, CE020, CE023, CE025, CE027, CE036]公开可见的 Eve 技术栈:从外部系统和原始案件记录,到检索、代理执行,再到面向律师的工作流。
[CE007, CE008, CE023, CE024, CE025, CE027]Eve 依赖已连接的上游系统、检索基础设施、基础模型伙伴和人工检查点,才能让输出在法律工作流中可用。
[CE016, CE023, CE024, CE029, CE036, CE037]公司页面和案例研究反复写到的工作流,是 Eve 最成熟的部分;操作系统级分析和互操作性仍有 beta 阶段缺口。
成熟度判断来自已保留公开来源的定性评估,反映的是证据密度,不是经审计的内部就绪度。
[CE023, CE025, CE027, CE029, CE030, CE038]5.3 验证控制可见,但公开资料无法消除模型风险担忧
Eve 在营销其验证姿态时异常明确。医疗概览页面称,每个数据点都会链接回来源文件和页码,直接引用与 AI 摘要在视觉上分开。Working With Eve 进一步声称内联来源、一键验证和 AI 验证框架;安全页面则强调隔离、针对基础模型训练的提示和输出零留存处理、年度审计、HIPAA 合规和 SOC 2 Type II。这些控制与原告律所买方应当想要的东西方向一致:敏感工作流中的来源可追溯、人工批准和数据隔离。 但公开证据也显示,应该把这些控制视为缓释,而不是证明。Eve 隐私政策描述会与服务提供商、分析伙伴和广告伙伴共享网站个人数据;这是营销网站行为,不是事项文件处理,但仍说明公司有不同隐私表面和不同风险画像。更重要的是,2026 年负面来源直接反驳任何简单的「幻觉已解决」读法。LLRX 重点提到一封 2026 年 3 月律师道歉信:信中称 Eve 培训强调防护和「Trust but Verify」,但不准确引用仍进入法院。该律师无法判断失败来自软件还是自己的工作流,这正是问题所在:验证必须在混乱的真实使用中生效,而不只是营销演示。EDRM、LawAccounting 和 TechNewsWorld 都展示了更严酷的 2026 年背景:制裁、逻辑层幻觉和 shadow-AI 治理失败,让原生可审计性和人工审查成为基本要求,而不是单靠自身就能差异化。[CE010, CE016, CE024, CE030, CE031, CE032]
| 控制 / 认证 / 质量信号 | 状态 | 范围 | 缺口 |
|---|---|---|---|
| 输出离开律所前先经人工审阅 | 公司和 Lawnext 描述已确认 | 草稿、智能体和 Atlas 不确定性处理 | 没有公开数据说明改写率或升级频率 |
| 页级内嵌来源 | Medical Overview 和 Working With Eve 已确认 | 医疗记录和带引文答案 | 没有公开基准说明所有模块的来源链接完整性 |
| AI 验证 / 信任但核验框架 | 公司声称 | 营销语言和产品定位 | 2026 年 3 月的提交错误说明这只是缓解措施,不是低幻觉风险的证明 |
| SOC 2 Type II 和 HIPAA 声称 | 公司声称 | 包括医疗记录在内的敏感原告律所数据 | 未发现可下载审计报告或认证范围 |
| 加密、隔离、零留存姿态 | 公司声称 | 组织、用户和工作流隔离;不使用客户提示词 / 输出训练模型 | 没有详细公开密钥管理、租户隔离或子处理方架构 |
| 网站隐私 / 数据共享政策 | 已确认 | 网站和营销界面的个人数据 | 政策允许分析和广告伙伴接入;这与案件文件处理分属两件事,但仍是治理暴露面 |
| 外部法律 AI 制裁环境 | 由负面来源确认 | 法庭提交、法律运营和律所治理 | 制裁增多意味着买方应直接测试可审计性和审阅控制,而不是依赖口号 |
控制栈方向上较强,但多数证据仍停留在政策或营销层面,而不是经过基准验证的运营报告。
[CE010, CE016, CE024, CE030, CE031, CE032]5.4 切换成本看起来真实,但关键技术尽调项仍未披露
Eve 可能的护城河是运营嵌入,而不是单一突破性模型声明。Atlas 旨在把记录、文件、沟通和案件管理数据规范化进一个案件文件;Clio sync 直接把联系人、笔记和案件细节拉入 Eve;客户故事描述了律所专属 playbook、由证词驱动的起草触发和夜间审查循环。一旦这些工作流被调到某家律所的模板、获客逻辑和审批检查点上,替换成本就会上升,因为买方替换的不只是软件席位,还有被编码的流程。内部搜索基础设施文章强化了这一推断:逐案检索设计对需要事项专属回忆和可追溯性的律所说得通,也意味着切换会涉及工作流和数据架构迁移。 即便如此,从技术尽调角度看,这个产品仍无法完全公开承销。已审阅来源没有找到公开状态页、正常运行指标、事故披露、外部 API 参考、除 Clio 和通用 CMS 声明之外的具名集成目录,也没有关于幻觉率、引用精度或审查通过表现的基准证据。公司控制的规模口径也变化很快——1 月 500-plus 家律所、3 月赞助报道 800-plus、招聘信息 1,000-plus、6 月官网 1,200-plus——但没有独立分母说明活跃使用、留存或模块渗透。综合来看,Eve 像是少数几个朝原告律所工作流原生 OS 方向建设的法律 AI 供应商之一,具备真实开发者信号深度和可见产品广度。但技术尽调论点仍依赖私有 data room 来确认可靠性、评估纪律、部署摩擦,以及验证控制能否在法庭级压力下稳定运转。[CE007, CE023, CE025, CE026, CE029, CE034]
| 主题 | 已确认内容 | 仍由公司声称的内容 | 区分原因 | 尽调请求 |
|---|---|---|---|---|
| 工作流广度 | 独立报道和客户故事确认了 intake、起草、审计、分析和庭审支持使用 | 各模块的具体附加率和生产使用情况 | 只有模块被广泛采用,而不只是发布,广度才会带来粘性 | 索取模块级 DAU/WAU 和按队列渗透率 |
| 律所数量规模 | 公开来源显示,2026 年各类材料中的叙事从 500+ 家上升到 1200+ 家律所 | 活跃 logo、留存 logo,以及按模块拆分的付费律所 | 规模质量比标题 logo 数更重要 | 索取活跃客户数、GRR/NRR,以及付费与试点拆分 |
| 产品结果 | 案例研究显示了真实使用、笔录触发和法庭支持 | 多数生产率和和解提升指标 | ROI 主张支撑估值和切换成本叙事 | 索取前后对照队列、基线和可独立审阅的案件样本 |
| 核验质量 | 内嵌来源和人工审阅控制是真实产品概念 | 公开基准化的幻觉率、引文率和批准通过率 | 法庭风险取决于控制实际表现,而不是 UI 主张 | 索取评测包、抽样方法和事故后 RCA 流程 |
| 集成和数据可携性 | Clio 同步和 Atlas 意图已确认 | 支持系统广度、导出完整性和 API 成熟度 | OS 级定位取决于真实互操作性 | 索取 API 文档、伙伴清单和导出 / 导入测试数据 |
| 安全 / 合规 | 公开声称覆盖 SOC 2、HIPAA、AES-256、隔离和零留存 | 认证范围、已测试控制和事故历史 | 买方需要控制细节,而不只是标签 | 索取信任中心材料、最新报告日期和安全架构审阅 |
| 部署速度 | 公司称律所可在 90 天或更短时间内转为 AI-native | 中位上线时间、失败实施和服务负担 | 快速部署是销售叙事的一部分,也影响真实切换成本 | 索取上线漏斗指标和实施人员配置模型 |
本表把公开文件可见的产品事实,与仍需技术或运营数据确认的营销主张分开。
[CE023, CE029, CE030, CE031, CE032, CE040]5.5 图表
06客户
6.1 客户分层、执业领域契合与可见证据
公开证据支持一个比泛泛「法律 AI」更窄、更可信的客户故事。Eve 明确面向原告律所营销,而不是整个律所宇宙;其公开执业领域足迹已经覆盖人身伤害、劳动就业、工伤赔偿、Social Security disability 和医疗过失。具名客户集合也比单纯 auto-PI 切口更多样。Mike Morse 和 Jeffrey Glassman 锚定大型人身伤害客户引用;James Scott Farrin 增加了一个规模化 Mid-Atlantic 原告平台,且具备工伤赔偿和 SSDI 深度;Hershey Law 和 Frontier Law Center 则说明 Eve 可以进入原告侧劳动就业业务。地域上,可见引用以美国为中心但覆盖较广:Michigan、North and South Carolina、California 和 Massachusetts 都出现在公开材料中。这足以说明产品可以跨原告细分市场和州级工作流迁移,但不足以证明全国客户组合均衡或国际触达强。最佳解读是,Eve 已从单一律所或单一执业领域 proof point 走向多 logo 原告样本,但仍依赖精心挑选的公开故事,而不是完整披露的客户名册。[CU001, CU002, CU007, CU012, CU017, CU022]
| 细分 | 买方 / 推动者 | 核心用户 | 主要工作流 | 可见证据 | 扩张解读 |
|---|---|---|---|---|---|
| 高量个人伤害平台 | 所有人 / 管理合伙人 / 运营负责人 | 律师、律师助理、接案、案件经理 | 接案、医疗审阅、索赔函、证据开示、和解准备 | Mike Morse;Jeffrey Glassman | 目前最能证明产能和起草 ROI |
| 规模化原告地区律所 | 业务负责人 + 法律科技 | 部门负责人、律师、运营 | 医疗摘要、证词准备、证据开示、案件周期压缩 | James Scott Farrin | 多团队铺开和政策工作可接受时匹配度高 |
| 原告侧雇佣法精品所 | 创始人 / 管理合伙人 | 运营、接案、诉讼律师、庭审团队 | 索赔评估、文件请求、索赔函准备、庭审支持 | 客户:Hershey Law、Frontier Law Center | 显示 Eve 能从汽车个人伤害扩展到雇佣法工作流 |
| 全律所领导层 / 商业智能买方 | 管理合伙人 / COO / 业务负责人 | 领导层和运营分析师 | 收入节奏、和解历史、律师业绩、案件分布 | EveOS / Analyst 定位;具名律所故事 | 核心案件工作嵌入后,可能成为向上销售点 |
| 寻求工作流提升的小型律所 | 创始人或主办律师 | 律师 + 精简支持人员 | 索赔函、时间线、接案分流 | 软件评测页面和推荐语 | 可能适配,但公开打法看起来仍偏咨询式,不像纯自助的个人律师需求 |
细分边界来自具名客户故事和官方定位;公开证据在美国原告业务最强,在个人律师或非原告细分更弱。
[CU001, CU002, CU007, CU017, CU022, CU025]| 客户 | 细分市场 | 部署 / 使用场景 | 生产使用还是试点 | 公开成效 | 局限 |
|---|---|---|---|---|---|
| Mike Morse Law Firm | 密歇根大型人身伤害律所 | 医疗记录、索赔函起草、理赔员电话、由行动手册驱动的律师工作 | 生产使用 / 日常使用 | 律师称每天使用 Eve 约 75 次;产能翻倍或增至三倍;节省工作流耗时 | 案例研究由公司托管,未披露合同条款或基线控制 |
| Law Offices of James Scott Farrin 律所 | 北卡 / 南卡规模化原告平台 | 医疗摘要、证言准备、证据开示、证物选择、Jove 集成路径 | 生产使用 / 分阶段全所铺开 | 覆盖 300 人;医疗摘要从数周压到数分钟;一名律师将案件周期从 7-8 个月缩短到 4 个月 | 仍是轶事且限定部门,不是全所续约指标 |
| Hershey Law 律所 | 加州原告侧雇佣法精品所 | 索赔评估、缺失文件检查、实时庭审准备、团队行动手册 | 生产使用 / 全所基础设施 | 律所称每个团队都使用 Eve;提到在 $27.5M 判决案中提供实时庭审支持 | 公开证据无法隔离 Eve 对判决结果的因果贡献 |
| Jeffrey Glassman Injury Lawyers | 波士顿人身伤害与群体侵权业务 | 索赔函生成、医疗时间线、证言摘要、MRI 对比 | 生产使用 / 快速扩张 | 约 3 个月内 90% 的索赔函经由 Eve;时间线制作缩短到约 20 分钟 | 未披露留存率或席位数;产出占比为自报 |
| Frontier Law Center | 原告侧雇佣法律所 | Jenny 接案代理、询问书起草、案件摘要 | 生产使用 / AI 原生运营模式 | 线索转化率从 10% 到 35%;接案少 50 分钟;平均案件价值 +90%;同样时间内完成 5 倍工作 | 指标来自营销式客户证明,未经独立审计 |
这只是公开具名部署样本,不是完整客户普查。成效陈述由 Eve 或被展示客户自报,应当作为参考证据,而不是投资组合平均结果。
[CU012, CU015, CU017, CU020, CU022, CU023]按具名客户拆分公开证据质量,把原始 logo 价值与结果具体度、留存可见度分开看。
矩阵标签反映公开证据质量,不代表客户价值。每一行都有具名证据,但没有任何客户公开披露 NRR、GRR 或续约情况。
[CU012, CU017, CU022, CU025, CU028, CU029]6.2 采用轨迹与公开披露 ROI 信号
Eve 现在有足够多带日期的公开里程碑,可以显示真实采用势头,尽管准确实时总数仍会随每次公告移动。LawNext 报道 2026 年 1 月超过 500 家原告律所;Above the Law 3 月引用 800-plus;Eve 及其 4 月发布称 1,000-plus 家律所和超过 200,000 个案件;6 月材料把公开区间推到 Eve 自家网站的 1,200-plus,LawNext 对 EveOS 的报道则高达 1,400。这个进展方向上强劲,尽管 6 月分母在不同来源之间并未完全协调。ROI 证据也强于泛泛「省时间」承诺,但仍是案例研究级别,而不是经审计 cohort 数据。Mike Morse 称产能翻倍或翻两倍;James Scott Farrin 称医疗摘要从数周降到数分钟,一名律师的案件周期从七到八个月降到四个月;Frontier 称线索转化率从 10% 升至 35%,获客接待时间减少 50 分钟,平均案件价值上升 90%;Jeffrey Glassman 称约三个月内 90% 索赔函都通过 Eve 运行。这些是有意义的运营指标,因为它们触及转化、周期、产能和案件经济性,但它们由特色客户自报,应被视为可能性证明,而不是组合平均表现。[CU003, CU004, CU005, CU006, CU015, CU020]
| 日期 / 来源 | 公开指标 | 数值 | 置信度 | 含义 | 缺失分母 |
|---|---|---|---|---|---|
| 2026 年 1 月 / LawNext | 服务的原告律所 | 500+ | 中 | 2026 年初装机基础已具备意义 | 未按付费、试点或活跃使用拆分 |
| 2026 年 3 月 / Above the Law | 使用 Eve 的原告律所 | 800+ | 中 | 暗示环比扩张很快 | 赞助式报道;没有队列细节 |
| 2026 年 4 月 7 日 / Eve + PRNewswire + TMCNet | 原告律所客户 | 1,000+ | 高 | 清楚证明其在原告细分具备商业规模 | 没有每账户收入或留存数据 |
| 2026 年 4 月 7 日 / Eve + PRNewswire + TMCNet | 平台案件数 | 200,000+ | 高 | 显示有意义的工作流渗透,而不只是 logo 数 | 未披露活跃案件定义 |
| Jun 11 2026 / Eve 首页 | 受信任律所 | 1,200+ | 中 | 公司仍在强调快速增长 | 首页时间戳和统计方法未说明 |
| Jun 11 2026 / LawNext EveOS | 平台上的原告律师事务所 | 1,400+ | 中 | 独立报道显示,增长延续到 6 月 | 与首页总数冲突;准确实时数量仍未厘清 |
这是一张带日期的公开主张表,不是经审计的台账。它有助于判断轨迹,但 2026 年 6 月客户总数在不同来源之间不一致,应按区间看待,而不是当作单一权威数字。
[CU003, CU004, CU005, CU006]Eve 的公开销售动作,如何从营销证据推进到原告律所内部实施和扩展。
[CU003, CU009, CU010, CU018, CU021, CU033]6.3 买方、用户、付款方与上手动作
从客户故事看,用户地图如今相当清楚。Eve 被获客接待团队、律师助理、律师、业务负责人、运营人员和法律科技负责人使用,因为它位于案件工作流之中,而不是外围。买方和付款方地图不那么明确,但仍可推断:公开部署由所有者、管理合伙人、律所运营负责人和法律技术负责人推动,而不是某个律师刷卡购买。Eve 自家网站也强化了这一模式,因为商业动作是预约沟通、产品 walkthrough 和候补名单驱动,而不是公布自助计划。实施看起来也偏咨询式。James Scott Farrin 建立实施团队、验证 SOC2 和 HIPAA 姿态,并签署 Acceptable Use Policy;Mike Morse 通过内部演示处理恐惧;Hershey 按阶段建立 playbook;Frontier 则围绕现场展示真实用例来推动采用。反复出现的模式是,先切入一个急性瓶颈——获客接待、医疗时间线、索赔函起草或审判准备——再扩展到相邻工作流,最终进入律所层面情报。这是合理的原告律所销售动作,但也意味着上手劳动密集、变革管理重,今天可能更有利于区域性和规模化律所,而不是纯 solo 自助服务。[CU009, CU010, CU013, CU018, CU019, CU021]
| 信号 | 公开数值 | 细分 / 来源 | 置信度 | 指向什么 | 尽调问题 |
|---|---|---|---|---|---|
| NRR / GRR / 流失 | 无公开披露 | 低 | 公开资料没有队列耐久性证明 | 按客户细分索取队列留存、续约和流失数据 | |
| 正式合同期限 / 续约条款 | 无公开披露 | 低 | 无法从合同结构判断粘性 | 索取标准 MSA 期限、续约节奏和扩张机制 | |
| 运营依赖轶事 | “如果有人把 Eve 拿走,我会辞职” | Mike Morse 案例研究 | 中 | 至少在一家大型 PI 律所内部,用户依赖度很强 | 用席位使用数据和续约证据验证 |
| 工作流占比 | 约 3 个月内 90% 的索赔函 | Jeffrey Glassman 案例研究 | 中 | 瓶颈清楚后,重复使用上量很快 | 索取前 3-6 个月之后的持续使用曲线 |
| 评价情绪 | 12 条已验证评价给出 4.9 / 5 | Software Finder | 中低 | 存在正向用户情绪和支持感知 | 用更大的评价池或客户推荐交叉验证 |
| 全所嵌入 | 每个团队都使用 Eve | Hershey Law 案例研究 | 中 | 显示广泛内部采用,不只是单个内部推动者 | 按角色索取活跃用户数和每周 / 每月使用深度 |
null 单元格表示公开留存数据缺失,不代表表现为零。现有信号是轶事、评价片段或嵌入式使用故事,不是正式队列分析。
[CU014, CU023, CU026, CU038, CU039]| 阶段 | 谁主导 | 公开证据 | 为什么重要 | 观察到的摩擦 |
|---|---|---|---|---|
| 发现 / 申请演示 | 创始人、管理合伙人、运营或法律科技负责人 | 公开动作以预约电话、产品导览、演示和报价引导页面为主 | 商业入口是顾问式销售,不是自助购买 | 早期就有定价不透明和厂商依赖 |
| 问题映射 | 实施团队 + Eve | Farrin 访谈各部门;Frontier 聚焦真实用例;Mike Morse 牵头现场演示 | 销售似乎先锚定最糟糕的工作流瓶颈 | 需要内部时间和推动者带宽 |
| 安全 / 政策审查 | 运营 / 法律科技 / 律所管理层 | Farrin 核验 SOC2 和 HIPAA;NYC 与 NYSBA 指引强调保密和监督 | 规模化部署前必须过这一关 | 治理审查会拉长周期 |
| 内部推动者试点 | 怀疑者或高频用户 | Jeffrey Glassman 让持怀疑态度的律师先用;Mike Morse 和 Hershey 打造定制行动手册 | 受尊敬的用户转化后,采用更容易粘住 | 变革阻力和岗位焦虑是真实阻碍 |
| 部门铺开 | 运营 + 部门负责人 | Hershey 按团队分阶段铺开;Farrin 扩展到一家 300 人律所 | 把成功用例变成嵌入式工作流 | 培训和支持负担会快速上升 |
| 扩张 / 反馈循环 | 管理层 + Eve 团队 + 开发者 | Farrin 每周会议、Jove 集成、CMS 补充定位 | 先落地再扩张依赖定制化和集成质量 | 客户成功负担可能限制小账户 |
这张表综合了具名部署和官方 CTA 中反复出现的公开模式。它反映的是 Eve 看起来如何销售和落地,而不是已披露的内部销售行动手册。
[CU009, CU010, CU018, CU019, CU021, CU032]公开证据显示,原告律所从发现 Eve 到扩展工作流,会经历一条路径;图中标出参与者,以及推广可能卡住的环节。
各阶段由公开客户故事和官方行动号召综合而来。顺序有证据支撑,但确切转化率和耗时并未公开披露。
[CU009, CU018, CU019, CU021, CU031, CU032]6.4 扩张空间、持久性边界与商业约束
Eve 的扩张机会真实存在,因为它的切入点不是单一文档任务。官方材料和客户案例显示,Eve 可以从线索受理延伸到诉前、诉讼支持和律所情报;参考客户也横跨高量级 PI 律所和原告侧雇佣法律所。与现有 CMS 共存也扩大了可触达买方,因为它降低了推倒重来的压力;Eve 现在把自己描述为案件管理系统的补充,公开引用也提到 CRM 同步和即将接入 Jove。但耐久性证据仍明显薄于采用证据。公开来源没有披露 NRR、GRR、流失率、合同期限或头部客户集中度,可见评论样本也太小,不能替代续约证据。定价不透明也是拖累:买方仍需演示和报价,外部评测也指出,集成、迁移、上线和培训成本都是真实预算变量。最后,法律 AI 治理仍是门槛。纽约州律协指引和制裁评论都强调,任何规模化部署仍需要保密控制、监督、核验和费用敏感度。也就是说,Eve 最好的扩张飞轮,是在成熟原告律所里以咨询式打法先落地再扩张——但这套打法也比轻量席位销售更慢、更贵,也更受治理约束。[CU011, CU024, CU027, CU033, CU037, CU038]
| 扩张驱动 | 约束 / 集中度风险 | 影响 | 当前公开判断 | 尽调路径 |
|---|---|---|---|---|
| 从接案扩到诉前、诉讼和分析 | 工作流变宽会抬高实施范围 | 可提高 ACV 和粘性,但上线会更慢 | 官方产品地图和客户故事支持这一点 | 索取按模块划分的附加购买率和实施周期 |
| 横向进入原告侧各业务领域 | 每个专业领域都需要调优行动手册和数据模型 | TAM 不再只靠车祸 PI 支撑 | 雇佣、工伤赔偿、SSDI、医疗事故等引用支持这一点 | 索取按业务领域划分的客户数,以及精选故事之外的证明 |
| 与现有 CMS / CRM 共存 | 集成深度可能成为瓶颈 | 降低替换旧系统阻力,但可能限制对记录系统的控制 | CMS 补充定位和 Jove 集成引用支持这一点 | 索取实际运行的集成、数据回写范围和失败率 |
| 从一个内部推动者团队铺到全所 | 变革管理和政策工作可能拖慢扩张 | 这是超越试点成功、实现规模化的关键 | Mike Morse、Farrin 和 Hershey 案例强力支持 | 索取实施行动手册、管理员负担和多团队铺开所需时间 |
| 用头部参考客户撬动销售 | 头部客户集中度仍未披露 | 有利于销售效率,但无法评估收入集中度 | 公开资料未解决 | 索取前 10 大客户收入占比和 logo 级 ARR 集中度 |
| 销售主导定价和服务 | 定价不透明会放大小律所采购摩擦 | 可能让产品更适合更大或需求更急的买方 | 评价页和无公开定价支持这一点 | 索取价格表、套餐定义,以及服务与软件收入构成 |
扩张路径有证据支撑;集中度字段仍为空,因为没有公开来源披露头部客户收入占比、渠道构成或续约队列。
[CU010, CU011, CU033, CU037, CU040, CU042]6.5 图表
07风险
7.1 监管、伦理与无证执业风险
Eve 并没有把自己包装成后台记笔记工具。其首页和平台页面称,产品处理线索受理、医疗时间线、动议、索赔函、证据开示回复、研究和律所分析;2026 年发布材料还描述了会在新信息到来时自动行动的代理,能够起草所需文件、管理常规受理和状态更新,并由律师在工作离开律所前审查批准。这个定位商业上很有力,但也把 Eve 推近了律协指引如今视为法律 AI 核心风险的边界:软件不再只是起草辅助,而是在法律服务交付里像准代理人一样运转。Eve 自己的信息试图让人类保留控制权,但它的差异化恰恰取决于继续深入监管机构和法院视为专业判断领域的工作流。 有约束力的基线仍很传统。ABA Formal Opinion 512 规定,律师使用生成式 AI 时,仍负有能力、保密、客户沟通和合理计费义务。Model Rule 1.1 要求胜任代理,Rule 1.6 要求采取合理措施防止未经授权的披露或访问,Rule 5.5 禁止协助无证执业。更尖锐的 2026 年来源是加州更新后的实务指引,该指引明确谈到代理式 AI,并称律师不得部署系统,使其在缺乏有意义律师监督和审查的情况下作出实质性法律判断、传达法律建议、准备并提交诉状,或以其他方式以代表身份行动。这一点重要,因为 Eve 如今把产品描述为带有执行层、审查层和情报层的 AI 劳动力,而不是被动软件工具。 眼前风险不是 Eve 明天就会被认定为律所。真正的风险在于,产品越多自动化线索资格判断、法律起草、问题识别和下一步建议,各州 UPL、监督和披露问题就越会从抽象政策争论进入客户采购和理赔处理。Thomson Reuters Institute 的 2026 年 UPL 分析强调,各州对法律执业仍没有统一定义,敏感边界在于把法律适用于具体事实并建议行动路径。对 Eve 来说,这种模糊性双向作用:它保留了经营空间,但也意味着少数不利的州律协或法院案例,可能比缓慢推进的成文法更快冷却采用。在风险章节里,应把它评为高严重性、中等可能性问题,因为即使没有任何正式执法直接点名 Eve,下行影响也可能先通过客户犹豫、法院审视,或保险公司 / 律所政策收紧体现出来。[CR001, CR002, CR003, CR010, CR011, CR012]
| 风险 / 规则 / 触发点 | 司法辖区 | 状态 | 可能性 | 严重性 | 当前缓释 | 剩余暴露 | 尽调路径 |
|---|---|---|---|---|---|---|---|
| 智能体式起草和办案滑入受监管的法律执业 | 多州 / 由州律师协会监管 | 未发现 Eve 专项行动;2026 年加州指引收紧了对智能体式 AI 的监督预期 | 中 | 关键 | 律师审阅和批准关口;产品定位为增强律师,而不是替代律师 | 少数州律师协会投诉、保险公司限制或采购冻结,就可能在正式执法前压冷采用 | 审查客户委托函、人工审阅日志,以及关于 UPL 的任何州别外部律师备忘录 |
| Rules 1.1 和 5.5 下的律师胜任能力与监督失效 | 美国全部司法辖区 | 只要在客户工作中使用 AI,义务就持续存在 | 高 | 高 | 引文、来源链接、内部政策和客户培训材料 | 如果律所把 Eve 输出当成终稿,而不是待审草稿,制裁和渎职风险会很快进入客户关系 | 检查培训完成度、审批工作流,以及高风险输出的审计轨迹 |
| Rule 1.6 下的客户保密失效 | 美国全部司法辖区 | 敏感案件数据进入系统时,风险始终在线 | 中 | 高 | 对齐 SOC 2 的控制、最小权限、加密、客户数据所有权条款 | 特权 / 保密问题取决于具体工作流,仍可能因客户误用或集成范围蔓延而失效 | 索取架构图、DPA/BAA 文件包和案件级权限控制 |
| 法院审查 AI 生成提交文件和缺乏坦诚 | 联邦和州法院 | 活跃;2023-2026 年制裁记录持续扩大 | 中 | 关键 | 打开并核验引文、带来源链接的输出、发布前律师签署确认 | 只要一起公开制裁点名 Eve 支持的工作,就会在原告律师行业外溢声誉影响 | 索取内部事件登记、质量升级政策,以及任何与 AI 输出有关的客户诉讼保全 |
| 医疗记录工作流中的 HIPAA / 商业伙伴身份误分类 | 联邦 / HHS OCR | 从已审材料无法完全确定公开范围 | 低中 | 高 | Eve 公开声称符合 HIPAA,且不使用 PHI 训练 | 如果客户数据流实际上构成商业伙伴身份,泄露义务和合同暴露会显著变尖 | 核验 Eve 何时签署 BAA、哪些数据流触发 BAA,以及分包商是否承继义务 |
| 州律师协会或保险公司限制智能体式法律 AI 的负面指引 | 州律师协会 / 渎职险承保方 | 未识别到 Eve 专项公告 | 中 | 高 | 人在回路定位和客户披露指引 | 如果渎职险承保方或大型律所收窄获批用例,采用可能明显放缓 | 收集客户承保方政策、采购异议,以及头部账户获得的任何外部伦理意见 |
各行按剩余严重性排序,依据截至 2026-06-11 的公开法律指引和已披露产品范围。公开来源未显示任何 Eve 专项律师协会行动或保险公司限制。
[CR010, CR011, CR012, CR013, CR014, CR015]按发生概率和影响程度,审视 Eve 最主要的法律、隐私、平台和竞争风险。
单元格是基于公开法律指引、产品主张和不利判例作出的定性判断,并非已披露的事件概率。
[CR015, CR016, CR021, CR032, CR038, CR041]7.2 幻觉、证据与法律过失风险
近期判例打消了一个幻想:法院不会因为牵涉工具就原谅 AI 错误。Mata v. Avianca 仍是标志性警示:律师提交并随后辩护 ChatGPT 生成的虚假案例后,Castel 法官制裁了律师及其律所,并强调现有规则要求律师把关。2026 年第九巡回法院在 Lnu v. Blanche 中的命令把教义又推进一步,而且尤其重要,因为它来自上诉层级、时间很近、表述明确:法院称,违规不发生在研究或起草环节,而发生在签署和提交环节。律师提交含有不存在案例的简报后,又未能就 AI 是错误来源及时、坦诚披露,法院因此处以金钱制裁和六个月停业。贯穿其中的逻辑很简单:法律 AI 可以辅助,但不会稀释责任。 这些先例直接关系到 Eve,因为产品面向真实案件中的索赔函、动议、证据开示回复、质询书、时间线和研究销售。部分输出不是法院文件,但仍会造成证据和法律过失暴露。错误时间线会扭曲案件估值。幻觉或夸大的医疗事实会污染索赔函、和解谈判、调解简报或证词提纲。核验不足的证据开示回复会引发证据问题、制裁或信誉损害。Eve 显然知道这类风险存在:其平台称研究引文可打开并核验,医疗概览称每个数据点都链接到页级来源材料,证据开示工作流称律师审查必不可少,同时应从选定输入中排除特权或工作成果材料。这些控制很好,但也等于承认准确性和过度披露都是现实失效模式。 2025 年 Morgan & Morgan 制裁令进一步说明,内部法律 AI 不是安全港。该案中,律所自有平台生成了虚假案例,法官强调核验义务不可委托。这正是审查 Eve 时应采用的框架。关键问题不是 Eve 是否比消费级聊天机器人用了更好的检索、模板或引用,而是在截止期限压力下,原告律所能否可靠地以足够人工核验运行产品,在输出触达保险公司、对方律师、调解员或法院之前,拦住事实漂移、引用幻觉和过度主张。公开材料没有披露 Eve 自身的独立错误率、索赔经验或制裁历史。没有这些指标,投资人应假定输出完整性风险会随着客户工作量和产品自主性放大,而不只是随模型质量变化。[CR017, CR018, CR019, CR020, CR021, CR022]
| 失效模式 | 可能性 | 严重性 | 缓释成熟度 | 剩余暴露 | 尽调问题 |
|---|---|---|---|---|---|
| 幻觉引文或法律依据进入法院提交文件 | 中 | 关键 | 来源核验 UX;律师审阅;研究产品承诺打开并核验引文 | 近期法院施加罚款、停业和坦诚义务制裁;签署责任无法外包 | 要求提供提交文件工作流样本、引文 QA 指标和异常日志 |
| 事实幻觉或错误时间线扭曲索赔、调解或估值 | 中 | 高 | 到源文件的页面级链接;夜间审计发现路由给人工 | 问题进入法院前,和解姿态就可能被扭曲 | 审查时间线、损害提取和来源不匹配的错误率仪表盘 |
| 证据开示回复过度披露特权或工作成果材料 | 中 | 高 | 工作流提示用户排除特权文件并核验每句话 | 产品指引有帮助,但一次文件误选仍可能放弃保护或引发动议程序 | 在客户环境中测试特权筛查控制和文件选择默认项 |
| 智能体式审计 / 建议让超负荷员工形成隐性依赖 | 中高 | 高 | 律师签署确认;客户治理政策;AI 使用培训 | 量上来后,团队可能不再复查底层证据,就信任队列和风险分数 | 检查审阅人负载、升级阈值,以及低置信度输出是被阻断还是只打标签 |
| 缺少公开质量基准掩盖尾部风险频率 | 高 | 中高 | 客户证言和带来源链接的产品设计是正向但间接的证据 | 已审材料中未发现公开独立准确率、制裁或理赔历史数据集 | 索取队列级 QA 指标、客户投诉日志,以及保险公司或 E&O 通知 |
这些是工作流风险,不是对已知缺陷的指控。公开证据支持缓释功能存在,但不能证明它们在规模化下的实测效果。
[CR017, CR018, CR019, CR020, CR021, CR022]Eve 的主要风险向采用率、收入耐久性和估值传导的方式。
[CR016, CR021, CR022, CR032, CR043, CR045]7.3 隐私、HIPAA、特权与平台依赖
Eve 的公开材料反复强调,它为敏感的原告法律数据而建。首页称客户数据私密、隔离并受特权保护;安全页面称公司遵循 SOC 2 标准,进行年度渗透测试,使用最小权限访问控制,开展季度访问审查,并在 AWS 和 Azure 上托管服务,AWS 数据库位于美国;医疗概览页面称 Eve 符合 HIPAA,医疗记录在传输和静态存储中均加密,受保护健康信息不会用于训练 AI 模型。这些都是有意义的信号,企业条款也通过要求安全保障、持有 E&O 和网络保险、界定客户数据所有权,增加了更正式的结构。但同一批文件也暴露出隐私和平台风险为何仍然重要。Eve 获得一项许可,可在提供服务所需范围内托管、复制、传输和展示客户数据;它可因安全风险或法院命令暂停服务;它不保证服务不中断或无错误;并明确称分包处理方可能停止托管,且 Eve 不对这些中止承担责任。 HIPAA 风险需要谨慎表述。HHS 指引和 45 CFR 164.410 明确列出 covered entities 和 business associates 的违规通知义务,包括 business associate 需在无不合理延误且 60 天内通知 covered entity。HHS 还把 business associates 定义为代表 covered entity 或另一 business associate 创建、接收、维护或传输 PHI 的供应商。公开来源无法证明 Eve 是否在每个客户工作流中都作为 HIPAA business associate 行事,因为原告律所往往通过授权获取医疗记录,并非每个事项都会形成直接的 covered-entity/vendor 链条。因此,精确的 HIPAA 角色无法从公开来源完全证明。不过,Eve 显然在规模化处理敏感医疗记录,这意味着即便适用理论是合同保密、州隐私法或法律过失,而非仅由 HIPAA 执法驱动,数据泄露、模型误路由或访问控制失败也会造成严重暴露。 特权风险同样有细微差别。该领域最强的 2026 年公开案例是 Warner v. Gilbarco,密歇根联邦法院认定 AI 辅助诉讼材料保留工作成果保护,使用 ChatGPT 不会自动放弃保护。这有利于多头论点。但它不是一揽子特权背书,也不会把产品营销变成法律教义。法院处理的是具体诉讼事实,其他 2026 年评论和案卷活动也显示,公开或第三方 AI 使用如何影响特权仍存在真实分歧。对 Eve 客户来说,实际教训是:特权是一种工作流属性,不是软件功能;它取决于数据选择纪律、合同控制、平台保密性和律师监督。再加上 SLA 把云服务商和集成应用故障排除在停机计算之外,运营结论很尖锐:如果 AWS、Azure 或关键 CMS 集成在截止期限附近失效,原告律所仍会暴露在风险中,即便 Eve 自身技术上没有做错任何事。[CR004, CR005, CR006, CR007, CR008, CR009]
| 依赖 / 失效模式 | 可能性 | 严重性 | 当前缓释 | 剩余暴露 | 尽调问题 |
|---|---|---|---|---|---|
| 医疗记录或 PHI 泄露 | 中 | 关键 | 安全计划、加密、年度渗透测试、事件响应、网络保险 | 公开来源未显示详细 BAA 范围、分包商清单或泄露演练结果 | 索取安全文件包、BAA 模板、事件行动手册和近期审计结果 |
| 尽管营销称“特权保护”,特权仍受到挑战 | 中 | 高 | 案件隔离、保密条款、律师审阅、受控输入 | 特权由工作流、法庭和保密事实决定,不由产品标签决定 | 审查客户特定配置下的日志、留存和数据使用企业控制 |
| 云宕机或集成故障让临近截止日期的案件工作停摆 | 中 | 高 | 99.5% SLA、备份、监控 | SLA 排除云服务商和集成应用宕机;服务抵扣无法补救错过截止日期 | 索取故障切换设计、RTO/RPO、离线导出路径和截止日期应急流程 |
| 分处理方停服或模型 / 供应商政策变化 | 中 | 高 | 供应商审查计划;AWS/Azure 引用暗示多系统架构 | 条款称分处理方可停止托管,Eve 不对这些停止承担责任 | 获取关键供应商名单、可移植性条款和迁移流程 |
| 公网站隐私实践和分析工具造成信息不一致 | 低中 | 中 | 存在退出选项,且营销网站收集可与案件处理分离 | 隐私政策提到分析和广告合作伙伴;如果未与案件数据控制清晰切开,可能让买方警惕 | 确认营销遥测、产品遥测和客户案件环境之间的隔离 |
平台和隐私各行把合同、技术和法律暴露合在一起,因为原告律师工作流在实践中会把这些风险绑在一起。
[CR004, CR005, CR006, CR007, CR008, CR009]横跨云、集成、客户和既有市场结构的关键上游依赖。
[CR003, CR007, CR008, CR039, CR040, CR042]7.4 买方集中、竞争与执行风险
Eve 的增长动能是真实的,但风险轮廓取决于这些动能落在哪里。2026 年 1 月,公司称拥有 500 多家领先原告律所,收入增长 10 倍;到 6 月,公司称更广泛平台已在 900 多家原告律所投产;首页现在称 1,200+ 家律所。这些都是强采用信号,但公开客户证据仍集中在 Mike Morse、James Scott Farrin 等相对可见的旗舰原告律所。这不能证明存在危险的收入集中,但确实说明参考价值、工作流学习和声誉杠杆集中在少数大型成熟客户手中。公开材料没有披露流失率、NRR、平均合同价值、席位集中度,或最大 logo 贡献的使用份额。这一缺口很重要,因为如果少数高量级律所暂停、切换或缩小范围,Eve 可能同时失去收入和推动获客的可信度。 Eve 想建的东西本身放大了执行风险。狭义法律 copilot 即便令人失望,也不一定击穿业务。一个承诺线索受理、分析、起草、夜间审计和跨案件情报的「AI 操作系统」,必须同时同步产品质量、数据摄取、集成、客户变更管理和法律风险控制。Eve 自己的案例研究提到,需要 Chief AI Officers、可接受使用政策、治理审查和审慎的变更管理,来克服员工恐惧和工作流扰动。从一个角度看,这是利好,因为它说明严肃客户正在深度嵌入产品;从另一个角度看,这是利空,因为部署成功取决于客户运营成熟度,而不只是产品需求。如果客户需要专门 AI 领导者、定制 playbook 和大量实施支持,毛利率杠杆和续约耐久性可能低于纯自助 SaaS 叙事所暗示的水平。 竞争风险也不再是假设。Thomson Reuters 和 RELX 没有等着验证法律 AI 是否有效;它们已经在用 AI 采用推动核心法律业务增长。Thomson Reuters Institute 的市场数据显示,法律 GenAI 采用在上升,而政策和培训滞后;RELX 的 2025 Form 20-F 称,其 Law Firms & Corporate Legal 分部在 Lexis+ AI 和 Protégé 带动下实现两位数增长。同一份文件还警告,法律市场高度竞争且动态变化,AI 和新竞争者可能快速改变需求,隐私、网络安全、云、软件和大语言模型依赖也都可能伤害业绩。这些表述说的是 RELX,但同样说明 Eve 的护城河为何脆弱:既有巨头拥有经验证的法律内容、深植的分销,以及吸收产品错误的资本。只要原告侧专业化和工作流深度仍明显领先既有巨头,Eve 仍能赢;但现在证明责任已经落在执行、治理和可衡量客户 ROI 上,而不再只是功能新鲜感。[CR034, CR035, CR036, CR038, CR039, CR040]
| 风险 | 发生概率 | 严重度 | 现有缓释 | 剩余暴露 | 尽调要求 |
|---|---|---|---|---|---|
| 大型标杆律所客户集中 | 中 | 高 | 律所数量增长、logo 覆盖更广,表面上降低了单一 logo 依赖 | 公开材料未披露收入集中度、NRR 或分 cohort 流失 | 要求提供前 10 大客户收入占比、席位集中度和 cohort 留存 |
| 客户变革管理摩擦拖慢扩张 | 高 | 中高 | 案例研究显示,律所有 AI 负责人、政策和定制 playbook 后可以跑通 | 如果采用需要大量赋能,规模效率和利润率杠杆可能低于预期 | 按客户规模衡量部署周期、支持强度和扩张速度 |
| 既有法律内容平台压缩功能差异 | 高 | 高 | 原告律所专精和工作流深度目前仍是差异点 | LexisNexis 和 Thomson Reuters 拥有已验证内容和分发,AI 增长也已在申报文件和市场报告中显现 | 跟踪对阵既有厂商的赢单 / 输单数据,以及买方愿意把内容与工作流拆开的证据 |
| 超高速增长跑在治理和 QA 控制前面 | 中 | 高 | Eve 公开强调通过 Auditor 建立复核闭环和质量监控 | 收入增长 10 倍、客户快速扩张,会提高实施质量不均或潜在控制失灵的概率 | 审查治理人员、QA 配置,以及每个已部署客户的事故频率 |
| 客户采购围绕披露、政策和保险方要求趋严 | 中 | 中高 | Eve 发布负责任 AI 和客户披露指引,可帮助客户更安全地采购 | GenAI 采用面在扩大,但法律买方的政策和培训仍滞后 | 检查销售周期是否拉长、AI 条款被修改的情况,以及保险方或律协指引驱动采购时的赢率 |
公开材料提供了很强的标杆客户证据,但单位经济和留存披露稀少,因此集中度和执行风险按保守口径评估。
[CR034, CR035, CR036, CR038, CR039, CR040]7.5 缓释措施、监控与止损标准
Eve 并没有忽视这些风险。公开可见的缓释因素包括:发布材料中的律师审查关卡、带来源链接的医疗和研究输出、映射到 SOC 2 的安全计划、年度渗透测试、最小权限控制、明确的客户数据所有权语言、E&O 和网络保险,以及一项虽有限但至少建立合同基线的 uptime 承诺。公司似乎也理解安全采用的组织侧问题:其博客内容鼓励律所制定负责任 AI 政策、客户披露、培训要求,以及针对畸形输出或泄露的升级流程。这些都是合规级法律 AI 供应商需要的正确要素。 问题在于,大多数可见缓释因素都是流程主张,而非独立衡量的结果。本文审阅的公开来源没有显示外部引用准确性审计、公开事故历史、已披露制裁记录、客户留存数据,或 Eve 自主工作流在多个司法辖区接受对抗性或监管审查测试的证据。这意味着投资人应把管理成熟度视为有希望但未被证明,尚不足以承保一个嵌入真实诉讼与和解工作流的平台。下一阶段尽调应聚焦硬证据:样本 QA 仪表盘、红队结果、泄露响应演练、保险公司历史、模型治理政策、特权数据架构,以及客户层面的续约 / cohort 数据。 对承保来说,最干净的 thesis-break 触发器也可监控。如果公开制裁、律协投诉或不利法院命令把 Eve 输出与律师不当行为联系起来,就是重大红旗,因为它会在 Eve 试图成为系统工作底座的关键点攻击信任。涉及医疗记录或特权案件数据的重大安全事件,也会产生同样不成比例的后果。旗舰律所显著流失,或证据显示大型既有公司能在更便宜地捆绑可信法律内容的同时,匹配原告侧工作流宽度,也会触发同类风险。在这些事件缺席时,Eve 仍有通往品类领导者的合理路径。但投资论点应以持续的人类在环纪律、低事故率、耐久参考客户,以及治理扩张速度跟上自主产品范围为条件。[CR004, CR008, CR015, CR018, CR019, CR024]
| 风险 | 可监测触发项 | 阈值 / 事件 | 行动含义 |
|---|---|---|---|
| UPL / 监督风险 | 州律协或保险方认为,在没有更严格人工复核时,agentic 起草 / intake 超出获准用途 | 主要原告律所州、保险方或前 20 大客户发布任何限制 | 暂停扩展自主功能,直到复核控制和准许使用边界重新验证 |
| 幻觉 / 制裁风险 | 法院命令、制裁或公开更正文件与 Eve 支持的输出有关 | 一起已确认、点名 Eve 输出的公开事件,或客户险情呈现模式 | 升级为红色尽调;要求事故根因、整改证明和客户侧封堵 |
| 隐私 / HIPAA / 特权风险 | 涉及医疗或特权案件数据的安全事件或保密争议 | 重大泄露、监管问询,或不利的特权裁定,且原因指向平台设计而非孤立误用 | 围绕法律责任准备金、客户流失和 go-to-market 放缓重新定价 |
| 平台依赖风险 | 云 / CMS 中断阻塞有时限要求的工作流 | 反复宕机,或一次影响关键期限的宕机且没有可信恢复方案 | 在提高敞口前,要求完成故障切换整改、离线工作流和更强合同保护 |
| 买方集中风险 | 标杆 logo 流失或使用量急剧收缩 | 失去顶级标杆客户,或有证据显示大型律所在试点后缩小范围 | 重新承销留存耐久性和销售效率假设 |
| 竞争 / 执行风险 | 既有厂商把已验证法律内容与可比的原告工作流打包 | 对 Lexis/Westlaw 关联工具的赢率持续恶化,或 CAC 膨胀且扩张收入无法抵消 | 除非 Eve 证明原告原生 ROI 可持续,否则估值立场转向执行折价 |
终止标准不是机械交易信号,而是投资人复核的升级触发项;每一项都可能比单看收入指标更早击穿信任。
[CR004, CR008, CR015, CR020, CR021, CR032]7.6 图表
08估值
8.1 融资背景,以及 2025 年 9 月轮次实际披露了什么
公开记录中最干净的披露事实很简单:Eve 2025 年 9 月 Series B 融资 $103 million,投后估值超过 $1 billion。Eve 在自己的融资文章和发布的新闻稿中重复了这个数字,LawNext 和 Artificial Lawyer 等独立法律科技报道也呼应了同一标题。但对估值更重要的,不只是融资规模,而是这个价格落在什么规模点上。Series B 公告前后,Eve 公开称其八个月内新增 350 多家律所,客户律所超过 450 家,每年处理 200,000 多个法律案件,并帮助律所追回超过 $3.5 billion 的和解金与判决款。这些都是有意义的增长标记,但不是收入披露。 与 Eve 自己 2025 年 1 月 Series A 的对比很有用,因为它显示叙事加速得有多快。Series A 时,公司宣布 $47 million 融资,并称 100 多家律所已采用产品,同时声称收入同比增长 500%。到 2026 年 6 月运行日,Eve 首页已经换成更大的 1,200+ 家律所标题,并把叙事扩展为「原告法律 AI 操作系统」,覆盖线索受理、诉前、诉讼、AI-ready 数据和律所级情报。这暗示轮后增长有意义,但也引入了尽调问题:公开记录从未解释 450 家律所和 1,200+ 家律所是否按同一定义计算,是否全是付费客户,或其下的平均合同价值是多少。 缺失的收入桥就是核心估值问题。对一家快速成为原告律所记录系统和行动层的公司来说,$1B+ 投后标记可以自洽。但公开证据仍未披露 ARR、NRR、毛利率或 Series B 背后的优先权结构。因此,这一轮应被理解为投资人对 Eve 品类潜力的强信号,而不是一个已经被公开市场式收入证据验证的价格。[CV001, CV002, CV003, CV004, CV005, CV006]
8.2 可比公司、市场规模支持,以及公开基准说明了什么
最有用的法律科技基准是 Clio,不是因为它与 Eve 完全可比,而是因为它是少数在实质规模上公开披露估值和 ARR 的私有法律软件平台之一。Clio 2024 年 7 月融资将公司定价为 $3 billion,当时 ARR 已高于 $200 million,隐含 ARR 倍数低于 15x。收购 vLex 并完成 2025 年 11 月融资后,Clio 估值达到 $5 billion,ARR 为 $400 million,隐含约 12.5x ARR。这些标记仍带有优质法律科技定位,但它们落在 Eve 尚未公开披露的收入水平上。相比之下,Harvey 明确是离群值:CNBC 报道其 2026 年 3 月 ARR 为 $190 million,估值 $11 billion,约 58x ARR。这让 Harvey 只能用来提醒市场存在前沿 AI 稀缺溢价,而不能作为原告法律工作流软件的直接承保锚。 私有原告侧和工作流同业提供方向性支持,但不是干净的倍数锚。EvenUp 2025 年 10 月 Series E 将其定价为 $2B+,管理层称 ARR 同比翻倍,平台上有 2,000+ 家律所。Supio 披露累计融资 $91 million,且 Series A 以来 ARR 增长 4 倍,但没有披露绝对 ARR 基数。Filevine 披露近 6,000 名客户、100,000 名法律专业人士、96% 总留存和高于 120% 的 NDR,但没有披露估值。Litify 披露 450+ 家企业客户和 70,000+ 名法律专业人士,此前完成 $50 million Series A,但同样没有披露当前 ARR。 公开市场框架更严格。Intapp 是一家上市专业软件平台,具有法律敞口,2026 年 3 月季度收入 $146.0 million,2026 年 6 月市值约 $1.78 billion,约为年化收入 3.0x。SaaS Capital 的公开基准显示,2025 年 6 月 SaaS ARR 倍数中位数为 6.7x,并把当前环境描述为 6-8x 的「新常态」。市场规模支撑足以证明 Eve 相比通用 SaaS 落后者应享有溢价:ABA 给出的美国律师基数为 1.37 million,Clio 的人身伤害统计页显示 PI 子集超过 135,000 名律师,每年近 400,000 起索赔。但 Thomson Reuters 2026 年法律市场报告提醒,买方是在不稳定经济基础上投资 AI,这更支持价格纪律,而不是自动 AI 倍数膨胀。[CV012, CV013, CV014, CV015, CV016, CV017]
| 可比对象 | 指标 | 倍数 / 估值 / 状态 | 相关性 | 局限 |
|---|---|---|---|---|
| Clio(2024 Series F) | 估值与已披露 ARR | > $200M ARR 对应 $3B 估值(<15x) | vLex 抬升前,已披露私有法律软件轮次中最好的锚点 | 广义法律工作流平台,不是原告原生 AI |
| Clio(2025 Series G + vLex,融资轮) | 估值与已披露 ARR | $400M ARR 对应 $5B 估值(约 12.5x) | 展示大 ARR 规模下高端法律平台定价的样子 | 包含 vLex 交易背景和债务融资安排 |
| Intapp(公开市场,2026 年 6 月) | 市值与年化收入 | 约 $584M 年化收入对应 $1.78B 市值(约 3.0x) | 有法律敞口的受监管专业软件公开可比公司 | 非原告律所特定,且公开市场通常更严苛 |
| Harvey(2026 年 3 月) | 估值与已披露 ARR | $190M ARR 对应 $11B(约 58x) | 只能作为明确标注的 AI 异常值上限,用于前沿稀缺定价 | 通用法律 / 专业服务 AI,不是原告工作流 |
| EvenUp(2025 年 10 月) | 私有轮次 / 规模标记 | $2B+ 估值;2,000+ 家律所;ARR 同比翻倍 | 最接近的已披露原告 AI 估值可比对象,并有有意义的规模标记 | 未披露绝对 ARR |
| Supio(2025 年 4 月) | 私有轮次 / 增长标记 | 总融资 $91M;Series A 以来 ARR 增长 4x | 相关的原告 AI 同行,定位偏重验证 | 无公开估值或绝对 ARR |
| Filevine(2025 年 9 月) | 私有规模 / 留存标记 | $400M 融资;约 6,000 个客户;96% 总留存;120%+ NDR | 展示法律软件中既有工作流厂商的规模和留存质量 | 无公开估值倍数 |
| Litify(2023 + 当前规模) | 私有规模 / 所有权标记 | 450+ 企业客户;70,000+ 法律专业人士;多数股权售予 Bessemer | 显示原告和高案件量业务中的平台宽度与买方价值 | 无公开估值倍数,且融资披露已陈旧 |
表中混合了干净的倍数锚点和方向性的私有规模标记,因为多数私有法律科技同行披露的 ARR 不足以支撑同口径计算。
[CV012, CV013, CV014, CV015, CV020, CV021]支撑 $1B 估值所需的 ARR,会随所用倍数体系大幅变化。
数值只是用 $1,000M 除以参考倍数,意在给出估值门槛,不是经审计的预测。
[CV031, CV032, CV036, CV037, CV038, CV051]8.3 用谨慎情景框架推算隐含 ARR 倍数
关键纪律步骤是把 $1B 轮次翻译成 ARR 要求,而不是重复标题估值。公开基准让门槛变得可见。若按 SaaS Capital 的 6.7x 公开 SaaS 中位数交易,Eve 需要约 $149 million ARR。若要看起来像 Clio 2025 年融资,需要约 $80 million ARR。若要看起来像 Clio 2024 年轮次,需要约 $67 million ARR。若要看起来像 Intapp 更严苛得多的公开市场比率,则需要超过 $330 million ARR。这些门槛不是精确目标,但显示了公开市场支撑与主要按私有可选性定价的一轮之间,数量级差距有多大。 因为 Eve 不披露 ARR,公开材料唯一能翻译这轮价格的方法,是使用已披露客户数和谨慎合同价值区间。按轮次时披露的 450 家律所计算,假设每家混合 ARR 为 $20,000,只隐含约 $9 million ARR 和约 111x 倍数。假设 $40,000,则隐含约 $18 million ARR 和约 56x 倍数。即使采用非常慷慨的每家 $80,000 混合 ARR,仍只隐含约 $36 million ARR 和约 27.8x 倍数。这些数字对热门 AI 平台并非不可能,但明显高于已披露法律科技和公开 SaaS 可比公司通常能支撑的水平。 唯一能缓和这一结论的公开事实,是 Eve 当前首页披露的 1,200+ 家律所。如果其中大多数是付费客户,且混合约为每家 $50,000 ARR,当前 ARR 可能接近 $60 million,并回头把 2025 年 9 月价格压缩到约 16.7x ARR。这仍然昂贵,但远没有轮次时 450 客户数学那么极端。问题在于,这个去风险情景依赖公开记录没有提供的信息:付费还是试点状态、每家律所席位数、服务组合、续约行为,以及高端计划客户占比。因此,正确结论不是这一轮明显错误,而是它是一个押注执行的期权式价格,仍需要私下验证。[CV036, CV037, CV038, CV039, CV040, CV041]
估值判断不太取决于融资标题,更取决于当前付费 ARR 和留存能否支撑法律科技的溢价倍数。
[CV034, CV036, CV037, CV039, CV041, CV050]8.4 正论、反论与上行 / 下行情景
Eve 的建设性论点是连贯的。原告法律不是玩具切口。PI 和更广泛原告工作流宇宙足够大、运营足够重复、经济上足够敏感,值得用更好的线索受理、起草和和解执行来自动化。Eve 的公开故事也符合成长投资人对垂直 AI 赢家的期待:logo 快速增长、覆盖广泛工作流,并有野心成为操作系统而不是点状解决方案。如果管理层能证明当前 1,200+ 家律所的标题转化为真实付费 ARR、强留存,以及可扩展、轻服务的部署模型,那么 2025 年 9 月估值就会开始像是一次激进但可理解的品类领导者押注。 反论同样清晰。公开证据支持市场机会,但不支持价格。Clio 披露的轮次是在大得多的 ARR 规模上完成的。Intapp 的公开市场比率远低。SaaS Capital 的基准显示,SaaS 中位数环境仍清醒,而非亢奋。EvenUp 和 Harvey 说明 AI 稀缺溢价可能发生,但 Harvey 是通用法律 AI 离群值,EvenUp 仍未披露绝对 ARR。因此 Eve 处在一个尴尬估值区间:用公开法律软件可比公司看太贵,但又不够透明,拿不到 Harvey 式豁免传统纪律的资格。 因此剩下三个实用情景。多头情景要求 Eve 把当前动能转化为约 $250-300 million 退出 ARR,并以 12-14x 的优质退出定价成交,得到约 $3.0-4.2 billion 价值,相比 $1 billion 标记实现体面的多倍回报。基准情景严苛得多:$120-160 million 退出 ARR、7-9x 倍数,只得到约 $0.84-1.44 billion,几乎只能保本。空头情景——$60-90 million、4-6x——只产生 $0.24-0.54 billion,并摧毁价值。这种不对称意味着投资观点对证据质量高度敏感,而不只是对公司质量敏感。[CV044, CV045, CV046, CV047, CV048]
| 论点 | 支撑 | 反向权重 | 什么会改变判断 |
|---|---|---|---|
| 原告律所是真实的垂直软件市场 | 135k+ PI 律师、每年约 400k 起索赔,以及上升的 AI 支出,支撑品类规模 | 市场大不自动支撑异常高的进入倍数 | 在有意义规模上证明可持续变现和留存 |
| Eve 看起来正在走向操作系统级宽度 | 当前产品叙事覆盖 intake、诉前、诉讼、AI-ready 数据和律所智能 | 如果毛利率弱,产品宽度可能掩盖服务重交付 | 披露毛利率和产品化部署指标 |
| 快速客户增长可以支撑溢价 | 公开记录显示,律所数量从 Series A 的 100+ 增至 Series B 的 450+,当前官网为 1,200+ | 公开记录未定义付费与试点,也未说明按办公室还是按律所计数 | 提供付费客户、席位和 ACV 定义 |
| 法律 AI 存在稀缺价值 | Harvey 和 EvenUp 显示,投资人愿为品类领导者付高价 | Harvey 是 AI 异常值,Clio 在 ARR 高得多时拿到的倍数更低 | 证明 Eve 属于溢价集群,而不是中位数集群 |
| 平台上行仍可能很大 | 原告原生工作流赢家可能支撑数十亿美元退出价值 | 如果公开市场倍数保持克制,从 $1B 进入的基准退出数学偏弱 | 证明通往 $250M+ ARR 和高端留存的路径 |
本表把公司论点质量与当前价格公平性拆开;两者可以同时成立。
[CV043, CV044, CV045]| 情景 | 假设 | 估值 / 回报逻辑 | 关键风险 | 概率信号 |
|---|---|---|---|---|
| 牛市 | 退出 ARR 为 $250-300M,退出倍数 12-14x;Eve 证明平台级留存,并把 1,200+ 律所足迹变现 | 价值 $3.0-4.2B,或从 $1B 进入约 3.0-4.2x 总 MOIC | 执行、留存,以及溢价倍数可持续性 | 低到中;需要溢价集群级执行 |
| 基准 | 退出 ARR 为 $120-160M,退出倍数 7-9x;增长仍好,但公开市场纪律持续存在 | 价值 $0.84-1.44B,或约 0.8-1.4x 总 MOIC | 倍数压缩、中等 ACV、有限经营杠杆 | 没有超常证明时最可能 |
| 熊市 | 退出 ARR 为 $60-90M,退出倍数 4-6x;变现落后于 logo 增长,或竞争压缩定价 | 价值 $0.24-0.54B,或资本大幅减损 | Logo 质量、留存弱、服务拖累,或定价硬重置 | 如果证明质量仍薄,下行显著 |
情景输出刻意保持粗粒度,用来展示回报敏感性,而不是给出精确估值点估计。
[CV046, CV047, CV048]| 触发项 | 阈值 | 对论点的传导 | 行动含义 |
|---|---|---|---|
| 付费 ARR 证明低于当前去风险阈值 | 当前付费 ARR 仍低于约 $60M | 1,200+ 家律所的标题数字不足以压低进入倍数 | 不按上一轮估值支付;要求重置价格或退出 |
| 留存质量普通,而非高端 | 总留存或净留存低于软件领导者常模 | 稀缺溢价定价不再匹配平台论点 | 重估至公开市场中位数或更低 |
| 客户数量高度偏向试点 | 当前律所 logo 中,大量是试点、低 ACV 试用或非付费席位 | Logo 增长会夸大已变现规模 | 任何承销前,只从付费 cohort 重建 ARR |
| 新融资带来沉重优先权包袱 | 后续轮次加入激进优先权、参与权或反稀释条款 | 标题估值会夸大普通股价值 | 按 waterfall 重新计算回报,或避开交易 |
| 竞争性价格压力冲击原告 AI | 可比原告 AI 产品压缩有效 ACV,或迫使销售变得服务重 | 退出 ARR 和退出倍数同时恶化 | 立即切换到熊市情景假设 |
这些触发项聚焦会直接压垮估值论点的事实,而不是一般创业公司噪音。
[CV050, CV053]以 $1B 进入后的前瞻回报高度不对称,取决于 Eve 能否达到支撑溢价的 ARR 规模。
数值是由宽 ARR 和倍数区间推导出的隐含未来股权价值,单位为百万美元,不是折现现金流输出。
[CV046, CV047, CV048]8.5 建议、价格纪律与最终尽调问题
有证据支撑的建议是价格敏感的谨慎,而不是广泛乐观。对现有内部投资人来说,如果 Eve 已经显著超越公开 2025 年披露的规模,并能展示更接近优质平台而非重服务 AI 供应商的留存、利润率和扩张经济性,那么 2025 年 9 月标记仍可能跑通。但对新资金来说,公开记录还不足以仅凭已披露基本面就承保上一轮价格公平。至少,管理层需要证明当前付费 ARR 明显高于 $60-80 million 区间、续约行为强劲,并且优先权悬置有限,才可能让估值讨论更接近 Clio 式而非 Harvey 式数学。 因此,本章立场是「拉伸,但有条件可挽救」,而不是「显然破裂」。市场足够大、客户动能足够真实、工作流宽度足够广,仍能保留上行。但承保不能忽视 ARR 披露缺失、资本表条款缺失,以及当前 logo 数可能夸大当前货币化规模的可能性。最简单的实用纪律是要求证据或价格。没有私下验证时,更审慎的新资金区间约为 $700-800 million;在这个价格,哪怕不假设离群倍数,双位数回报仍有可能。 下一步尽调很直接。要求管理层提供从 2025 年 1 月至今的实际 ARR 桥、续约和总留存数据、分部 ACV、每家律所席位数,以及完整优先权瀑布。如果这些数据表明 Eve 已经进入类似 Clio 的扩张通道,估值立场可以迅速改善。否则,本章默认观点应维持为:在 2025 年 9 月标记下跟踪 / 继续研究,只在更低入场价格参与。[CV049, CV050, CV051, CV052, CV053]
| 维度 | 评估 | 决策含义 | 依据 |
|---|---|---|---|
| 建议 | 按上一轮价格跟踪 / 继续研究 | 没有私有 ARR 证明前,不要按 $1B 标记为新钱承销 | 公开证据支持增长牵引,但不足以给出干净的收入倍数 |
| 置信度 | 中 | 证据足以判断方向,不足以精确定价 | 融资事实清楚;ARR、留存和 cap table 条款不清楚 |
| 风险评级 | 高 | 回报结果高度取决于证明质量和倍数压缩 | 基准和熊市情景会很快抹掉风险投资式上行 |
| 估值立场 | 偏高 | 将 $1B 视为稀缺性溢价标记,而非有公开可比公司支撑的公允价值 | 融资时客户数量加上谨慎 ACV 假设,意味着倍数很高 |
| 新钱纪律 | 无证明时 $700-800M;只有强有力的当前 ARR 证明才支持 $1B | 要求降价,或证明当前付费 ARR 已显著高于 $60-80M | 10-12.5x 级法律科技倍数需要远高于公开披露的 ARR |
建议仅基于公开证据;对 ARR、留存和优先权条款的私有尽调,可能显著改变立场。
[CV050, CV051, CV052]| 主题 | 缺失证据 | 重要性 | 负责人或尽调路径 |
|---|---|---|---|
| ARR 和留存 | 2025 年 9 月至今按 cohort 划分的实际 ARR、NRR、GRR 和毛利率 | 这是稀缺溢价轮次与可支撑估值倍数之间的核心桥梁 | CFO 数据室加董事会报告 |
| 客户定义 | 付费客户、席位数、按分部划分的 ACV,以及 450+ 与 1,200+ 家律所披露口径的对账 | 情景数学取决于当前 logo 数量代表付费收入,还是更广泛的采用声明 | CRO / 销售副总裁审查并导出 cohort |
| Cap table 和 waterfall | 股份类别、清算优先权、参与权和反稀释条款 | 不能只靠标题估值评估普通股回报 | 公司法律顾问加 cap table 导出 |
| 单位经济 | 服务组合、CAC、回收期,以及按 cohort 划分的贡献利润率 | 如果实施高效、软件毛利率强,溢价定价可信度会高得多 | 财务和 RevOps 尽调 |
| 二级交易和内部定价 | Series B 以来的任何二级交易或内部标记 | 如果内部悄悄打折,会显著改变对 $1B 标题估值的解读 | 领投方和公司财务尽调 |
这些要求是从估值框架走向真实承销所需的最低材料包。
[CV049, CV050, CV052]IC 式评分对市场和产品宽度较为正面,但估值支撑和证据质量偏弱。
[CV044, CV045, CV049, CV050]8.6 图表
免责声明
本报告依赖公开来源,不能替代管理层访谈、客户访谈、融资文件、产品演示或内部经营数据。
证据索引
| 编号 | 陈述 | 可信度 | 来源 |
|---|---|---|---|
| CO001 | Eve is a legal AI platform built specifically for plaintiff law firms and positioned across the full case lifecycle. | 高 | SO001, SO004, SO017 |
| CO002 | The current homepage brands the platform as EveOS and says it is trusted by more than 1,200 firms. | 中 | SO001 |
| CO003 | Official and investor materials identify Jay Madheswaran, Matt Noe, and David Zeng as Eve’s co-founders. | 高 | SO002, SO021 |
| CO004 | Jay Madheswaran’s public background includes Facebook, Rubrik, and Lightspeed Venture Partners. | 高 | SO002, SO018, SO024 |
| CO005 | Matt Noe is publicly presented as co-founder and chief product officer and is described as a former Rubrik founding engineer. | 高 | SO002, SO014, SO021 |
| CO006 | David Zeng is publicly presented as co-founder and head of engineering with an AI and machine-learning background, and Lightspeed describes him as an early Rubrik engineer. | 中 | SO002, SO021 |
| CO007 | Lightspeed says it first met the founding team in 2020, making 2020 the best-supported origin anchor in the reviewed public record. | 中 | SO021 |
| CO008 | Eve announced a $47 million Series A in January 2025 led by Andreessen Horowitz with Lightspeed and Menlo Ventures supporting. | 高 | SO005, SO020 |
| CO009 | Eve announced a $103 million Series B in September 2025 at a valuation above $1 billion led by Spark Capital with Andreessen Horowitz, Lightspeed, and Menlo participating. | 高 | SO006, SO015, SO017 |
| CO010 | The two publicly announced 2025 rounds imply at least $150 million of disclosed capital raised since January 2025. | 高 | SO005, SO006, SO015 |
| CO011 | Series B materials say Eve added more than 350 firms after Series A and surpassed 450 customer firms within eight months. | 高 | SO015, SO016, SO017 |
| CO012 | Series B materials say Eve processes more than 200,000 legal cases annually. | 高 | SO006, SO015, SO021 |
| CO013 | Series B materials say Eve has helped customer firms collectively recover more than $3.5 billion in settlements and judgments. | 高 | SO006, SO015, SO021 |
| CO014 | Independent funding coverage repeatedly describes Eve as San Francisco-based. | 中 | SO016, SO018 |
| CO015 | Public identity surfaces do not cleanly verify “Eve Legal, Inc.”; Eve’s privacy materials refer to Butler Labs and SiliconANGLE says the company is officially Butler Labs Inc. | 高 | SO008, SO019 |
| CO016 | Official pages describe Eve as a full-lifecycle plaintiff-law workflow platform spanning intake, pre-litigation, litigation support, analytics, and AI-ready case data. | 高 | SO001, SO004, SO023 |
| CO017 | Eve publicly claims SOC II Type 2 certification, HIPAA compliance, and that customer prompts or documents are not used to train central foundation models. | 高 | SO007, SO012 |
| CO018 | Eve’s launch materials say firms using the product can increase case capacity about 2.5 times without adding headcount. | 中 | SO004, SO020 |
| CO019 | Andreessen Horowitz wrote that Eve could help plaintiff attorneys handle three to four times more cases than before. | 高 | SO020, SO024 |
| CO020 | By January 2025, Eve said more than 100 firms had adopted the product and those customers were driving 500% year-over-year revenue growth. | 中 | SO005 |
| CO021 | Andreessen Horowitz said Eve had added more than 80 law firms and increased its customer base 800% within a year around the Series A. | 高 | SO020, SO024 |
| CO022 | LawNext reported in January 2026 that Eve had achieved 10x revenue growth over the prior year and served more than 500 plaintiff firms. | 中 | SO022 |
| CO023 | Above the Law reported in March 2026 that more than 800 plaintiff firms were using Eve. | 中 | SO025 |
| CO024 | LawNext reported in June 2026 that Eve served more than 1,400 plaintiff law firms and had more than 200,000 active matters on the platform. | 中 | SO023 |
| CO025 | Public 2026 firm-count disclosures therefore range from 500-plus in January to 1,400-plus in June, which supports strong momentum but leaves the exact run-date total imprecise. | 中 | SO001, SO022, SO023, SO025 |
| CO026 | The January 2026 AI Workforce release introduced AI Agents, an AI Auditor, and an AI Analyst as distinct role-based components. | 高 | SO014, SO022 |
| CO027 | The June 2026 EveOS release added Atlas, Communication Agents, Eve Research, and expanded Analyst capabilities around a unified data layer. | 高 | SO001, SO023 |
| CO028 | The current homepage says Eve offers 24/7 intake in 28 languages and every fact is cited with legal research built in. | 中 | SO001 |
| CO029 | Barrett & Farahany said Eve saved attorneys more than 20 hours per week and reduced discovery-response work from 10 to 20 hours down to 30 to 45 minutes. | 高 | SO011, SO020 |
| CO030 | Mike Morse Law Firm said attorneys using Eve doubled or tripled their capacity and that the tool had become indispensable in daily practice. | 中 | SO010 |
| CO031 | James Scott Farrin’s case study says one attorney’s average case length fell from seven to eight months to four months when Eve was used throughout the case. | 中 | SO012 |
| CO032 | Hershey Law says it used Eve through trial preparation and won a $27.5 million California employment verdict. | 中 | SO013 |
| CO033 | The current homepage markets a 4.9 out of 5 G2 rating and says 15 to 30 percent higher settlements are reported across 1,200-plus customers. | 低 | SO001 |
| CO034 | SiliconANGLE says Eve can save customers up to 15 hours per week by automating intake and document review work. | 中 | SO019 |
| CO035 | LawNext and Eve’s own Series B copy both say attorneys keep judgment and review authority while AI handles routine or preparatory work. | 高 | SO006, SO022 |
| CO036 | LLRX documented that Eve marketing and training used “Trust but Verify” language around hallucination safeguards and treated Eve as part of a wider legal-tech overclaiming debate. | 中 | SO026, SO007 |
| CO037 | A March 27, 2026 apology letter in Rushing v. Turner says attorney Ross LeBlanc had used Eve to help draft pleadings that contained an inaccurate quotation. | 中 | SO027 |
| CO038 | The same letter also says LeBlanc could not determine whether the mistake came from Eve or from his own cut-and-paste process, leaving causation unresolved. | 中 | SO027 |
| CO039 | Stateline reported in January 2026 that bars and courts were issuing AI guidance on confidentiality, competence, and verification as hallucinated legal content spread. | 中 | SO028 |
| CO040 | Reviewed public sources do not disclose Eve’s board composition, ownership percentages, or investor control rights. | 中 | SO003, SO015, SO017 |
| CO041 | Reviewed public sources do not disclose a current employee headcount for Eve. | 中 | SO001, SO003, SO015 |
| CO042 | Spark and Eve both frame the new capital as funding a broader AI-native law platform, deeper onboarding, and transformation services rather than a point tool. | 高 | SO006, SO015, SO017 |
| CO043 | Customer-story and webinar materials show Eve has expanded beyond drafting into continuous audit, operational review, and firm-level performance insight. | 中 | SO009, SO014, SO022 |
| CM001 | Eve defines its core target market as plaintiff law firms rather than the legal market in general. | 高 | SM001, SM005 |
| CM002 | Eve’s product positioning centers on plaintiff workflows such as medical-record review, demand drafting, and discovery preparation. | 中 | SM001 |
| CM003 | Eve explicitly says it supports plaintiff-side practice areas including personal injury, workers’ compensation, medical malpractice, labor and employment, mass torts, and Social Security disability. | 中 | SM001 |
| CM004 | Eve’s market narrative is that plaintiff firms face tedious tasks and rising costs that can be improved with AI. | 中 | SM005 |
| CM005 | Eve’s intake case study says AI-driven intake increased client conversions by 40% while keeping response times under 30 seconds. | 中 | SM003 |
| CM006 | Plaintiff-firm workflow is an intake-to-settlement operating stack, not just a legal-research workflow. | 中 | SM001, SM009, SM010, SM011, SM012 |
| CM007 | ABA Rule 1.5 requires contingent-fee agreements to be written and to state fee percentages, expense treatment, and remittance mechanics. | 中 | SM029 |
| CM008 | ABA Rule 1.15 requires client and third-party property to be held separate from a lawyer’s own property and client funds to be kept in a separate account. | 中 | SM027 |
| CM009 | ABA trust-account-record rules require lawyers to maintain current financial records and retain them for five years after a representation ends. | 中 | SM028 |
| CM010 | ABA Formal Opinion 512 says lawyers using generative AI still owe duties of competence, confidentiality, communication, supervision, candor, and reasonable fees. | 高 | SM023, SM024 |
| CM011 | NYSBA says AI hallucinations continue to generate sanctions and other disciplinary consequences for counsel. | 中 | SM025 |
| CM012 | Thomson Reuters says lawyers still must verify AI-generated citations and should treat AI like an assistant rather than a lawyer. | 中 | SM026 |
| CM013 | The ABA says the U.S. lawyer population reached 1.37 million in 2025. | 中 | SM006 |
| CM014 | Clio says the United States has more than 135,000 personal-injury lawyers, about 10% of practicing attorneys. | 中 | SM007 |
| CM015 | Clio says nearly 400,000 personal-injury claims are filed annually in the United States, predominantly in state courts. | 中 | SM007 |
| CM016 | Clio says approximately 95% of personal-injury lawsuits resolve before trial. | 中 | SM007 |
| CM017 | Clio cites 2025 U.S. personal-injury-lawyer industry revenue of $61.7 billion. | 中 | SM007 |
| CM018 | Clio says 79% of legal professionals now incorporate AI tools into their daily work. | 中 | SM007 |
| CM019 | Clio says growing law firms are 18% more likely to adopt workflows such as electronic payments, online scheduling, and automated communications. | 中 | SM007 |
| CM020 | Thomson Reuters says the average law firm achieved 13% profit growth in 2025. | 中 | SM014 |
| CM021 | Thomson Reuters says smaller firms captured the lion’s share of 2025 demand growth as clients shifted work away from the most expensive firms. | 中 | SM014 |
| CM022 | Thomson Reuters says technology spending and talent costs are rising rapidly as firms invest in AI capabilities. | 中 | SM014 |
| CM023 | Wolters Kluwer says over 90% of respondents report using at least one AI tool in their daily workflow. | 中 | SM015 |
| CM024 | The ABA Law Practice summary of the 8am report says legal-AI adoption has more than doubled year over year. | 中 | SM016 |
| CM025 | The same 8am / ABA summary says many firms still lack formal policies, structured training, and governance frameworks for AI. | 中 | SM016 |
| CM026 | Thomson Reuters says contingent-fee plaintiff wins historically accrued to generally small plaintiff firms while large firms mostly defended those matters. | 中 | SM013 |
| CM027 | Because contingent-fee economics differ from hourly defense work, plaintiff workflow software is closer to an operating system for case throughput than to a BigLaw research product. | 中 | SM013, SM014 |
| CM028 | Smokeball’s PI workflow examples include reduction requests, settlement receipt, disbursement sheets, and signed disbursement confirmation. | 中 | SM009 |
| CM029 | CloudLex says plaintiff legal workflows include client intake, medical-record retrieval, legal drafting, and case management. | 中 | SM010 |
| CM030 | CasePeer says PI paralegals manage intake forms, medical records, deadlines, and post-settlement work. | 中 | SM011 |
| CM031 | Paxton says personal-injury litigation runs from intake and case evaluation to final settlement, with early emphasis on police reports, medical records, photos, and witness statements. | 中 | SM012 |
| CM032 | Eve says most firms can become operational within 90 days without replacing their core case-management or document workflows. | 中 | SM001, SM002 |
| CM033 | Eve says partners, associates, paralegals, and operations professionals all fit naturally into the product’s workflow. | 中 | SM001 |
| CM034 | Eve says no work product is sent to clients or opposing counsel without a human in the loop and that agent claims are cited to source documents. | 中 | SM001 |
| CM035 | Eve says users have seen a 2-3x increase in attorney capacity. | 中 | SM001 |
| CM036 | Eve says complaint drafting can shrink from five hours to one hour. | 中 | SM001 |
| CM037 | Eve says agents can help secure settlements and verdicts up to 15% faster. | 中 | SM001 |
| CM038 | Lex Machina says premises-liability and motor-vehicle tort cases reached record levels in its 2025 torts report. | 中 | SM017 |
| CM039 | The PFAS AFFF MDL page says that single MDL has approximately 10,000-plus associated cases and tens of thousands of plaintiffs. | 中 | SM021 |
| CM040 | Duane Morris says its 2025 class-action review analyzed more than 1,441 decisions from the prior year. | 中 | SM022 |
| CM041 | The U.S. Courts civil data page exposes current federal civil and product-liability filing tables through 2025. | 中 | SM020 |
| CM042 | NCSC says its caseload dashboards are sourced directly from state courts and are built to monitor changing patterns across state systems. | 中 | SM018 |
| CM043 | The Federal Judicial Center’s Integrated Database is a primary federal-cases source for class-action, MDL, and nature-of-suit analysis. | 中 | SM019 |
| CM044 | AAJ describes the plaintiff bar as a network with U.S. and Canadian affiliates in diverse areas of trial advocacy. | 中 | SM030 |
| CM045 | The served market is best described as a fragmented plaintiff bar spanning small PI firms, scaled regional litigation shops, mass-tort platforms, and class-action plaintiff practices. | 中 | SM006, SM013, SM017, SM021, SM022, SM030 |
| CM046 | Because most PI cases settle before trial, pre-trial throughput and settlement operations matter more to ROI than trial-only or research-only features. | 中 | SM007, SM009, SM012 |
| CM047 | Because plaintiff firms must handle contingent-fee remittance and segregated client funds, trust-accounting controls are part of product-market fit rather than only back-office compliance. | 高 | SM027, SM028, SM029 |
| CM048 | Because AI use is already common but governance is uneven, successful plaintiff-workflow deployment depends on leadership-approved rollout, training, and verification controls rather than on attorney-by-attorney experimentation. | 中 | SM015, SM016, SM023, SM024, SM025, SM026 |
| CM049 | Public sources provide bounded plaintiff-market proxies but do not isolate a clean SAM or SOM for plaintiff-workflow AI. | 中 | SM006, SM018, SM019, SM020 |
| CM050 | Generic legal AI or defense-oriented tools do not fully replace plaintiff-native workflows such as 24/7 intake, medical chronology, demand drafting, discovery prep, and settlement operations. | 中 | SM001, SM003, SM009, SM010, SM011, SM012 |
| CM051 | Mass-tort and class-action plaintiff practices add bursty claimant coordination and large-scale litigation complexity on top of the core personal-injury workflow stack. | 中 | SM021, SM022 |
| CP001 | Eve markets EveOS as the AI operating system for plaintiff law and says it is trusted by 1,200-plus firms. | 中 | SP001 |
| CP002 | Eve publicly spans 24/7 intake, medical overviews, drafting, research, discovery, communications, and firm analytics, and says it can work with an existing CMS or stand alone. | 中 | SP001 |
| CP003 | EvenUp markets itself as leading AI for personal injury lawyers and publicly covers intake, treatment, demands, negotiation, discovery, trial, analytics, and communication agents. | 中 | SP002 |
| CP004 | EvenUp says more than 2,000 personal injury firms choose its platform. | 中 | SP002 |
| CP005 | EvenUp’s public funding coverage says its Claims Intelligence Platform uses a proprietary model trained on hundreds of thousands of injury cases and millions of medical records. | 中 | SP028 |
| CP006 | Artificial Lawyer reports that EvenUp raised a $150 million Series E at a $2 billion-plus valuation and $385 million total capital. | 中 | SP028 |
| CP007 | EvenUp’s public story remains centered on claims intelligence and personal-injury operations rather than a plaintiff-firm operating system that explicitly spans every attorney, case, and dollar. | 中 | SP001, SP002 |
| CP008 | Supio markets itself as the only agentic legal AI platform built for plaintiff law and mass torts. | 中 | SP003 |
| CP009 | Supio explicitly markets intake, chronologies, demand letters, litigation drafting, cross-case analysis, and firm intelligence in one platform. | 中 | SP003 |
| CP010 | Supio says it integrates with Westlaw, Litify, MyCase, and CasePeer, which implies it is designed to layer onto existing systems rather than always replace them. | 中 | SP003 |
| CP011 | Supio’s competitive materials claim 97% verified accuracy with human experts validating outputs before delivery. | 中 | SP004, SP005 |
| CP012 | Supio’s public comparison pages say it supports the full PI spectrum from single-injury cases to mass torts and markets flat pricing with no platform fees or page limits. | 中 | SP004, SP005 |
| CP013 | Supio’s 2025 financing announcement says the company has raised $91 million in total and is scaling product, research, and go-to-market around plaintiff-law demand. | 中 | SP006 |
| CP014 | Supio’s public attack lines against Eve focus on hallucination risk, shallow integrations, and insufficient litigation depth, showing where direct rivalry is already active. | 中 | SP004 |
| CP015 | Darrow markets legal-exposure management that detects emerging litigation opportunities from regulatory filings, incident reports, market activity, and litigation patterns before harm fully surfaces. | 中 | SP007 |
| CP016 | Darrow serves law firms, insurers, and compliance teams, which makes it an adjacent origination and portfolio-intelligence competitor rather than a plaintiff-firm operating system. | 中 | SP007 |
| CP017 | Darrow says it has surfaced more than $22 billion of legal exposure, detected more than 5 million signals monthly, and connected more than 10,000 active matters through its intelligence. | 中 | SP007 |
| CP018 | Plaintifi’s public website describes a service that matches people with personal injury claims to vetted attorneys, not an AI workflow platform for plaintiff firms. | 中 | SP008 |
| CP019 | ProPlaintiff publicly markets AI demand letters, AI summaries, an AI paralegal, a case manager, AI medical chronologies, and document generation from intake to settlement. | 中 | SP009 |
| CP020 | LawPro.ai publicly markets visual chronology, file analysis, case valuation, citation-backed answers, and legal-document generation for personal injury lawyers. | 中 | SP010 |
| CP021 | Filevine’s public feature set includes AI medical-record analysis, AI data mapping, case validation workflows, deposition intelligence, calendaring, and payments. | 中 | SP011 |
| CP022 | Litify publicly markets legal intelligence embedded in every field and workflow and sells personal-injury workflows from intake to settlement with configurable questionnaires, matter plans, analytics, and medical-record requests. | 高 | SP018, SP019 |
| CP023 | Litify’s public PI solution language shows that incumbent workflow suites can already cover intake, communications, reporting, deadlines, and settlement progression without adopting a new AI-first operating system. | 中 | SP019 |
| CP024 | CloudLex says it is built exclusively for personal injury firms from intake through settlement and includes AI lead capture, demand drafting, medical summaries and timelines, a 24/7 client web assistant, and paralegal services. | 中 | SP020 |
| CP025 | CasePeer says it is case-management software designed exclusively for personal injury law firms and highlights reporting, marketing, CRM, client intake, document automation, accounting hubs, and legal AI. | 中 | SP021 |
| CP026 | Clio’s public feature stack covers client intake, CRM, workflow automation, document automation, billing, client portals, and Manage AI, but it is a general legal practice platform rather than a plaintiff-native one. | 中 | SP012 |
| CP027 | Clio publicly prices its software from $49 per user and routes large firms to custom quotes. | 中 | SP013 |
| CP028 | MyCase publicly positions itself as practice management from client intake to billing with workflow automation and legal AI, which makes it a general-practice substitute rather than a plaintiff operating system. | 中 | SP014 |
| CP029 | MyCase publicly lists annual pricing from $50 to $130 per user per month, with monthly billing from $60 to $150 and no long-term contract requirement. | 中 | SP015 |
| CP030 | PracticePanther publicly markets custom workflows, automated intake forms, calendaring, billing, accounting, and general practice management across many practice areas. | 中 | SP016 |
| CP031 | PracticePanther publicly prices from $49 to $89 per user per month on annual plans, with monthly pricing up to $99. | 中 | SP017 |
| CP032 | Smokeball remains a document-heavy status-quo substitute rather than a plaintiff-native AI operating system. | 低 | SP022 |
| CP033 | Harvey publicly sells agents that execute complex legal work end to end across knowledge, contract intelligence, litigation, and broader legal transformation. | 中 | SP023 |
| CP034 | CNBC reports that Harvey raised $200 million at an $11 billion valuation in 2026 and that its products are used by more than 100,000 lawyers across 1,300 organizations. | 中 | SP024 |
| CP035 | Thomson Reuters’ CoCounsel page publicly positions the product across drafting, legal research, due diligence, spend and matter management, and cites a customer story claiming 5x ROI and doubled litigation capacity. | 中 | SP025 |
| CP036 | Lexis+ with Protégé publicly positions itself for drafting, research, and analysis while grounding outputs in LexisNexis content, Shepard’s citations, web and news sources, and connected document-management systems. | 中 | SP026 |
| CP037 | LawNext reports that Harvey and LexisNexis are partnering to integrate Lexis content, gen-AI technology, and advanced legal workflows inside Harvey. | 中 | SP030 |
| CP038 | Paxton publicly markets itself as an all-in-one AI legal assistant for drafting, document analysis, contextual research, medical chronologies, and billing summaries across practice areas including personal injury. | 中 | SP027 |
| CP039 | Crunchbase News says disclosed funding for EvenUp, Eve, Supio, and Darrow totals roughly $682 million, with plaintiff-focused companies accounting for about 71% of disclosed legal-AI capital in the covered set. | 中 | SP029 |
| CP040 | Crunchbase News argues that plaintiff-side legal AI has attracted capital because intake, case evaluation, medical review, and demand generation are unusually standardized workflows. | 中 | SP029 |
| CP041 | Artificial Lawyer explicitly groups EvenUp, Eve, and Supio together as fast-growing plaintiff-side legal-AI companies in the same part of the market. | 中 | SP028 |
| CP042 | Public integration and overlay language from Eve and Supio implies that plaintiff firms can plausibly multi-home AI tools on top of an existing case-management system. | 高 | SP001, SP003 |
| CP043 | Eve’s strongest public differentiation is the combination of plaintiff specialization, workflow breadth from intake through discovery, and explicit firm-level operating-system ambition. | 中 | SP001, SP002, SP003 |
| CP044 | Eve faces direct pressure from EvenUp on scaled pre-lit operations, from Supio on medical-depth and litigation-readiness messaging, and from incumbent suites on system-of-record control. | 中 | SP002, SP003, SP019, SP020 |
| CP045 | General legal-AI vendors can commoditize drafting and research layers, but the reviewed public materials do not show plaintiff-native intake, settlement, or case-operations depth comparable to Eve’s positioning. | 中 | SP001, SP023, SP025, SP026, SP027 |
| CP046 | Public pricing transparency is materially better among Clio, MyCase, and PracticePanther than among Eve, EvenUp, Supio, Litify, Filevine, CloudLex, and CasePeer. | 中 | SP001, SP002, SP003, SP013, SP015, SP017 |
| CP047 | The deepest switching costs in this market likely sit with incumbent matter-management suites because they already control intake, documents, analytics, calendars, and other case-operation records. | 中 | SP019, SP020, SP021, SP012 |
| CP048 | Plaintiff specialization still matters because the most heavily funded and fastest-growing direct peers all organize around repeated plaintiff workflows such as intake, medical review, demand generation, and litigation support. | 中 | SP002, SP003, SP006, SP029 |
| CP049 | Reviewed public sources do not disclose exact 2026 price cards, implementation fees, retention metrics, or NRR for Eve and most direct plaintiff-AI peers. | 中 | SP001, SP002, SP003, SP006 |
| CP050 | Supio’s public comparison pages show that verification, integration depth, and litigation-grade output quality are already live competitive attack lines in the plaintiff-AI category. | 中 | SP004, SP005 |
| CI001 | Eve announced a $47 million Series A led by Andreessen Horowitz with Lightspeed and Menlo participating. | 高 | SI002, SI030 |
| CI002 | Eve announced a $103 million Series B at a valuation above $1 billion, led by Spark Capital with Andreessen Horowitz, Lightspeed, and Menlo participating. | 高 | SI003, SI004, SI031 |
| CI003 | Eve’s 2026 careers and job materials say the company has raised more than $160 million, implying earlier capital beyond the two disclosed 2025 rounds. | 中 | SI007, SI018, SI019 |
| CI004 | The two disclosed 2025 rounds alone imply at least $150 million of capital raised before any earlier seed financing. | 高 | SI002, SI003, SI004 |
| CI005 | The publicly named investor base includes Spark Capital, Andreessen Horowitz, Lightspeed Venture Partners, and Menlo Ventures. | 高 | SI003, SI004, SI018 |
| CI006 | No reviewed public source discloses Eve’s cash on hand, monthly burn, or runway months. | 中 | SI003, SI007, SI018 |
| CI007 | Eve markets a firm-wide plaintiff-law AI platform spanning intake, drafting, discovery, and analytics rather than a single-task tool. | 高 | SI001, SI008, SI009 |
| CI008 | Eve does not publish list pricing on its public product pages and instead routes prospects to demos or sales calls. | 高 | SI001, SI008, SI009, SI010 |
| CI009 | Eve’s Head of Legal posting references MSAs, customer agreements, partnership agreements, vendor contracts, and Deal Desk, indicating contract-led B2B selling. | 中 | SI018 |
| CI010 | Eve AI Intake reached general availability after a beta with more than 40 plaintiff firms. | 中 | SI011 |
| CI011 | Eve’s 2026 product narrative positions EveOS, AI Intake, and Agents as modules inside a broader plaintiff-law operating system. | 中 | SI005, SI006, SI008 |
| CI012 | Official Series B materials say Eve added more than 350 firms in eight months and reached more than 450 customer firms by September 2025. | 高 | SI003, SI004, SI031 |
| CI013 | Official Series B materials say Eve processes more than 200,000 legal cases annually. | 高 | SI003, SI004, SI031 |
| CI014 | Official Series B materials say Eve has helped firms recover more than $3.5 billion in settlements and judgments. | 高 | SI003, SI004, SI031 |
| CI015 | 2026 job postings say Eve is trusted by more than 1,000 law firms. | 高 | SI018, SI019 |
| CI016 | 2026 job postings say Eve is growing revenue two times quarter over quarter. | 高 | SI018, SI019 |
| CI017 | No reviewed source discloses an absolute 2026 revenue or ARR figure for Eve. | 中 | SI003, SI018, SI019 |
| CI018 | LawNext reported that EveOS was launched partly because AI agents and humans working across disconnected systems create data drift and audit gaps. | 中 | SI006 |
| CI019 | Archuleta Law Firm says it handles roughly 1,000 leads per month across all 50 states. | 中 | SI014 |
| CI020 | Archuleta Law Firm says roughly 50% of inbound callers voluntarily choose AI intake when given the option. | 中 | SI014 |
| CI021 | Archuleta Law Firm says Eve doubled intake capacity without adding staff. | 中 | SI014 |
| CI022 | Eve’s AI Intake launch post says the beta cohort let firms scale capacity without adding staff and treat intake as a growth lever. | 中 | SI011 |
| CI023 | Laurel Employment Law says it reached more than 1,500 active clients and more than 100 employees across five continents within 24 months of launching with Eve. | 中 | SI015 |
| CI024 | Laurel Employment Law says demand-letter drafting dropped from two to four hours to about 15 minutes with Eve. | 中 | SI015 |
| CI025 | Laurel says its demand-drafting agent increased weekly mailed demand letters from 48 to 104 and cut human touch time to about three to five minutes per letter. | 中 | SI016 |
| CI026 | Frontier Law Center says Eve turned several days of interrogatory work into a 45-minute task and now lets the team do five times more work in the same time. | 中 | SI017 |
| CI027 | Eve’s public ROI framing emphasizes higher settlements, faster case movement, and more capacity, but those are company-claimed outcomes rather than audited financial metrics. | 中 | SI001, SI009, SI010, SI011 |
| CI028 | MyCase publicly lists annual per-user pricing of $50 for Basic, $100 for Pro, and $130 for Advanced. | 中 | SI023 |
| CI029 | PracticePanther publicly lists annual per-user pricing of $49 for Solo, $69 for Essential, $89 for Business, and $114 for Business Pro. | 中 | SI024 |
| CI030 | Clio publicly starts at $49 per user, while higher-growth, AI, and PI-oriented modules move to demo or custom pricing. | 中 | SI022 |
| CI031 | Because Eve discloses no list price, adjacent legal-software price cards only bound a plausible spend range rather than reveal Eve’s realized ACV. | 中 | SI022, SI023, SI024 |
| CI032 | A conservative benchmark scenario of 1,000 firms paying five seats at $50 per user per month implies roughly $3.0 million of annualized software spend. | 低 | SI019, SI023 |
| CI033 | An upper benchmark scenario of 1,000 firms paying 20 seats at $130 per user per month implies roughly $31.2 million of annualized software spend. | 低 | SI019, SI023 |
| CI034 | Jobera listed 34 open positions for Eve in June 2026 across engineering, revenue operations, finance, security, marketing, sales, success, and legal. | 中 | SI021 |
| CI035 | Named June 2026 compensation bands span at least $82,000 to $450,000 of base salary across public Eve openings. | 中 | SI021 |
| CI036 | Eve’s careers materials advertise equity, 401(k) matching, benefits, stipends, and team gatherings, so fully loaded labor cost exceeds base salary alone. | 高 | SI007, SI018, SI019 |
| CI037 | The Head of Legal role alone carries a $280,000 to $340,000 base range, indicating premium compliance and contracting cost. | 中 | SI018 |
| CI038 | The ML Engineer role carries a $195,000 to $350,000 base range, indicating premium model and infrastructure hiring cost. | 中 | SI019 |
| CI039 | Intapp’s March 2026 filed results showed 27% SaaS revenue growth, 31% cloud ARR growth, and 123% cloud net revenue retention. | 中 | SI025 |
| CI040 | DISCO’s first-quarter 2026 results showed 14% total revenue growth, 12% software revenue growth, 347 customers above $100,000, and a continued GAAP net loss. | 中 | SI026 |
| CI041 | These public comp disclosures suggest scaled legal and professional software can combine strong recurring growth with improving but not necessarily positive earnings. | 高 | SI025, SI026 |
| CI042 | Eve discloses no public gross margin, CAC, payback, NRR, or revenue mix metrics, so any unit-economics view remains proxy-based. | 中 | SI018, SI019, SI025 |
| CI043 | A March 27, 2026 court letter shows attorney Ross LeBlanc accepted responsibility for filing motions with unverified quotations. | 中 | SI029 |
| CI044 | B17 reported that the incident created a public blame dispute over whether Eve hallucinated the bad cites, which is a reputational and QA risk for the vendor. | 高 | SI027, SI029 |
| CI045 | EDRM reported at least $145,000 of U.S. sanctions from AI-generated fake citations in the first quarter of 2026, raising governance costs for legal-AI vendors and buyers. | 高 | SI028, SI032 |
| CI046 | The ABA article says legal AI adoption rose to 69% from 31% year over year, but 54% of firms still had no generative-AI training plans. | 中 | SI032 |
| CI047 | Post-Series B capital materially reduced near-term financing pressure versus early 2025, but the public record still cannot support a precise runway calculation. | 高 | SI002, SI003, SI004, SI006 |
| CI048 | Eve looks like a fast-growing, contract-sold plaintiff-law SaaS platform with credible customer ROI and capital access, but its valuation case still rests on growth signals rather than disclosed revenue-quality metrics. | 中 | SI003, SI004, SI018, SI021 |
| CI049 | The gap between 450-plus firms in September 2025 and the 1,000-plus-law-firm claim in 2026 suggests continued expansion, but the exact current paying-customer count and seat mix are not externally reconciled. | 中 | SI004, SI018, SI019 |
| CI050 | The main financial diligence blockers are absolute ARR or revenue, realized pricing and ACV, gross margin, retention, CAC, payback, cash, burn, debt, and customer concentration. | 中 | SI018, SI019, SI021, SI025 |
| CE001 | Eve markets EveOS as an AI operating system for plaintiff law firms. | 中 | SE001 |
| CE002 | The homepage positions Eve across intake, pre-litigation, litigation, and AI-ready data rather than a single drafting task. | 中 | SE001 |
| CE003 | Eve says its intake layer answers calls and emails 24/7 in 28 languages. | 中 | SE001, SE017 |
| CE004 | Eve says qualified clients can sign engagement documents live on the intake call. | 中 | SE001, SE017 |
| CE005 | Eve says pre-litigation workflows include chronologies, motions, and demands drafted in the firm's style with cited facts. | 中 | SE001 |
| CE006 | Eve says litigation workflows include cross-examination chapters, deposition summaries, and full discovery-response drafting. | 中 | SE001 |
| CE007 | Eve says its AI-ready data layer can work with an existing CMS or as a standalone system. | 中 | SE001 |
| CE008 | Eve says the product auto-extracts and structures records, bills, and calls with no manual data entry. | 中 | SE001 |
| CE009 | The Medical Overview module is described as including a narrative summary, visit chronology, bad facts, ICD codes, an economic damages ledger, and a non-economic damages estimate. | 中 | SE002 |
| CE010 | The Medical Overview page says every data point is hyperlinked to a source document and page number, with direct quotes visually distinguished from AI summaries. | 中 | SE002 |
| CE011 | Eve says a Medical Overview is generated in roughly 15–20 minutes. | 中 | SE002 |
| CE012 | Eve says the Medical Overview can process tens of thousands of pages and use OCR on scanned and handwritten records. | 中 | SE002 |
| CE013 | Eve's discovery guide says the product can draft discovery requests and responses, suggest objections, summarize productions, and support deposition prep and analysis. | 中 | SE003 |
| CE014 | The AI Agents page says agents monitor case updates and act automatically instead of waiting for prompts. | 中 | SE007, SE008 |
| CE015 | The AI Agents page says agents handle intake, record review, chronologies, demand drafting, discovery work, and routine status updates. | 中 | SE007 |
| CE016 | Eve says agent outputs are queued for attorney review and approval before they leave the firm. | 中 | SE007, SE014 |
| CE017 | The AI Agents page claims a 2–3x increase in attorney capacity. | 低 | SE007 |
| CE018 | The AI Agents page claims cases resolve up to 15% faster when agents keep work moving. | 低 | SE007 |
| CE019 | Eve 2.0 introduced a role-based architecture centered on Agents, Auditor, and Analyst. | 中 | SE008, SE014, SE016 |
| CE020 | Lawnext independently reports that Eve is positioning the product as a proactive AI workforce rather than a prompt-only assistant. | 中 | SE014 |
| CE021 | Lawnext reports that Auditor reviews every document in every case for missed deadlines, overlooked injuries, factual gaps, and risk exposure. | 中 | SE014 |
| CE022 | Lawnext reports that Analyst is meant to surface firm-level performance differences, bottlenecks, and return patterns across a docket. | 中 | SE014 |
| CE023 | Lawnext reports that Atlas is a self-updating case-data layer that pulls from case-management software, emails, court filings, medical records, and client communications. | 中 | SE015 |
| CE024 | Lawnext reports that Atlas flags uncertainty for human review when the system is unsure of the correct source of truth. | 中 | SE015 |
| CE025 | EveOS launch materials say Analyst moved into beta for plain-English operational reporting with live dashboards. | 中 | SE015, SE017 |
| CE026 | EveOS launch materials say Communication Agents automate outbound follow-ups, records requests, onboarding, and status updates across 31 languages. | 中 | SE015, SE017 |
| CE027 | EveOS launch materials say Eve Research has native access to court opinions across U.S. jurisdictions, flags overruled rulings, and links citations to source passages. | 中 | SE015, SE017 |
| CE028 | EveOS launch materials claim that intake with signing increased qualified leads by 50% for firms using Jenny. | 低 | SE017 |
| CE029 | The Clio integration page says Eve syncs Clio case details, contacts, and notes directly into Eve. | 中 | SE004 |
| CE030 | Working With Eve says the product provides inline sourcing, one-click verification, and an AI validation framework intended to reduce hallucination risk. | 中 | SE010 |
| CE031 | Eve security pages say the platform uses isolation at the organization, user, and workflow levels, does not use customer prompts or outputs to train foundation models, and is reviewed through annual audits. | 中 | SE005 |
| CE032 | Working With Eve says Eve uses AES-256 encryption and claims SOC 2 Type II and HIPAA compliance. | 中 | SE010 |
| CE033 | Eve's privacy policy says website personal data may be shared with service providers, analytics partners, and advertising partners. | 中 | SE006 |
| CE034 | Eve's careers page says the company builds in small, fast-moving pods and uses tight feedback loops with direct client-call exposure. | 中 | SE011 |
| CE035 | Built In lists Django, PostgreSQL, Python, React, and TypeScript as technologies used at Eve. | 中 | SE026 |
| CE036 | The ML Engineer posting says Eve fine-tunes models on domain-specific data including code, natural language, and product usage signals. | 中 | SE025 |
| CE037 | The ML Engineer posting says the role owns evaluation frameworks and production deployment and notes collaboration with OpenAI and Anthropic. | 中 | SE025 |
| CE038 | Eve's search-infrastructure post says the company built per-case search indices and left OpenSearch after shard instability and costs above $10,000 per month on AWS. | 中 | SE028 |
| CE039 | Eve's WALL-E engineering post says an internal background agent now authors 22% of monorepo merge requests and 53% of those are one-shot. | 中 | SE027 |
| CE040 | Laurel's case study says demand drafting now auto-triggers from uploaded client-call transcripts and takes 3–5 minutes of human time. | 中 | SE012 |
| CE041 | Laurel's case study says weekly mailed demand letters increased from 48 to 104 after deploying the Demand Drafting Agent. | 中 | SE012 |
| CE042 | Hershey Law's case study says the firm built case-phase playbooks in Eve and used the product for real-time deposition citations, limine tracking, and overnight trial prep. | 中 | SE013 |
| CE043 | Above the Law sponsored coverage says one Eve customer raised lead conversion from 10% to 35% using Jenny. | 低 | SE019 |
| CE044 | The same Above the Law coverage says the firm shortened intake by 50 minutes and increased average case value by 90%. | 低 | SE019 |
| CE045 | Artificial Lawyer's intake walkthrough says Eve classifies calls, auto-generates intake forms, integrates with case-management systems, and is built on an open ecosystem with API availability. | 中 | SE018 |
| CE046 | Public adverse evidence does not support treating Eve's verification marketing as a guarantee: LLRX highlighted a March 2026 attorney letter saying Eve safeguards were discussed, yet an inaccurate quotation still reached the court. | 中 | SE020, SE024 |
| CE047 | EDRM says U.S. courts imposed at least $145,000 of sanctions in Q1 2026 for AI-generated fake citations, raising procurement and governance stakes for legal-AI tools. | 中 | SE021 |
| CE048 | LawAccounting argues the citation-audit trail must live natively inside the matter platform rather than in a disconnected AI workspace. | 中 | SE023 |
| CE049 | TechNewsWorld reports that hallucinated legal logic can survive cite checks and that shadow-AI usage is a growing governance risk in law firms. | 中 | SE022 |
| CE050 | Eve's public scale claims rose from 500-plus firms in January 2026 to 800-plus in March sponsored coverage, 1,000-plus in hiring pages, and 1,200-plus on the June homepage, which suggests rapid adoption but not independently audited active usage. | 中 | SE001, SE008, SE019, SE025 |
| CU001 | Eve publicly targets plaintiff law firms across personal injury, labor and employment, workers’ compensation, Social Security disability, medical malpractice, and related plaintiff workflows. | 中 | SU001, SU015 |
| CU002 | Public product materials position Eve across intake, pre-litigation, litigation, and firm intelligence rather than as a narrow drafting-only tool. | 中 | SU001, SU013 |
| CU003 | On April 7, 2026, Eve said it had surpassed 1,000 plaintiff law firm customers and powered more than 200,000 cases on the platform. | 高 | SU007, SU009, SU010 |
| CU004 | LawNext reported in January 2026 that Eve served more than 500 plaintiff firms. | 中 | SU011 |
| CU005 | Above the Law reported in March 2026 that more than 800 plaintiff firms were using Eve. | 中 | SU012 |
| CU006 | By June 2026, public sources diverged on Eve’s current footprint, with the homepage saying 1,200-plus firms and LawNext saying more than 1,400 plaintiff law firms. | 中 | SU001, SU013, SU014 |
| CU007 | The visible public reference set spans multiple U.S. geographies, including Michigan, North and South Carolina, California, and Massachusetts. | 中 | SU002, SU003, SU004, SU005, SU022, SU023, SU024, SU025 |
| CU008 | Public materials position Eve for plaintiff firms at multiple case-management stages rather than only one attorney persona or one isolated workflow. | 中 | SU001, SU015 |
| CU009 | Eve’s public commercial motion is schedule-a-call, walkthrough, waitlist, and quote led rather than self-serve checkout led. | 中 | SU001, SU008, SU015 |
| CU010 | Eve does not publish standard pricing publicly, and outside reviewers describe the commercial motion as demo and quote based. | 中 | SU015, SU016 |
| CU011 | External pricing commentary says integration, onboarding, staff training, and data migration can materially affect implementation cost. | 中 | SU016 |
| CU012 | Mike Morse Law Firm is a large personal-injury customer reference; the Eve case study says it has 250 legal professionals, more than 60 lawyers, and thousands of cases per year, while the firm’s own site calls it the largest PI firm in Michigan. | 中 | SU002, SU022 |
| CU013 | Mike Morse described fear of job loss and change aversion as the main barrier to AI adoption inside the firm. | 中 | SU002 |
| CU014 | At Mike Morse, one attorney said she uses Eve roughly 75 times a day for demand drafting, medical-record review, and live adjuster calls. | 中 | SU002 |
| CU015 | Mike Morse said attorney capacity doubled or tripled after Eve and that the firm had saved thousands of hours of workflow time. | 中 | SU002 |
| CU016 | Mike Morse tied Eve’s document search and synthesis to faster client callbacks and potentially higher case value by surfacing details buried in large files. | 中 | SU002 |
| CU017 | The Law Offices of James Scott Farrin is a large Mid-Atlantic plaintiff firm with more than 60 attorneys across personal injury, workers’ compensation, social security, mass tort, and eminent domain in North and South Carolina. | 中 | SU003, SU023 |
| CU018 | James Scott Farrin rolled Eve out through a 300-person firm implementation led by legal-technology and department leaders rather than through purely individual attorney adoption. | 中 | SU003 |
| CU019 | Before broad rollout, James Scott Farrin verified Eve’s SOC2 and HIPAA posture and required an Acceptable Use Policy firmwide. | 中 | SU003, SU021 |
| CU020 | James Scott Farrin reported that medical summaries moved from weeks to 30–60 minutes and that one workers’ compensation attorney cut average case length from seven or eight months to four months when Eve was used throughout. | 中 | SU003 |
| CU021 | James Scott Farrin says the firm meets weekly with Eve and its developers, implying expansion depends on ongoing customization and vendor involvement. | 中 | SU003 |
| CU022 | Hershey Law is a California employment-law customer, showing that Eve’s public customer base extends beyond personal injury into plaintiff-side employment matters. | 中 | SU004, SU024 |
| CU023 | Hershey Law built stage-specific playbooks and rolled Eve out across intake, pre-litigation, litigation, and operations until every team at the firm was using it. | 中 | SU004 |
| CU024 | Hershey Law publicly credits Eve with trial-prep and case-prep speed, but the public record does not isolate how much of the firm’s $27.5 million verdict was attributable to Eve versus lawyering and case facts. | 中 | SU004 |
| CU025 | Jeffrey Glassman Injury Lawyers is a large Boston plaintiff practice spanning personal injury, workers’ compensation, Social Security disability, medical malpractice, and mass torts. | 中 | SU005, SU025 |
| CU026 | Jeffrey Glassman said roughly 90 percent of the firm’s demands were going out through Eve within about three months of rollout. | 中 | SU005 |
| CU027 | Jeffrey Glassman’s customer story says Eve cut medical chronology work from hours to about 20 minutes and freed more time for client interaction. | 中 | SU005 |
| CU028 | Frontier Law Center focuses primarily on plaintiff-side employment law, showing Eve can land in employment-litigation workflows as well as personal injury. | 中 | SU017, SU018, SU019 |
| CU029 | Frontier’s public metrics moved from 10 percent to 35 percent lead conversion, shortened intake by 50 minutes, and raised average case value by 90 percent after using Eve’s intake agent. | 中 | SU012, SU017 |
| CU030 | Frontier said Eve enabled about 5x more work in the same timespan and automatically responded to more than 30 percent of over one hundred special interrogatories. | 中 | SU017, SU018 |
| CU031 | The visible user map spans intake teams, paralegals, associates, partners, operations staff, and leadership users. | 中 | SU001, SU002, SU003, SU004 |
| CU032 | The visible commercial sponsors are firm owners, managing partners, operations leaders, and legal-technology heads rather than rank-and-file end users. | 中 | SU002, SU003, SU018, SU015 |
| CU033 | Eve’s current expansion story depends on working with existing case-management systems rather than demanding immediate replacement; official materials say it works with existing CMS, LawNext calls it complementary, and James Scott Farrin highlighted an upcoming Jove integration. | 中 | SU001, SU003, SU013, SU015 |
| CU034 | Official and independent June 2026 materials say Eve intake runs 24/7 in 28 languages and can sign qualified clients live on the call, making intake automation a lead-conversion wedge rather than a back-office feature. | 高 | SU001, SU013 |
| CU035 | Official and independent descriptions say Eve’s agents draft work automatically but route it to attorney review before anything is filed or sent. | 高 | SU007, SU011, SU013 |
| CU036 | Customer stories repeatedly frame adoption as consultative and customized: live demos, tailored tools, playbooks, and ongoing feedback loops are central to successful rollout. | 中 | SU002, SU003, SU004, SU005 |
| CU037 | Public evidence supports expansion across several plaintiff practice areas and U.S. regions, but it does not disclose customer concentration, top-account exposure, or channel mix. | 中 | SU001, SU002, SU003, SU004, SU005 |
| CU038 | No public source in the reviewed set discloses NRR, GRR, churn, contract duration, or renewal cohorts for Eve. | 中 | SU001, SU007, SU015 |
| CU039 | Software Finder’s 4.9 rating across 12 verified reviews is a positive satisfaction signal, but the sample is too small and reseller-mediated to stand in for cohort retention proof. | 中 | SU015 |
| CU040 | Legal bar guidance and sanctions commentary show that law-firm AI rollouts still require confidentiality controls, supervision, verification, and fee sensitivity. | 中 | SU020, SU021 |
| CU041 | Change-management friction is real: Mike Morse described fear as the top barrier, and James Scott Farrin said firms have to win over hearts and minds during adoption. | 中 | SU002, SU003 |
| CU042 | Public materials imply best fit with mid-size through enterprise plaintiff firms more than pure self-serve solo buyers because pricing is custom, rollout is consultative, and multiple team roles are involved. | 中 | SU003, SU015, SU016 |
| CU043 | Current public evidence supports broad U.S. coverage plus multilingual intake capability, but it does not support a claim of diversified international customer footprint. | 中 | SU001, SU007, SU013 |
| CR001 | Eve markets itself as the AI operating system for plaintiff law and says it is trusted by 1,200+ firms. | 中 | SR001 |
| CR002 | Eve publicly says it supports intake, medical chronologies, motions, demands, discovery responses, research, and firm analytics. | 高 | SR001, SR002 |
| CR003 | Eve says it works with existing CMS and CRM systems, offers open APIs and custom integrations, and can also stand on its own. | 高 | SR001, SR002 |
| CR004 | Eve publicly claims client data is private, isolated, and privilege-protected, and says its security program follows SOC 2 criteria with annual penetration testing and incident-response procedures. | 高 | SR001, SR003 |
| CR005 | Eve’s medical-overview page says the product is HIPAA compliant, encrypts records in transit and at rest, hosts medical-record workloads on U.S.-based AWS infrastructure, and does not use PHI to train AI models. | 中 | SR007 |
| CR006 | Eve’s enterprise terms give Eve a limited-term license to host, copy, transmit, and display customer data as reasonably necessary to provide the service. | 中 | SR005 |
| CR007 | Eve’s enterprise terms say customer-facing service may be affected by subprocessors and that Eve is not liable for discontinuation of subprocessor hosting services. | 中 | SR005 |
| CR008 | Eve’s SLA promises 99.5% monthly uptime but excludes cloud-provider and integrated-application-provider outage events from downtime calculations. | 中 | SR006 |
| CR009 | Eve’s terms allow suspension of service for significant data-security risk or if court or administrative order requires it. | 中 | SR005 |
| CR010 | ABA Formal Opinion 512 says lawyers using generative AI remain bound by duties of competence, confidentiality, communication, and reasonable fees, and must verify outputs before relying on them. | 高 | SR017, SR018, SR019 |
| CR011 | ABA Model Rule 1.1 requires competent representation and Rule 1.6 requires reasonable efforts to prevent unauthorized disclosure or access to client information. | 高 | SR018, SR019 |
| CR012 | ABA Model Rule 5.5 forbids lawyers from practicing law in a jurisdiction in violation of legal-profession rules or assisting another in doing so. | 中 | SR020 |
| CR013 | California’s 2026 practical guidance says lawyers must not deploy agentic AI so that it makes substantive legal determinations, communicates legal advice, prepares and files pleadings, or otherwise acts in a representative capacity without meaningful lawyer supervision and review. | 高 | SR021, SR020 |
| CR014 | California’s 2026 guidance says the greater the level of system autonomy, the greater the lawyer’s obligation to implement oversight, verification, and periodic reassessment of the system’s risks and capabilities. | 高 | SR021, SR017 |
| CR015 | Eve’s 2026 launch materials say AI Agents advance casework automatically as new information arrives, draft required documents, and manage routine intake and status updates, with work queued for attorney review and approval before it leaves the firm. | 高 | SR039, SR040 |
| CR016 | Because Eve differentiates on autonomous execution inside live plaintiff workflows, its product scope now sits close to the agentic-AI boundary that bar guidance identifies as legally sensitive. | 中 | SR002, SR021, SR039 |
| CR017 | Eve’s drafting pages say demand letters are generated in about 10-15 minutes and discovery-response drafts can reach a servable product in under an hour, but still require lawyer review and refinement. | 高 | SR008, SR009 |
| CR018 | Eve’s discovery workflow warns that privileged memos or work product should be excluded from selected inputs and that attorney review is essential before anything is served. | 高 | SR009, SR013 |
| CR019 | EveOS markets research outputs whose citations can be opened and verified at the source, which shows that citation traceability is central to the company’s risk-mitigation story. | 中 | SR002 |
| CR020 | Mata v. Avianca sanctioned lawyers and their law firm after fake ChatGPT-generated cases were filed and defended, emphasizing that attorneys retain a gatekeeping role over accuracy. | 高 | SR026, SR029 |
| CR021 | In June 2026, the Ninth Circuit sanctioned two lawyers for briefs containing nonexistent cases and misquoted authority, stating that legal papers violate rules at the point of signing and filing, not merely at drafting. | 高 | SR027, SR029 |
| CR022 | The Morgan & Morgan sanctions episode shows that even a law firm’s internal AI tooling does not remove Rule 11 exposure because the duty to verify cited law is nondelegable. | 中 | SR030, SR026 |
| CR023 | The ABA’s 2025 Gauthier coverage shows courts are ordering penalties and AI-related CLE when lawyers fail to confirm that generated citations and quotations actually exist. | 中 | SR029 |
| CR024 | Eve says every medical-record data point is hyperlinked to its source document and page number, which mitigates but does not eliminate factual-hallucination risk. | 中 | SR007 |
| CR025 | Eve’s marketed outputs include medical chronologies, demands, discovery requests and responses, motions, and research, all of which can shape evidentiary positions and malpractice exposure even before a court filing exists. | 中 | SR001, SR008, SR009, SR010 |
| CR026 | Warner v. Gilbarco held that AI-assisted litigation materials were protected work product in that case and that ChatGPT use was not automatically a waiver to an adversary. | 中 | SR028 |
| CR027 | Because work-product outcomes remain fact-specific and public sources do not show one settled rule for AI-assisted privilege, customers cannot assume that product-level privilege marketing will hold across forums and workflows. | 中 | SR028, SR018, SR019 |
| CR028 | HHS says the HIPAA Breach Notification Rule applies to covered entities and business associates after a breach of unsecured PHI. | 中 | SR022 |
| CR029 | HHS says a business associate is a person or entity that creates, receives, maintains, or transmits PHI for a covered entity or another business associate. | 中 | SR023 |
| CR030 | 45 CFR 164.410 requires a business associate to notify the covered entity of a breach without unreasonable delay and no later than 60 calendar days after discovery. | 高 | SR024, SR022 |
| CR031 | Public sources reviewed do not establish whether Eve acts as a HIPAA business associate across all customer workflows, so HIPAA applicability appears to depend on customer-specific data flows and contracting. | 低 | SR007, SR023 |
| CR032 | Even where full HIPAA status is not proven from public materials, Eve plainly processes large volumes of sensitive plaintiff matter data and medical records, so a privacy or confidentiality failure would still create severe contractual, malpractice, and reputational exposure. | 中 | SR005, SR007, SR018, SR019 |
| CR033 | Eve’s public privacy policy references analytics and advertising partners on the website surface, underscoring that marketing-site privacy language is distinct from matter-handling privilege and confidentiality controls. | 中 | SR004 |
| CR034 | Thomson Reuters’ 2025 professional-services survey found 41% of respondents personally use publicly available GenAI tools, 17% personally use industry-specific GenAI tools, and 95% expect GenAI to become central to workflow within five years. | 中 | SR031 |
| CR035 | The same Thomson Reuters work says 52% of respondents report no GenAI policy and 64% report no GenAI training at work. | 高 | SR031, SR032 |
| CR036 | Thomson Reuters’ legal-specific summary says 26% of legal professionals are already using GenAI, 59% of law firms believe it should be applied to their work, and 71% of law-firm clients do not know whether outside firms are using it. | 中 | SR032 |
| CR037 | Thomson Reuters Institute’s 2026 UPL analysis says there is no uniform definition of the practice of law across states and identifies applying law to specific facts and recommending a course of action as the sensitive boundary. | 中 | SR033 |
| CR038 | RELX’s 2025 Form 20-F says its businesses operate in highly competitive and dynamic markets where AI, new competitors, and rapid change can reduce demand if the company fails to adapt quickly. | 高 | SR036, SR037 |
| CR039 | RELX’s 20-F also says data-privacy changes, cybersecurity compromise, failures by cloud, software, and large-language-model providers, and significant platform interruption can all adversely affect business performance. | 高 | SR036, SR037 |
| CR040 | RELX says double-digit growth in its Law Firms & Corporate Legal segment is being driven by Lexis+ AI and Protégé, showing that incumbent legal platforms already have AI traction with law-firm buyers. | 中 | SR036 |
| CR041 | Eve’s January and June 2026 launch materials claim 10x revenue growth and expansion from 500+ to 900+ plaintiff firms, implying rapid operational scaling at the same time autonomous features are broadening. | 高 | SR039, SR040 |
| CR042 | Eve’s public customer proof prominently features large plaintiff firms such as Mike Morse and James Scott Farrin, but public materials do not disclose revenue concentration, NRR, or churn. | 中 | SR015, SR016, SR039 |
| CR043 | Because flagship customer logos serve as both revenue sources and proof points, losing a small number of large reference firms could damage both retention economics and new-customer credibility. | 中 | SR015, SR016, SR039 |
| CR044 | Eve’s own case studies describe AI adoption as requiring governance policies, dedicated AI leadership, training, and explicit reassurances to employees about job-displacement fears, showing customer-side execution friction. | 中 | SR015, SR016 |
| CR045 | Taken together, UPL ambiguity, sanction precedent, sensitive medical-data workflows, platform reliance, and incumbent competition make governance maturity—not mere feature breadth—the key diligence issue for Eve. | 中 | SR021, SR027, SR036, SR039 |
| CR046 | Eve’s enterprise terms require professional liability/errors-and-omissions insurance of $1 million annual aggregate and cyber liability insurance of $3 million per occurrence and in the aggregate. | 中 | SR005 |
| CR047 | Eve’s security page says the company performs vendor reviews, annual risk assessments, least-privilege access control, and quarterly access reviews. | 中 | SR003 |
| CV001 | Eve announced a $103 million Series B at a $1B+ valuation in September 2025. | 高 | SV001, SV002, SV005 |
| CV002 | The Series B was led by Spark Capital with continued participation from Andreessen Horowitz, Lightspeed, and Menlo Ventures. | 高 | SV001, SV002 |
| CV003 | Eve said more than 350 firms partnered with it during 2025 before the Series B announcement. | 中 | SV001 |
| CV004 | PR Newswire said Eve had added over 350 new firms in eight months and reached over 450 customer firms by the Series B announcement. | 中 | SV002 |
| CV005 | At the Series B announcement, Eve said the platform processed more than 200,000 legal cases annually and had helped firms recover over $3.5 billion in settlements and judgments. | 高 | SV001, SV002 |
| CV006 | Eve announced a $47 million Series A in January 2025. | 中 | SV004 |
| CV007 | At the Series A, Eve said more than 100 firms had adopted the product and that revenue had grown 500% year over year. | 中 | SV004 |
| CV008 | As of the report run date, Eve’s homepage says the platform is trusted by 1,200+ firms. | 中 | SV003 |
| CV009 | Eve’s homepage markets a firm-wide workflow spanning intake, pre-litigation, litigation, AI-ready data, and firm intelligence. | 中 | SV003 |
| CV010 | The public record shows a jump from 450+ firms at the September 2025 round to 1,200+ firms on the June 2026 homepage, but it does not disclose whether both figures use the same paid-customer definition. | 中 | SV002, SV003 |
| CV011 | No reviewed public source discloses Eve’s ARR, NRR, gross margin, or cap-table terms. | 中 | SV001, SV002, SV003, SV004, SV005, SV006 |
| CV012 | Clio raised $900 million at a $3 billion valuation in July 2024. | 高 | SV007, SV008 |
| CV013 | At that 2024 round, Clio said ARR was above $200 million and TechCrunch reported use by more than 150,000 legal professionals. | 高 | SV007, SV008 |
| CV014 | By November 2025, Clio closed a $500 million Series G plus a $350 million debt facility at a $5 billion valuation and said the combined business had $400 million ARR. | 高 | SV009, SV030 |
| CV015 | Clio’s disclosed financing implies a sub-15x ARR multiple in 2024 and roughly 12.5x ARR in 2025, both well below Harvey-style AI outlier pricing. | 中 | SV007, SV008, SV009, SV030 |
| CV016 | Clio’s current entry pricing starts at $49 per user. | 中 | SV012 |
| CV017 | MyCase publicly prices at $50, $100, and $130 per user per month across its main annual tiers. | 中 | SV013 |
| CV018 | PracticePanther’s annual pricing runs roughly from $69 to $114 per user per month across its main tiers. | 中 | SV014 |
| CV019 | These transparent incumbent pricing pages imply that general legal-workflow seat costs are modest relative to the ARR burden required to support a $1 billion valuation without enterprise-scale expansion or premium AI pricing. | 中 | SV012, SV013, SV014 |
| CV020 | EvenUp raised $150 million at a $2B+ valuation in October 2025 and said total capital raised reached $385 million. | 高 | SV015, SV016 |
| CV021 | EvenUp says over 2,000 firms use the platform, case volume reached 10,000 cases per week, and ARR doubled year over year. | 高 | SV015, SV016 |
| CV022 | Supio raised a $60 million Series B in April 2025, bringing total funding to $91 million. | 高 | SV017, SV018 |
| CV023 | Supio said ARR had grown 4x since its Series A and that it serves many top personal injury and mass tort firms. | 中 | SV017 |
| CV024 | Filevine announced a $400 million financing in 2025 and said it had nearly 6,000 customers, 100,000 legal professionals, 96% gross retention, and net dollar retention above 120%. | 高 | SV019, SV020 |
| CV025 | Litify says it serves 450+ enterprise customers, 70,000+ legal professionals, and 5M+ cases handled annually. | 中 | SV021 |
| CV026 | LawNext reported that Bessemer acquired a majority stake in Litify in 2023 and that Litify had previously raised $50 million in Series A funding in 2019. | 中 | SV022 |
| CV027 | Harvey raised $200 million at an $11 billion valuation in March 2026. | 高 | SV023, SV024 |
| CV028 | CNBC said Harvey reached $190 million ARR in January 2026, implying an approximately 58x ARR financing multiple at the $11 billion valuation. | 高 | SV023, SV024 |
| CV029 | Intapp reported $146.0 million of quarterly revenue, $146.8 million of cash, and $791.4 million of remaining performance obligations as of March 31, 2026. | 中 | SV025 |
| CV030 | CompaniesMarketCap reported Intapp at $1.78 billion market capitalization in June 2026. | 中 | SV026 |
| CV031 | Using annualized March 2026 revenue, Intapp traded at roughly 3.0x market-cap-to-revenue in June 2026, or about 2.8x if cash is netted and debt is ignored. | 中 | SV025, SV026 |
| CV032 | SaaS Capital says the median public SaaS ARR multiple stood at 6.7x in June 2025 and that the 2023-2025 public-market regime was generally bounded in the 6-8x ARR range. | 中 | SV027 |
| CV033 | The ABA says the U.S. lawyer population reached 1.37 million in 2025. | 中 | SV028 |
| CV034 | Clio’s 2026 personal injury statistics page says the U.S. has over 135,000 personal injury lawyers and nearly 400,000 personal injury claims filed annually. | 中 | SV011 |
| CV035 | Thomson Reuters and Georgetown describe 2025 legal-market profits as strong but built on unstable ground, with rising technology spending and contraction risk emerging by mid-2026. | 中 | SV029 |
| CV036 | A $1 billion valuation would require about $149 million of ARR to line up with a 6.7x public SaaS median multiple. | 中 | SV027 |
| CV037 | A $1 billion valuation would require about $80 million of ARR at Clio’s 12.5x 2025 financing multiple and about $67 million at Clio’s sub-15x 2024 financing multiple. | 中 | SV007, SV008, SV009, SV030 |
| CV038 | A $1 billion valuation would require about $333 million of ARR to match Intapp’s roughly 3x public-market revenue ratio. | 中 | SV025, SV026 |
| CV039 | If Eve had 450 paying firms at the September 2025 round and blended ARR per firm was $20,000, ARR would be roughly $9 million and the implied valuation multiple would be about 111x. | 中 | SV002, SV012, SV013, SV014 |
| CV040 | If Eve had 450 paying firms at the round and blended ARR per firm was $40,000, ARR would be about $18 million and the implied multiple about 56x. | 中 | SV002, SV012, SV013, SV014 |
| CV041 | Even at $80,000 blended ARR per round-time customer, 450 firms would imply only about $36 million ARR and a roughly 27.8x multiple. | 中 | SV002, SV012, SV013, SV014 |
| CV042 | If the current 1,200+ firm homepage claim mostly reflects paying firms at roughly $50,000 blended ARR per firm, implied ARR could approach $60 million and compress the prior $1 billion price to about 16.7x ARR. | 中 | SV003, SV012, SV013, SV014 |
| CV043 | The public record therefore supports a view that Eve’s $1B+ round priced a scarcity premium on future category leadership rather than a disclosed revenue multiple. | 中 | SV001, SV002, SV015, SV023, SV027 |
| CV044 | The positive thesis is that plaintiff legal workflows are a real vertical market with large lawyer and claim volume, and AI budgets are rising even as firms demand operational ROI. | 中 | SV011, SV028, SV029 |
| CV045 | The anti-thesis is that public software multiples are sober, Clio’s disclosed financing was supported by far more ARR than Eve has publicly shown, and Eve still provides no audited revenue or preference disclosure. | 中 | SV007, SV008, SV009, SV025, SV027 |
| CV046 | A credible bull case requires Eve to scale to roughly $250-300 million of exit ARR with 12-14x exit pricing, supporting about $3.0-4.2 billion of equity value. | 中 | SV015, SV023, SV027, SV029 |
| CV047 | A base case of roughly $120-160 million exit ARR at 7-9x would support only about $0.84-1.44 billion of value, which is weak return math from a $1 billion entry. | 中 | SV027, SV029 |
| CV048 | A bear case of roughly $60-90 million exit ARR at 4-6x would support only about $0.24-0.54 billion of value and would impair capital. | 中 | SV027, SV029 |
| CV049 | Public evidence still does not reveal Eve’s liquidation preferences, secondary pricing, dilution overhang, paid-customer definition, retention, or gross margin. | 中 | SV001, SV002, SV003, SV011 |
| CV050 | The evidence-backed stance is that the September 2025 $1B+ post-money was stretched for new money and is only more defensible if current paid ARR is already materially above $60-80 million with strong retention. | 中 | SV003, SV007, SV009, SV027 |
| CV051 | At a still-premium 10x ARR threshold, a $1 billion valuation needs about $100 million ARR, highlighting the gap between public proof and the round price. | 中 | SV027 |
| CV052 | Without private diligence that closes the ARR and cap-table gaps, a more prudent new-money underwriting range is roughly $700-800 million rather than $1 billion. | 低 | SV027, SV029 |
| CV053 | Thesis-break triggers are paid ARR materially below $60 million by 2026, retention below software-leader levels, customer counts that prove heavily pilot-weighted, or a later financing that adds heavy preference overhang. | 中 | SV003, SV027, SV029 |
| 编号 | 出版方 | 标题 | 引文 |
|---|---|---|---|
| SO001 | Eve | Legal AI Software for Plaintiff Law Firms | Eve | |
| SO002 | Eve | About Eve | Legal AI Company for Plaintiff Law Firms | |
| SO003 | Eve | Eve Legal | Press & Media Coverage | |
| SO004 | Eve | Introducing Eve | |
| SO005 | Eve | The Dawn of a New Era in Plaintiff Law: Eve Secures $47M Series A | |
| SO006 | Eve | The Next Chapter in AI Transformation for Law Firms | |
| SO007 | Eve | Security and Compliance | The Eve platform is secure at all levels, preserving the confidentiality of your valuable work. |
| SO008 | Eve | Privacy Policy | You understand that Butler Labs owns the Services. |
| SO009 | Eve | Customer Stories | Legal AI Case Studies for Plaintiff Firms | Eve | |
| SO010 | Eve | How Mike Morse Law Firm made Eve indispensable | |
| SO011 | Eve | How Barrett & Farahany uses Eve to deliver justice | |
| SO012 | Eve | How James Scott Farrin rolled out AI to 300 employees | |
| SO013 | Eve | How Hershey Law built a different kind of firm with Eve | |
| SO014 | Eve | The first AI auditor for plaintiff law | |
| SO015 | PR Newswire | Eve Raises $103 Million at $1 Billion Valuation to Help Plaintiff Firms Deliver Justice Through AI Transformation | |
| SO016 | Tech Funding News | Legal AI startup Eve joins unicorn club with $103M to empower plaintiff law firms | |
| SO017 | LawNext | Eve, AI-Driven Platform for Plaintiff-Side Law Firms, Raises $103 Million in Series B Round | |
| SO018 | Legal IT Insider | Plaintiff legal AI startup Eve raises $103m Series B at a $1bn valuation | |
| SO019 | SiliconANGLE | Eve raises $103M to make law firms more efficient with AI | |
| SO020 | Andreessen Horowitz | Investing in Eve | |
| SO021 | Lightspeed Venture Partners | Eve’s Series B: Another Milestone as Eve Rapidly Transforms Legal Firms with AI | |
| SO022 | LawNext | Eve Launches AI Workforce to Reshape the PI Firm Org Chart | |
| SO023 | LawNext | Eve Builds on AI Workforce Launch with EveOS, An AI-Native Operational Platform for Plaintiff Firms | |
| SO024 | LawNext | LawNext: Eve CEO Jay Madheswaran on Building AI-Native Law Firms for the Plaintiffs’ Bar | |
| SO025 | Above the Law | Meet Eve — The AI Used By 800+ Top Plaintiff Firms | |
| SO026 | LLRX | Tracking hallucination marketing claims from legal tech vendors | Earlier this year, I began using an AI program called EVE to help me draft pleadings. |
| SO027 | U.S. District Court exhibit file | Rushing v. Turner apology letter exhibit | Earlier this year, I began using an AI program called EVE to help me draft pleadings. |
| SO028 | Stateline | As AI-generated fake content mars legal cases, states want guardrails | |
| SM001 | Eve | Eve Agents | AI Teammates for Plaintiff Law Firms | Eve Agents support the full range of plaintiff law, including personal injury, workers’ compensation, medical malpractice, labor and employment, mass torts, Social Security disability, and others. |
| SM002 | Eve | Legal AI Guides for Plaintiff Firms | Eve describes in-depth resources on using AI across the full case lifecycle — from intake to resolution. |
| SM003 | Eve | How Plaintiff Law Firms Convert Leads to Cases with AI Intake | AI-driven intake resulted in a 40% increase in client conversions, with response times under 30 seconds. |
| SM004 | Eve | Introducing Eve 2.0: The Proactive AI Workforce for Plaintiff Firms | Eve 2.0 brings Agents, Auditor, and Analyst to your law firm — three always-on AI teammates that fundamentally change the way plaintiff firms operate. |
| SM005 | Eve | About Eve | Legal AI Company for Plaintiff Law Firms | Eve says plaintiff law firms are burdened by tedious tasks and rising costs. |
| SM006 | American Bar Association | ABA Profile of the Legal Profession | The number of lawyers in the U.S. rose to 1.37 million in 2025. |
| SM007 | Clio | 2026 Personal Injury Law Statistics: What the Data Reveals | Of the over 1.3 million lawyers in the United States, over 135,000 are personal injury lawyers — roughly 10% of all practicing attorneys. |
| SM008 | Clio | Personal Injury Software | Medical Record & Settlement Management | Clio positions personal-injury software around medical record and settlement management. |
| SM009 | Smokeball | Personal Injury Workflows | Sample PI workflows include getting settlement funds, creating a disbursement sheet, and receiving the signed disbursement sheet from the client. |
| SM010 | CloudLex | Streamlined Legal Workflows in Action | CloudLex frames PI legal workflows around client intake, medical record retrieval, legal document drafting, and personal injury case management. |
| SM011 | CasePeer | The Ultimate Personal Injury Paralegal Guide: Checklists and Workflow Tips | CasePeer says PI paralegals manage everything from intake to post-settlement work. |
| SM012 | Paxton | A Roadmap for Personal Injury Litigation: Best Practices from Intake to Settlement | Paxton describes PI litigation as moving from intake and case evaluation to final settlement, with early collection of police reports, medical records, photos, and witness statements. |
| SM013 | Thomson Reuters Institute | Practice Innovations: Large Law Firms Embrace Contingent Fee Litigation | When contingent-fee cases were successful for plaintiffs, the generally small law firms made substantial amounts of money. |
| SM014 | Thomson Reuters Institute | 2026 Report on the State of the US Legal Market: Peak prosperity and the fault lines below | Technology spending and talent costs are rising rapidly, with firms aggressively investing in AI capabilities. |
| SM015 | Wolters Kluwer | The Wolters Kluwer Future Ready Lawyer Report: Building confidence in an AI era | Over 90% of respondents report using at least one AI tool in their daily workflow. |
| SM016 | American Bar Association Law Practice Magazine | AI for Law Firms: What the 8am Legal Industry Report Tells Us About AI Use | Legal AI adoption has more than doubled year over year, but many firms still lack formal policies, structured training, and governance frameworks. |
| SM017 | LexisNexis | Lex Machina 2025 Torts Litigation Report | LexisNexis Newsroom | Lex Machina says premises liability and motor vehicle cases reached record levels. |
| SM018 | National Center for State Courts | Data | NCSC says its caseload data is sourced directly from state courts and helps court systems monitor changing patterns. |
| SM019 | Federal Judicial Center | Federal Court Cases: FJC Integrated Database 1979 to Present | The FJC Integrated Database is a primary federal-cases resource that can be used for class-action and MDL analysis. |
| SM020 | United States Courts | Civil | The U.S. Courts civil data page includes current product-liability and civil-cases-filed tables through 2025. |
| SM021 | U.S. District Court for the District of South Carolina | SCD - MDL | The AFFF MDL has approximately 10,000+ associated cases and tens of thousands of plaintiffs. |
| SM022 | Duane Morris | Duane Morris Class Action Review - 2025: A Comprehensive Analysis of Class Action Litigation | Duane Morris reviewed more than 1,441 class action decisions in the past year. |
| SM023 | American Bar Association | ABA Issues First Ethics Guidance on Use of Gen AI Tools | Formal Opinion 512 mandates that lawyers adhere to ethical standards when using generative AI. |
| SM024 | National Conference of Bar Examiners | Generative Artificial Intelligence Tools: ABA Formal Opinion 512 Provides Needed Guidance on the Benefits and Burdens of Lawyers’ Use of GAI | The ethical issues discussed relate to lawyers’ duties of competence, confidentiality, communication, supervision, candor, and duty to charge reasonable fees. |
| SM025 | New York State Bar Association | Avoiding Sanctions in the Gen AI Era: Practical Guardrails for Lawyers | Hallucinations continue to generate sanctions or other disciplinary consequences for counsel. |
| SM026 | Thomson Reuters | ABA ethics rules related to Generative AI | Lawyers still have to verify the cases cited in documents filed with the court and should treat AI like an assistant, not a lawyer. |
| SM027 | American Bar Association | Rule 1.15: Safekeeping Property | Funds shall be kept in a separate account maintained in the state where the lawyer’s office is situated. |
| SM028 | American Bar Association | ABA Model Rules on Client Trust Account Records - Rule 1 Recordkeeping Generally | A lawyer shall maintain current financial records and retain them for five years after termination of the representation. |
| SM029 | American Bar Association | Rule 1.5: Fees | A contingent fee agreement shall be in a writing signed by the client and shall state the method by which the fee is to be determined. |
| SM030 | American Association for Justice | About Us | AAJ has members worldwide and a network of U.S. and Canadian affiliates involved in diverse areas of trial advocacy. |
| SP001 | Eve | Legal AI Software for Plaintiff Law Firms | Eve | EveOS is the AI operating system for plaintiff law. Every case, attorney, and dollar, in one place. Trusted by 1200+ firms. |
| SP002 | EvenUp | Leading AI for Personal Injury Lawyers - EvenUp Law | The First Proactive AI Platform for the Entire Case Lifecycle. |
| SP003 | Supio | Supio | Legal AI for Personal Injury Law Firms | Supio is the only agentic legal AI platform built for plaintiff law and mass torts cases. |
| SP004 | Supio | Eve Legal Alternatives: Supio vs Eve (2026) | Legal AI for Personal Injury | Supio is the only platform that delivers all of this—with 97% verified accuracy. |
| SP005 | Supio | EvenUp Alternatives: Supio vs EvenUp (2026) | Full-Lifecycle Legal AI | Predictable, flat pricing—no page limits, no demand-type tiers, no edit penalties. |
| SP006 | Supio | Supio Announces $60M Series B to Accelerate Adoption of Legal AI in Plaintiff Law | The round was led by existing investor Sapphire Ventures, with participation from new investors Mayfield and Thomson Reuters Ventures. The new investment brings Supio's total funding to date to $91 million. |
| SP007 | Darrow | Darrow AI | Darrow is the leader in Legal Exposure Management - working upstream to surface hidden signals and transform them into structured legal intelligence, before harm compounds. |
| SP008 | Plaintifi Lawyers | Plaintifi Lawyers | Connects individuals with personal injury cases to vetted attorneys, offering a fast, free, and no-obligation matching service. |
| SP009 | ProPlaintiff | AI Legal Software for Personal Injury Law Firms - Agentic Case Management | From intake to settlement, ProPlaintiff automates the operational work that slows your firm down — without losing control. |
| SP010 | LawPro.ai | LawPro.ai - Legal AI Platform for Personal Injury Lawyers | Instant, citation-backed answers from the timeline, for demands, depositions, and adjuster calls. |
| SP011 | Filevine | Features | Filevine | FilevineAI analyzes your uploaded medical records to surface what matters most - automatically. |
| SP012 | Clio | All Features | From automating everyday tasks to turning information into actionable insights, Manage AI (formerly Clio Duo) takes care of it so you can focus on more impactful work. |
| SP013 | Clio | Clio Legal Software Plans & Pricing | Starting at $49/user. |
| SP014 | MyCase | Best Law Practice Management Software | Start Free Today | MyCase | Manage firm cases from client intake to billing and everything in between. |
| SP015 | MyCase | MyCase Pricing | Start Your 10-Day Free Trial Today | MyCase | MyCase Basic plan is $50 per user/month (billed annually) or $60 per user/month (billed monthly). |
| SP016 | PracticePanther | Intuitive Law Firm Software | PracticePanther | Standardize your intake, workflows, and billing process so you never miss a step. |
| SP017 | PracticePanther | PracticePanther Pricing | Start for as Low as $49/Month | PracticePanther Pricing | Start for as Low as $49/Month. |
| SP018 | Litify | Legal Software: The Leading Platform of Action | Litify | Accelerate growth and maximize outcomes with legal intelligence that’s embedded in every field, button, action, and workflow. |
| SP019 | Litify | Legal Solutions For Every Practice | Litify | Secure the best outcomes for injured clients and get from intake to settlement faster with configurable intake questionnaires, automated matter plans, and robust analytics that track case progression, key deadlines, and medical record requests. |
| SP020 | CloudLex | The Complete Personal Injury Management Software | CloudLex | Complete practice management built exclusively for personal injury firms—from intake through settlement. |
| SP021 | CasePeer | Best Personal Injury Case & Practice Management Software | CasePeer is the only case management software driving better outcomes for your clients, and better outcomes for your firm. |
| SP022 | Smokeball | Features | Features. |
| SP023 | Harvey | Harvey | AI software for legal and professional services | Purpose built agents execute complex legal work end to end. |
| SP024 | CNBC | Legal AI startup Harvey valued at $11 billion in funding round, as VCs spread bets beyond model companies | Founded in 2022, Harvey offers AI tools for legal and professional services that can streamline contract analysis, compliance, due diligence and litigation. |
| SP025 | Thomson Reuters | CoCounsel Legal - AI Legal Assistant | Thomson Reuters | One query. Conversational AI. Expert-level output. |
| SP026 | LexisNexis | Lexis+ with Protégé | Legal AI Solution for Drafting & Research | Lexis+ with Protégé is a legal AI solution built for drafting, research, and analysis. |
| SP027 | Paxton | PAXTON | The all-in-one AI legal assistant | Experience the leading AI for lawyers. Rapidly conduct research, accelerate drafting, and analyze documents with Paxton. |
| SP028 | Artificial Lawyer | Plaintiff Bar AI Takes Off – EvenUp Bags $150m | The funding brings EvenUp’s total capital raised to $385m. |
| SP029 | Crunchbase News | Investors Have Poured Billions Into Plaintiff-Side Legal AI, But Defense Could Be The Next Big Opportunity | EvenUp has raised $370 million, Eve $164 million, Supio $85 million, and Darrow $63 million, for a combined total of roughly $682 million. |
| SP030 | LawNext | Legal AI Platform Harvey To Get LexisNexis Content and Tech In New Partnership Between The Companies | The generative AI legal startup Harvey has entered into a strategic alliance with LexisNexis Legal & Professional by which it will integrate LexisNexis’ gen AI technology, primary law content, and Shepard’s Citations within the Harvey platform and jointly develop advanced legal workflows. |
| SI001 | Eve | Legal AI Software for Plaintiff Law Firms | Eve | Don't just manage your firm. Master it. |
| SI002 | Eve | The Dawn of a New Era in Plaintiff Law: Eve Secures $47M Series A | We’re beyond thrilled to announce a $47 million Series A funding round led by Andreessen Horowitz, with support from Lightspeed and Menlo Ventures. |
| SI003 | Eve | The Next Chapter in AI Transformation for Law Firms | Eve has raised a record-setting $103M in Series B funding, with a $1B+ valuation. |
| SI004 | PR Newswire | Eve Raises $103 Million at $1 Billion Valuation to Help Plaintiff Firms Deliver Justice Through AI Transformation | Eve has raised $103 million in Series B funding at over a $1 billion valuation. |
| SI005 | LawNext | Eve Launches 'AI Workforce' to Reshape the PI Firm Org Chart: Not Replacing Lawyers, But Freeing them For Higher-Level Work | LawSites | |
| SI006 | LawNext | Eve Builds on AI Workforce Launch with EveOS, An AI-Native Operational Platform for Plaintiff Firms | LawSites | |
| SI007 | Eve | Legal AI Careers | Join Eve's Team | We’ve raised $164M+ from leading investors to bring Eve’s cutting-edge AI to plaintiff firms. |
| SI008 | Eve | AI Platform for Plaintiff Law Firms | EveOS | |
| SI009 | Eve | Why Plaintiff Firms Choose Eve | Legal AI Platform | |
| SI010 | Eve | AI Intake | Eve | |
| SI011 | Eve | The End of the Intake Bottleneck: Eve AI Intake is Now Generally Available | After a highly successful beta program with over 40 plaintiff firms, Eve AI Intake is now available. |
| SI012 | Eve | Best AI Tools for Personal Injury Lawyers (2026) | Eve Legal | |
| SI013 | Eve | PI Lawyer Marketing in 2026: 7 Levers That Fill Your Caseload | |
| SI014 | Eve | Scaling smarter: Archuleta Law Firm’s path to AI-powered intake | We're seeing roughly 50% of inbound callers voluntarily select the AI receptionist when given the option for expedited service. |
| SI015 | Eve | "Get Used to It": How Laurel Employment Law Went AI-Native from Day One | Twenty-four months later, Laurel has 100 employees across five continents and more than 1,500 active clients. |
| SI016 | Eve | Off to the Races: How Laurel Employment Law Automates Demand Drafting with AI Agents | After implementing Eve’s Demand Drafting Agent, the firm’s demand output increased from 48 to 104 demand letters mailed out in a week. |
| SI017 | Eve | Customer Story - Frontier Law Center | What would have been several days of work, we completed in 45 minutes. |
| SI018 | Eve | Candidature à un emploi de Head of Legal chez Eve | Explosive growth: We are growing 2X revenue Quarter over Quarter. |
| SI019 | Eve | Job Application for ML Engineer at Eve | Product-market fit: Eve is trusted by over 1000+ law firms, and we’re growing fast. |
| SI020 | Built In | Eve Careers, Perks + Culture | Built In | |
| SI021 | Jobera | Eve Careers | Remote, Hybrid | 34 Open Positions | June 2026 | 34 open positions |
| SI022 | Clio | Clio Pricing | Plans for Every Law Firm | Try it for Free Now | Starting at $49/user |
| SI023 | MyCase | MyCase Pricing | Start Your 10-Day Free Trial Today | MyCase | MyCase Basic plan is $50 per user/month ... MyCase Pro plan is $100 ... Advanced plan is $130. |
| SI024 | PracticePanther | PracticePanther Pricing | Start for as Low as $49/Month | Solo $49 ... Essential $69 ... Business $89 ... Business Pro $114 |
| SI025 | Securities and Exchange Commission / Intapp | Document | Cloud annual recurring revenue was $459.3 million ... cloud net revenue retention rate ... 123%. |
| SI026 | Business Wire / DISCO | DISCO Announces First Quarter 2026 Financial Results | DISCO grew to 347 customers with revenue in excess of $100,000 over the previous 12-month period. |
| SI027 | B17 News | The blame game over AI hallucinations in court filings has started - B17 News | If their software starts to embarrass customers in court, that trust erodes. |
| SI028 | EDRM | The AI Sanction Wave: $145K in Q1 Penalties Signals Courts Have Lost Patience with GenAI Filing Failures - EDRM | In the first quarter of 2026, U.S. courts imposed at least $145,000 in sanctions for fabricated citations. |
| SI029 | Rushing v. Turner filing letter | March 27 2026 Rushing v. Turner responsibility letter | I write to take responsibility for an error made in my briefings in the Rushing v. Turner case. |
| SI030 | Andreessen Horowitz | Investing in Eve | Eve acts as an intelligent partner for plaintiff attorneys, helping them take on 3-4x the number of cases. |
| SI031 | Lightspeed Venture Partners | Eve's Series B: Another Milestone as Eve Rapidly Transforms Legal Firms with AI - Lightspeed Venture Partners | Since its Series A, Eve has grown to add over 450 customers. |
| SI032 | American Bar Association | AI for Law Firms: What the 8am Legal Industry Report Tells Us About AI Use | Legal AI adoption has more than doubled year over year. |
| SE001 | Eve | Legal AI Software for Plaintiff Law Firms | Eve | |
| SE002 | Eve | Legal AI for Medical Overviews | Eve | |
| SE003 | Eve | AI for Legal Discovery: A Plaintiff Firm’s Guide to Drafting, Responding, and Review | |
| SE004 | Eve | Eve <> Clio Integration | Get on the waitlist | |
| SE005 | Eve | Security and Compliance | |
| SE006 | Eve | Privacy Policy | |
| SE007 | Eve | Eve Agents | AI Teammates for Plaintiff Law Firms | |
| SE008 | Eve | Introducing Eve 2.0: The Proactive AI Workforce for Plaintiff Firms | |
| SE009 | Eve | The End of the Intake Bottleneck: Eve AI Intake is Now Generally Available | |
| SE010 | Eve | Why Plaintiff Firms Choose Eve | Legal AI Platform | |
| SE011 | Eve | Legal AI Careers | Join Eve's Team | |
| SE012 | Eve | Off to the Races: How Laurel Employment Law Automates Demand Drafting with AI Agents | |
| SE013 | Eve | From Paper Intakes to a $27.5 Million Verdict: How Hershey Law Built a Different Kind of Firm with Eve | |
| SE014 | LawNext / LawSites | Eve Launches ‘AI Workforce’ to Reshape the PI Firm Org Chart: Not Replacing Lawyers, But Freeing them For Higher-Level Work | |
| SE015 | LawNext / LawSites | Eve Builds on AI Workforce Launch with EveOS, An AI-Native Operational Platform for Plaintiff Firms | |
| SE016 | PR Newswire | Eve Launches AI Workforce to Reshape the Law Firm Org Chart: Humans for Strategy, Agents for Execution | |
| SE017 | Morningstar / Business Wire | Eve Launches EveOS, the AI-Native Operating System Transforming Plaintiff Law Firms | |
| SE018 | Artificial Lawyer | Walk Through: Eve – AI-Driven Client Intake | |
| SE019 | Above the Law | Meet Eve — The AI Used By 800+ Top Plaintiff Firms | |
| SE020 | LLRX | Tracking hallucination marketing claims from legal tech vendors – LLRX | |
| SE021 | EDRM | The AI Sanction Wave: $145K in Q1 Penalties Signals Courts Have Lost Patience with GenAI Filing Failures | |
| SE022 | TechNewsWorld | Law Firms Grapple With Hallucinated Legal Logic, Shadow AI | |
| SE023 | LawAccounting | Q1 2026 AI Hallucination Sanctions Just Crossed $145,000 — Why Mid-Market Law Firms Need a Citation Audit Workflow Inside Their Practice Platform, Not Bolted On | |
| SE024 | U.S. District Court filing exhibit | Rushing v. Turner USA apology letter exhibit | |
| SE025 | Greenhouse / Eve | ML Engineer | |
| SE026 | Built In | Eve Careers, Perks + Culture | |
| SE027 | Eve engineering blog | WALL-E at Eve: An Always-On Agent That Actually Ships | |
| SE028 | Eve engineering blog | Why we chose Turbopuffer for our search infrastructure | |
| SU001 | Eve | Legal AI Software for Plaintiff Law Firms | Eve | EveOS is the AI operating system for plaintiff law. Every case, attorney, and dollar, in one place. Trusted by 1200+ firms. |
| SU002 | Eve | "They'd quit if I took it away": How Mike Morse Law Firm made Eve indispensable | I would say my capacity has doubled at minimum — maybe even tripled compared to what I was doing before I had Eve. |
| SU003 | Eve | How James Scott Farrin rolled out AI to 300 employees — and made them ask for more | Medical summaries that used to take weeks — we're talking 30 minutes, maybe an hour on a big case. |
| SU004 | Eve | From Paper Intakes to a $27.5 Million Verdict: How Hershey Law Built a Different Kind of Firm with Eve | Every team at the firm uses Eve. Operations. Intake. Pre-litigation. Litigation. |
| SU005 | Eve | How Jeffrey Glassman turned its toughest critics into AI champions | We started with Eve probably three or four months ago. I saw we are at about 90% of our demands now going out through Eve. |
| SU006 | Eve | AI-Native Law Firms | Legal AI for Plaintiff Firms | Eve | Trusted by 1000 elite plaintiff lawyers to deliver exceptional results. |
| SU007 | Eve | Eve Surpasses 1,000 Plaintiff Law Firm Customers | More than 1,000 plaintiff law firms now run their caseloads on Eve, with over 200,000 active cases on the platform. |
| SU008 | Eve | Eve 2.0 Early Access | Ready to go AI-Native in 90 days or less? |
| SU009 | TMCnet | Eve Surpasses 1,000 Plaintiff Law Firm Customers, Powers Over 200,000 Cases | Eve today announced it has surpassed 1,000 plaintiff law firm customers and powered more than 200,000 cases on its platform. |
| SU010 | PR Newswire | Eve Surpasses 1,000 Plaintiff Law Firm Customers, Powers Over 200,000 Cases | Customers are seeing measurable impact. Mike Morse Law Firm has increased attorney capacity by 2–3x. |
| SU011 | LawNext | Eve Launches ‘AI Workforce’ to Reshape the PI Firm Org Chart: Not Replacing Lawyers, But Freeing them For Higher-Level Work | The company says it has achieved 10x revenue growth over the past year and now serves more than 500 plaintiff firms. |
| SU012 | Above the Law | Meet Eve — The AI Used By 800+ Top Plaintiff Firms | Before, his team would only convert 10% of inquiries into cases. Now, they convert 35% of all leads. |
| SU013 | LawNext | Eve Builds on AI Workforce Launch with EveOS, An AI-Native Operational Platform for Plaintiff Firms | Eve says it now serves more than 1,400 plaintiff law firms with more than 200,000 active matters on the platform. |
| SU014 | FinancialContent | Eve Launches EveOS, the AI-Native Operating System Transforming Plaintiff Law Firms | Eve Launches EveOS, the AI-Native Operating System Transforming Plaintiff Law Firms. |
| SU015 | Software Finder | Eve Legal: Reviews, Pricing & Free Demo | Eve Legal offers integration with Clio. Users can schedule an Eve Legal demo to learn more about its integration. |
| SU016 | ProPlaintiff.ai | Eve Legal Pricing Explained: Features, Tiers, and How It Compares | Eve Legal does not publish pricing tiers publicly. To see real numbers, your firm has to schedule a demo and speak with sales. |
| SU017 | Eve | Customer Story - Frontier Law Center | What would have been several days of work, we completed in 45 minutes. |
| SU018 | Frontier Law Center | Frontier Law Center and Eve Launch Groundbreaking AI-Native Law Firm | By infusing AI throughout our processes, we can analyze cases faster and streamline our workflows, often 5-10X our abilities. |
| SU019 | PR Newswire | Frontier Law Center Taps Eve to Become the First AI-Native Law Firm | Frontier Law Center Taps Eve to Become the First AI-Native Law Firm. |
| SU020 | New York State Bar Association | Avoiding Sanctions in the Gen AI Era: Practical Guardrails for Lawyers | Generative AI can be a helpful drafting and research aid, but it does not, and cannot, displace the lawyer’s nondelegable duties of competence, supervision and candor. |
| SU021 | New York City Bar Association | Formal Opinion 2024-5: Generative AI in the Practice of Law | NYC Bar | When using generative artificial intelligence tools, a lawyer should take into account the duty of confidentiality and the duty of competence and diligence. |
| SU022 | Mike Morse | Mike Morse Attorney, Entrepreneur, Visionary, Educator | Mike Morse Law Firm is the largest personal injury law firm in Michigan. |
| SU023 | James Scott Farrin | James Scott Farrin | Injury Lawyers | North and South Carolina | Our 60+ attorneys have helped us recover +$2 billion total for +78,000 people since our firm began. |
| SU024 | Hershey Law | Leading Employment Lawyer in California | Hershey Law | We stand with California employees ... throughout the state. |
| SU025 | Jeffrey Glassman Injury Lawyers | Boston Personal Injury Lawyers | Massachusetts Accident Attorney | We have recovered hundreds of millions of dollars and helped over 25,000 injured clients in Boston and beyond. |
| SR001 | Eve | Legal AI Software for Plaintiff Law Firms | Eve | |
| SR002 | Eve | Meet EveOS | |
| SR003 | Eve | Security and Compliance | |
| SR004 | Eve | Privacy Policy | |
| SR005 | Eve | Master Service Agreement | The Customer understands that the Services are hosted by subprocessors of Eve, and such subprocessors may reserve rights to discontinue their hosting. Eve shall not be liable in any way for any discontinuation of such services provided by subprocessors. |
| SR006 | Eve | Butler Labs, Inc. SLA | |
| SR007 | Eve | Medical Overview | Yes. Eve is HIPAA compliant and SOC 2 Type 2 certified. Medical records are encrypted in transit and at rest, hosted exclusively on AWS servers in the United States, and isolated so nothing crosses between matters or firms. |
| SR008 | Eve | Demand Letters | |
| SR009 | Eve | Responding to Discovery | |
| SR010 | Eve | Propounding Discovery | |
| SR011 | Eve | AI Auditor for Plaintiff Law | |
| SR012 | Eve | 7 Things to Consider When Writing a Responsible AI Use Policy for Your Law Firm | |
| SR013 | Eve | Disclosing AI Usage to Your Clients: Best Practices for Legal Teams | |
| SR014 | Eve | ChatGPT Making Clients Think They’re Lawyers | |
| SR015 | Eve / Mike Morse Law Firm | Mike Morse Case Study | |
| SR016 | Eve / Law Offices of James Scott Farrin | Law Offices of James Scott Farrin Case Study | |
| SR017 | American Bar Association | ABA Formal Opinion 512 — Generative Artificial Intelligence Tools in Legal Practice | Lawyers must evaluate the risks and benefits of the specific GAI tool they intend to use, taking reasonable efforts to prevent the inadvertent or unauthorized disclosure of client information, and verify outputs before relying on them. |
| SR018 | American Bar Association | Model Rule 1.1 — Competence | |
| SR019 | American Bar Association | Model Rule 1.6 — Confidentiality of Information | |
| SR020 | American Bar Association | Model Rule 5.5 — Unauthorized Practice of Law; Multijurisdictional Practice of Law | |
| SR021 | State Bar of California COPRAC | Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law | Lawyers must not deploy agentic systems in a manner that allows the system to make substantive legal determinations, communicate legal advice, prepare and file pleadings, or otherwise act in a representative capacity without meaningful lawyer supervision and review. |
| SR022 | U.S. Department of Health and Human Services | Breach Notification Rule | |
| SR023 | U.S. Department of Health and Human Services | Business Associates | |
| SR024 | Electronic Code of Federal Regulations | 45 CFR 164.410 — Notification by a business associate | |
| SR025 | National Institute of Standards and Technology | Artificial Intelligence Risk Management Framework (AI RMF 1.0) | |
| SR026 | U.S. District Court, S.D.N.Y. | Mata v. Avianca, Inc. — Opinion and Order on Sanctions | But existing rules impose a gatekeeping role on attorneys to ensure the accuracy of their filings. |
| SR027 | U.S. Court of Appeals for the Ninth Circuit | Lnu v. Blanche — Order Imposing Sanctions for AI Hallucinations | However legal papers are prepared, and however legal technology develops, the court’s procedural and ethical rules apply with equal force. |
| SR028 | U.S. District Court, E.D. Michigan | Warner v. Gilbarco, Inc. et al. — Order on Discovery Motions | And ChatGPT (and other generative AI programs) are tools, not persons, even if they may have administrators somewhere in the background. |
| SR029 | American Bar Association Litigation Section | Lawyer Sanctioned for Failure to Catch AI “Hallucination” | |
| SR030 | LawNext | Federal Judge Sanctions Morgan & Morgan Attorneys for AI-Generated Fake Cases in Court Filing | |
| SR031 | Thomson Reuters | 2025 Generative AI in Professional Services Report | |
| SR032 | Thomson Reuters Institute | 2025 GenAI report: Executive summary for legal professionals | |
| SR033 | Thomson Reuters Institute | How AI-powered access to justice is impacting unauthorized practice of law | |
| SR034 | Thomson Reuters Institute | Scaling Justice: Unauthorized practice of law and the risk of AI over-regulation | |
| SR035 | Wolters Kluwer Legal & Regulatory | The 2024 Future Ready Lawyer Survey | |
| SR036 | Securities and Exchange Commission / RELX PLC | RELX PLC Form 20-F for fiscal year ended December 31, 2025 | |
| SR037 | RELX | Publication of 2025 Annual Report and Annual Report on Form 20-F | |
| SR038 | Thomson Reuters | Thomson Reuters Files 2025 Annual Report | |
| SR039 | Eve | Eve Expands Plaintiff AI Workforce Platform to General Availability Following Adoption by 900 Plaintiff Law Firms | |
| SR040 | Eve | Eve Launches AI Workforce to Reshape the Law Firm Org Chart: Humans for Strategy, Agents for Execution | |
| SV001 | Eve | The Next Chapter in AI Transformation for Law Firms | Today, I’m proud to share that Eve has raised a record-setting $103M in Series B funding, with a $1B+ valuation. |
| SV002 | PR Newswire | Eve Raises $103 Million at $1 Billion Valuation to Help Plaintiff Firms Deliver Justice Through AI Transformation | Eve has raised $103 million in Series B funding at over a $1 billion valuation. |
| SV003 | Eve | Legal AI Software for Plaintiff Law Firms | Eve | EveOS is the AI operating system for plaintiff law. Every case, attorney, and dollar, in one place. Trusted by 1200+ firms. |
| SV004 | Eve | The Dawn of a New Era in Plaintiff Law: Eve Secures $47M Series A | More than 100 firms have embraced this new paradigm, driving a staggering 500% year-over-year revenue increase. |
| SV005 | LawNext | Eve, AI-Driven Platform for Plaintiff-Side Law Firms, Raises $103 Million in Series B Round | Eve, AI-Driven Platform for Plaintiff-Side Law Firms, Raises $103 Million in Series B Round. |
| SV006 | Artificial Lawyer | Eve Bags $103m, Hits $1bn+ Valuation | Eve Bags $103m, Hits $1bn+ Valuation. |
| SV007 | Sixth Street | Clio announces US $900M investment at US $3B valuation to transform the legal experience for all | Clio has grown its revenue beyond US $200M ARR. |
| SV008 | TechCrunch | Clio raises $900M at a $3B valuation, plans to double down on AI and fintech | Clio recently crossed over $200 million in ARR. |
| SV009 | LawNext | Clio Completes Historic $1 Billion vLex Acquisition, Announces $500 Million Series G at $5 Billion Valuation | Clio becomes a company with $400 million in annual recurring revenue and a customer base of 400,000 legal professionals. |
| SV010 | American Bar Association | Profile of the Legal Profession | The ABA’s 2025 Profile of the Legal Profession reveals that the number of lawyers in the U.S. has increased significantly for the first time since 2020, rising to 1.37 million in 2025. |
| SV011 | Clio | 2026 Personal Injury Law Statistics: What the Data Reveals | Of the over 1.3 million lawyers in the United States, over 135,000 are personal injury lawyers—roughly 10% of all practicing attorneys. |
| SV012 | Clio | Clio Pricing | Plans for Every Law Firm | Starting at $49/user. |
| SV013 | MyCase | MyCase Pricing | Basic $50 USD/user/month; Pro $100; Advanced $130. |
| SV014 | PracticePanther | PracticePanther Pricing | Annual pricing runs from $69 to $114 per user/month across Essential, Business, and Business Pro tiers. |
| SV015 | EvenUp | EvenUp Raises $150M Series E at $2B+ Valuation to Redefine Personal Injury Law | The funding brings EvenUp’s total capital raised to $385 million and its valuation to over $2 billion. |
| SV016 | Business Wire | EvenUp Raises $150M Series E at $2B+ Valuation to Redefine Personal Injury Law | Today, over 2,000 firms—including 20% of the Top 100 U.S. personal injury firms—depend on EvenUp’s platform. |
| SV017 | Supio | Supio Announces $60M Series B to Accelerate Adoption of Legal AI in Plaintiff Law | Since emerging from stealth in August 2024 with its $25 million Series A funding, Supio has experienced four times Annual Recurring Revenue (ARR) growth. |
| SV018 | LawNext | Legal Tech Startup Supio Secures $60 Million in Series B Funding to Expand AI Platform for Personal Injury Law | Supio has secured $60 million in Series B funding to accelerate the development and adoption of its AI-powered platform for personal injury law firms. |
| SV019 | Filevine | Filevine Raises $400M with Insight, Accel, and Halo Fund to Scale Legal Intelligence | The company has grown rapidly since its founding, with nearly 6,000 customers and 100,000 legal professionals using Filevine. |
| SV020 | Insight Partners | Filevine Raises $400M with Insight, Accel, and Halo Fund to Scale Legal Intelligence | Filevine has proven its ability to sustain tremendous growth while simultaneously capturing new opportunities and markets. |
| SV021 | Litify | Litify Customers | 450+ Enterprise Customers; 70,000+ Legal Professionals; 5M+ Cases Handled Annually. |
| SV022 | LawNext | Bessemer Venture Partners Acquires Majority Stake in Legal Practice Management Company Litify | Litify had previously raised $50 million in Series A funding in 2019. |
| SV023 | Harvey | Harvey raises at $11 billion valuation to scale agents across law firms and enterprises | The round values Harvey at $11 billion. |
| SV024 | CNBC | Legal AI startup Harvey valued at $11 billion in funding round, as VCs spread bets beyond model companies | The company hit $190 million in annual recurring revenue in January. |
| SV025 | Securities and Exchange Commission | Intapp, Inc. Form 10-Q for the quarter ended March 31, 2026 | During the three months ended March 31, 2026, we generated total revenues of $146.0 million. |
| SV026 | CompaniesMarketCap | Intapp (INTA) - Market capitalization | As of June 2026 Intapp has a market cap of $1.78 Billion USD. |
| SV027 | SaaS Capital | SaaS Valuation Multiples | The SCI ... stands at 6.7x as of June, 2025. |
| SV028 | American Bar Association | Profile of the Legal Profession | The number of lawyers in the U.S. ... rising to 1.37 million in 2025. |
| SV029 | Thomson Reuters Institute | 2026 Report on the State of the US Legal Market | The year’s exceptional results are built on uncertain foundations. |
| SV030 | GeekWire | Clio hits $5B valuation after acquiring vLex | Clio hit a $5 billion valuation after closing a $500 million Series G round and $1 billion vLex acquisition. |