Why AI Could Turn Small Law Firms Into Billion-Dollar Businesses

在这期播客节目中，Andreessen Horwitz 合伙人、主持人亚历克斯·兰佩尔（Alex Rampell）与 EVE 联合创始人兼首席执行官杰伊（Jay）一同探讨了法律人工智能及其对原告律师行业的变革性影响。 硅谷的大多数人将法律行业与公司法以及精英律所每小时 2,000 美元的账单联系在一起。然而，原告律师行业——通常通过 101 号公路或 I-95 号州际公路旁的巨型广告牌进入人们的视野——是一个规模更大、分布更广、采用风险代理费用的行业。历史上，原告律师面临着严重的劳动力限制，会拒绝大约 90% 至 96% 的潜在案源，因为人工成本使得处理较小的案件在经济上不可行。这就造成了一种市场失灵，即拥有合理、较小损失的人因价格过高而无法获得法律代理。 EVE 解决了这一劳动力限制。EVE 没有充当传统的削减成本工具，而是作为原告律所的收入扩展平台，使其能够承接成倍增加的例如，洛杉矶的一家律所将每位律师管理的客户从大约 100 名增加到 500 名。EVE 端到端地处理案件——从最初的呼入电话和自动资格审查，到起草简报和达成和解讨论。 EVE 没有完全依赖像 ChatGPT 或 Claude 这样的纯原始智能模型，而是通过将不可抓取的原告法律数据、本地司法管辖区背景和保险公司策略编码到其工作流程中来弥合差距。由于只有不到 1% 的案件真正进入审判并在之前达成和解，这些数据历史上一直分散在各个地方律所中。EVE 整理了这些信息，提高了律所能够完成的工作的基线。 这场对话探讨了 EVE 如何改变个人原告与大型机构对手（如保险公司）之间的权力动态，这些机构传统上依赖拖延和拒绝等策略。随着 EVE 自动化常规任务和越界的证据披露回复，辩护律所开始在对方阵营遇到自动化代理。 展望未来，杰伊设想了一个未来，律所以更像软件公司的方式运营，利用以结果为导向的定价模型，并深入专精于特定类型的案件。虽然 EVE 尚未针对每一个案件完全实现端到端的自主运行，但它正在迅速向在人工干预极少的情况下处理数万次来回互动（例如向医疗保健公司拨打呼出电话和起草文件）迈进。 归根结底，EVE 旨在使法律体系民主化，改变原告法律的经济学，以便成千上万以前没有代理人的个人最终能够获得司法公正并获得公平的和解。

In this podcast episode, host Alex Rampell, partner at Andreessen Horwitz, is joined by Jay, co-founder and CEO of EVE, to discuss legal AI and its transformative impact on plaintiff law. Most people in Silicon Valley associate the legal industry with corporate law and $2,000-an-hour bills from white-shoe firms. However, plaintiff law—often seen via billboards on the 101 or I-95—is a larger, more distributed industry operating on contingency fees. Historically, plaintiff lawyers faced severe labor constraints and turned down about 90% to 96% of leads because the cost of labor made servicing smaller cases economically unviable. This created a market failure where people with legitimate, smaller damages were priced out of legal representation. EVE solves this labor constraint. Instead of acting as a traditional cost-cutting tool, EVE functions as a revenue expansion platform for plaintiff law firms, allowing them to take on exponentially more cases. For example, one firm in Los Angeles went from managing about 100 clients per attorney to 500 clients per attorney. EVE handles a case end-to-end—from the very first inbound call and automatic qualification to drafting briefs and settlement discussions. Rather than relying purely on raw intelligence models like ChatGPT or Claude, EVE bridges the gap by encoding non-scrapable plaintiff law data, local jurisdiction contexts, and insurance carrier tactics into its workflows. Because less than 1% of cases actually go to trial and settle beforehand, this data historically lived sharded across local firms. EVE organizes this information, raising the baseline of what a law firm can accomplish. The conversation explores how EVE changes the power dynamics between individual plaintiffs and large institutional adversaries like insurance companies, who traditionally rely on tactics like delay and denial. With EVE automating routine tasks and out-of-bounds discovery responses, defense firms are beginning to encounter automated agents on the other side. Looking forward, Jay envisions a future where law firms operate more like software companies, utilizing outcome-oriented pricing models and specializing deeply in specific types of cases. While EVE is not yet fully autonomous end-to-end for every single case, it is rapidly moving toward handling tens of thousands of back-and-forth actions—such as outbound calls to healthcare companies and drafting documents—with minimal human intervention. Ultimately, EVE aims to democratize the legal system, shifting the economics of plaintiff law so that thousands of previously unrepresented individuals can finally access justice and secure fair settlements.