A law firm does not buy AI simply because it can produce convincing paragraphs. It needs software that helps lawyers work with contracts, research questions, and client documents without creating an unacceptable confidentiality or accuracy problem. Harvey has built its business around that distinction.
Founded in 2022 by Winston Weinberg and Gabriel Pereyra, Harvey develops AI software for legal and professional-services work. Its appeal is not that it replaces a lawyer’s judgment. It is that a specialized product can make document-heavy assignments more manageable while fitting the requirements of an institutional buyer.
A startup with evidence beyond investor enthusiasm
Harvey’s February 2025 funding announcement provides a useful, dated snapshot of its progress: the company reported 235 customers across 42 countries and said its annual recurring revenue had grown fourfold during 2024. It also announced a $300 million Series D round led by Sequoia Capital, valuing the business at $3 billion.
Those figures should not be confused with profitability. Funding measures investor backing, while a valuation reflects the terms of a financing round. Nevertheless, customer adoption and recurring-revenue growth offer stronger evidence of a working business than a large funding headline alone. The operating figures were company-reported, rather than an independently audited financial account.
An early milestone was Allen & Overy’s 2023 rollout of Harvey, before the firm became part of A&O Shearman. Adoption by a major international law firm gave the startup a reference point in a market where buyers pay close attention to what their peers trust.
The product is built around the assignment
Harvey’s offering includes an assistant for legal tasks, tools for working across collections of documents, and reusable workflows. In practice, the opportunity lies in activities such as comparing contract provisions, extracting information from large document sets, and preparing an initial analysis for a lawyer to check.
Consider a team reviewing agreements during an acquisition. The useful output is not merely a readable summary. It is a consistent account of which agreements contain particular obligations, where those provisions appear, and what deserves closer inspection. That makes legal document analysis a more demanding product problem than generating fluent text.
The distinction also explains why human review of AI outputs remains central. An omitted exception or an unsupported legal statement can matter more than the time saved producing a draft. Harvey’s business case depends on helping professionals complete work, not making professional responsibility disappear.
Why specialization can support a substantial business
Harvey sells into organizations where skilled staff time is expensive and much of the work involves reading, comparing, and drafting. That creates room for an enterprise product, provided customers can establish that it improves actual assignments.
- Relevant workflows: Tools shaped around legal assignments can require less adaptation than a general-purpose chatbot.
- Institutional purchasing requirements: Confidentiality, access permissions, and security reviews make enterprise AI procurement part of the product’s commercial challenge.
- Repeatable use: A tool used across recurring assignments has a stronger subscription case than one purchased for occasional experimentation.
This is a useful example of vertical AI software: software designed for a particular profession rather than every possible user. The underlying model matters, but so do document handling, workflow design, customer support, and AI adoption in law firms.
The risks are equally specific. Established legal-information providers have valuable content and existing customer relationships. General-purpose AI products keep improving. Legal buyers also expect demanding standards for confidential client data. Harvey must demonstrate that its specialized experience remains worth paying for as those alternatives develop.
For other founders, the practical lesson is to examine the work surrounding an AI response. Harvey’s reported growth suggests that customers will pay for a product organized around their profession’s assignments and constraints. Whether that becomes a durable advantage depends on reliable outcomes and renewals—not the valuation attached to the next funding round.
Source basis: Harvey’s February 2025 Series D announcement and company product information, available through its official blog and website. Financial and customer figures above refer to that announcement, not a live estimate of the company’s current size.
