Legal Insights

Harvey AI is one of the most closely watched innovators in legal technology in recent years, successfully combining generative artificial intelligence with legal expertise to offer law firms and legal departments worldwide an unprecedented opportunity to transform how they work. Since its founding in 2022 by former O'Melveny & Myers lawyer Winston Weinberg and former Google DeepMind researcher Gabriel Pereyra, Harvey AI has rapidly drawn industry attention and, within just a few years, been put into practice at leading law firms and large enterprises around the world, becoming a key driver of the modernization of legal services.

At the core of Harvey AI's product is an AI platform designed specifically for the legal domain, with main modules including Harvey Assistant (a natural-language legal assistant), Vault (large-scale document storage and semantic analysis), Knowledge (a knowledge management system) and Workflows (an automation engine). These tools support the full span of tasks from legal research, contract review, compliance checks and case preparation through to litigation strategy analysis, dramatically reducing the time lawyers spend on routine work.

Among them, Harvey Assistant is the module lawyers use most: they can ask questions directly in natural language and receive instant legal research results, draft contract clauses and analytical opinions. Assistant can generate compliant content for the needs of different jurisdictions, with case law and statutory sources attached to ensure professional accuracy. Vault supports high-speed analysis of more than 10,000 documents, automatically extracting keywords and risk factors — saving firms substantial manpower on first-pass document screening.

Harvey's underlying technology is built on GPT-4 with vertical fine-tuning for the legal domain, trained on corpora spanning case law, statutes and regulations, contract samples and practice commentary, allowing it to understand legal context more accurately than general-purpose AI models. Harvey's legal AI models provide complete citations to sources and support customized training and private deployment, meeting the legal industry's exacting requirements for confidential data and dedicated systems.

One of its success stories comes from leading UK firm Allen & Overy. The firm began trialing Harvey AI internally as early as November 2022, and in February 2023 formally rolled the platform out to 3,500 lawyers across 43 offices worldwide. During the pilot, more than 40,000 queries were submitted, spanning legal research, due diligence, contract drafting and clause comparison. The firm's innovation lead David Wakeling remarked: "I have worked in legal tech for 15 years and have never seen a game-changing tool like Harvey."

Beyond Allen & Overy, Paul Weiss in the United States and A&L Goodbody in Ireland have also adopted Harvey AI. While Paul Weiss has been measured about the efficiency gains, it still sees AI as a powerful tool for first-pass analysis of large volumes of data, helping lawyers focus their attention on high-value decisions.

On the corporate legal side, PwC, one of the Big Four accounting firms, deployed Harvey AI in Singapore in 2024, applying it to contract analysis and regulatory interpretation. The same month, prominent Singapore firm WongPartnership became the first law firm in Southeast Asia to deploy the Harvey platform, marking its successful entry into Asia's professional services market.

Harvey AI's business model centers on enterprise-grade SaaS: it does not serve individual users, instead engaging law firms and enterprises directly through a "Request a Demo" process. Pricing runs on annual or quarterly subscriptions, with plans customized by user count and module selection — contract-intensive clients may prioritize Vault and Workflows, while litigation-focused clients prefer Assistant and the knowledge modules.

By 2025, Harvey AI's annual recurring revenue (ARR) had surpassed USD 75 million, with continued expansion into large international firms. Its investors include the OpenAI Startup Fund, Sequoia Capital, Kleiner Perkins and GV (Google Ventures); its valuation reached USD 1.5 billion after its Series C round, and its Series D in early 2025 lifted the valuation to USD 3 billion — a super-unicorn of legal AI.

Harvey's product also keeps evolving: from 2025 the platform supports a multi-model architecture — beyond the original OpenAI GPT-4 base, it also integrates Anthropic's Claude and Google's PaLM 2, letting clients choose the model best suited to their preferences and jurisdiction. The Harvey on Azure model likewise supports deployment in an enterprise's own cloud environment, strengthening data control.

Harvey AI's growth strategy includes aggressive international expansion and recruiting talent with legal backgrounds. By the end of 2024 its team was set to reach 160 people, with most product and marketing staff bringing experience at international law firms — former lawyers from Skadden, White & Case and Paul Weiss among them — strengthening its professional dialogue with clients.

As for Chinese-language legal support, Harvey has not yet been specifically trained on Chinese legal corpora. It can handle basic Chinese questions and document analysis, but its precision and professional depth remain strongest in English-language common law. Even so, its multilingual capability and open architecture give it the potential to expand into Chinese-language markets in the future, possibly by partnering with local players to build dedicated Chinese models.

In sum, Harvey AI demonstrates the feasibility and impact of putting generative AI into practice in the legal industry. Through partnerships with the world's top firms and sustained investment in specialized model training and security design, it is bringing profound change to legal services. Its commercial success also proves that when AI technology is deeply integrated into professional workflows, it not only unlocks productivity but may reshape the industry's division of labor and competitive landscape.

The Harvey AI case stands as a model of the fusion of law and technology — delivering significant efficiency gains for large firms and enterprises, and providing an evidence-based reference model for the development of legal AI worldwide.