Harvey, officially Counsel AI, secured $300 million in Series E funding led by Kleiner Perkins and Coatue, valuing the company at $5 billion.
This funding round marks a substantial increase in Harvey’s valuation, which stood at $2 billion earlier in the year following its previous fundraising effort. Additional investors participating in this Series E round included GV, the investment arm of Alphabet Inc., the OpenAI Startup Fund, and Sequoia. RELX plc, the parent company of LexisNexis, also contributed through its venture capital arm, REV.
Harvey specializes in developing artificial intelligence software specifically designed for legal professionals. The company’s platform is currently utilized by over 330 law firms and various other organizations across more than 50 countries globally. This extensive adoption has contributed to Harvey’s financial performance. As of April, Harvey reportedly generated $75 million in annualized recurring revenue. This figure represents a significant increase from $50 million in annualized recurring revenue recorded just a few months prior, indicating accelerated growth in its customer base and service utilization.
A core offering from Harvey is its flagship AI tool, Harvey Assistant, which aims to enhance the productivity of legal teams. The tool automates several time-consuming tasks traditionally performed by attorneys. When drafting new legal documents, lawyers typically begin by researching relevant laws and court decisions. Harvey Assistant streamlines this initial research phase, allowing legal teams to aggregate information efficiently from diverse sources such as court filings using natural language prompts. A key feature of this tool is its inclusion of citations within its prompt responses, enabling legal teams to verify the accuracy and origin of the generated output.
Beyond information aggregation, Harvey Assistant also automates the document drafting process itself after necessary documents have been collected. Following the initial draft, the tool checks the newly created document against an organization’s specific internal guidelines for legal filings. It also offers proofreading capabilities for these documents and can translate them into different languages. For contracts, Harvey Assistant can automatically populate required information into each designated field, further expediting the creation process.
The process of revising complex legal agreements, commonly known as redlining, often involves numerous iterations before finalization. Attorneys traditionally indicate sections for deletion or modification using a red strikethrough. Harvey’s platform introduces automation to this process by summarizing document changes automatically, thereby streamlining revision management for users. To power these redlining capabilities, Harvey developed a custom AI model. This model features approximately 100,000 parameters and utilizes a convoluted neural network architecture, a design commonly employed in image processing applications. Prior to processing a contract, Harvey Assistant converts the PDF file containing the contract into an image. Subsequently, the software applies a grayscale filter to this image. The company indicates that its engineers have determined this specific arrangement enhances the AI model’s accuracy and improves the speed of inference.
While Harvey’s redlining functions rely on a custom-built AI model, other features within its platform are powered by off-the-shelf large language models, including those from providers like OpenAI. Moving forward, Harvey intends to integrate AI models from LexisNexis, a prominent provider of legal document databases and software tools tailored for attorneys. The participation of RELX plc, LexisNexis’s parent company, in Harvey’s recent funding round through its REV venture capital arm underscores this planned collaboration.
