Sacra estimates that Harvey hit $100M in annual recurring revenue (ARR) in August 2025, up from $50M at the end of 2024.

After its first paid pilots in 2023, the San Francisco-based legal AI vendor expanded to more than 500 enterprise customers—including Comcast—while weekly active users grew fourfold year over year.

Revenue is seat-based: accounts typically start with a few hundred licenses for research, drafting and diligence, and internal usage data show that median seat count doubles within 12 months. Sector-specific modules for insurance and financial services, along with integrations into document-management systems, have lifted average contract values.

Harvey completed a $300M Series E at a $5B valuation in June 2025, co-led by Kleiner Perkins and Coatue, just four months after raising $300M Series D at a $3B valuation led by Sequoia in February 2025. This represents a 67x forward revenue multiple on current ARR.

Total funding now exceeds $506M across five rounds, with backing from Sequoia, Kleiner Perkins, Coatue, OpenAI Startup Fund, GV, and LexisNexis's RELX Group.

Harvey was founded in 2022 by Winston Weinberg, a former securities and antitrust litigator at O'Melveny & Myers, and Gabriel Pereyra, previously a research scientist at DeepMind and Meta AI.

Harvey found product-market fit as a legal AI copilot for large law firms and corporate legal departments, offering specialized capabilities for document analysis, legal research, and multi-language translation.

The platform helps lawyers analyze complex legal documents, conduct research across case law databases, and draft legal documents. When reviewing contracts or legal filings, Harvey can identify key provisions, flag potential issues, and generate summaries. For legal research, it can analyze precedents and relevant cases, helping attorneys build stronger arguments and identify potential weaknesses in their positions.

The company scrapped its proprietary vertical model after frontier reasoning models from Google, xAI, OpenAI, and Anthropic began outperforming Harvey's custom legal model on its own BigLaw Bench evaluation.

Harvey now positions itself around pre-configured agentic workflows that chain multiple LLMs and tools together to complete specific legal tasks. Rather than relying on a single specialized model, the platform orchestrates different AI models depending on the task—using one model for document analysis, another for legal research, and a third for contract drafting. This approach allows Harvey to leverage the best-performing model for each workflow step while maintaining the legal-specific context and enterprise security that law firms require.

Harvey is a subscription SaaS company that licenses customized large language models (LLMs) to law firms and corporate legal departments, with pricing based on both per-seat licensing and custom model development fees.

The company's base offering starts at $1,200 per lawyer per month with 12-month commitments and roughly 20-seat minimums.

Harvey's model combines software licensing with intensive "forward-deployed" services—Harvey dedicates roughly 10% of its team to ex-lawyers in customer success roles who drive change management, implementation, and adoption within law firms to ensure clients hit utilization thresholds needed for renewal.

The shift from custom model training to pre-configured agentic workflows reduces Harvey's implementation complexity while maintaining the high-touch service model that justifies premium pricing. This approach allows Harvey to scale more efficiently across different legal use cases without the resource-intensive custom training that previously limited growth velocity.

The legal AI landscape has fundamentally shifted as frontier reasoning models have commoditized legal reasoning as a core differentiator. Major LLM providers including Google Gemini, xAI Grok, OpenAI, and Anthropic now match or exceed specialized legal models on standardized benchmarks, forcing vertical AI companies to compete on workflow orchestration and enterprise integration rather than model performance.

This commoditization favors companies with strong distribution and implementation capabilities over those relying primarily on proprietary model training.

The most direct competitors are specialized legal AI platforms built for large law firms and enterprises. These include Casetext's CoCounsel (acquired by Thomson Reuters), ROSS Intelligence, and Blue J Legal. These platforms focus on legal research, document analysis, and case law interpretation using specialized AI models. Unlike Harvey's custom-trained models for specific firms, most competitors offer standardized solutions.

Established players like Thomson Reuters (Westlaw), LexisNexis, and Wolters Kluwer are integrating AI capabilities into their existing legal research and practice management platforms. These companies have massive databases of legal content and established relationships with law firms, but their AI offerings tend to be more limited in scope compared to pure-play AI companies.

Large language model providers like OpenAI (ChatGPT), Anthropic (Claude), and Microsoft (Azure OpenAI) offer capabilities that can be adapted for legal work. While these platforms have sophisticated AI technology, they lack the legal-specific training and enterprise security features that Harvey provides. Some law firms are attempting to build their own solutions on top of these platforms.

The market is seeing rapid consolidation, exemplified by Thomson Reuters' acquisition of Casetext for $650M in 2023. Major law firms are increasingly partnering with AI providers, as demonstrated by Allen & Overy's exclusive partnership with Harvey and PwC's strategic alliance for custom AI models in tax and legal services.

Harvey has tailwinds from the rapid advancement of LLM technology and growing enterprise acceptance of AI tools, with opportunities to expand beyond legal services into the broader professional services market and eventually become an AI super-app for knowledge workers.

The legal services market represents a $300B+ initial opportunity in the U.S. alone. Harvey's early success with firms like Allen & Overy and PwC demonstrates the massive efficiency gains possible through AI-assisted legal work. The company's ability to train custom models on firms' proprietary documents while maintaining security and compliance creates strong competitive moats.

Harvey's text-processing capabilities naturally extend to adjacent professional services markets like accounting, consulting, and financial services. These industries face similar challenges around document analysis, research, and compliance. The global professional services market exceeds $5 trillion annually, presenting an enormous expansion opportunity as Harvey develops industry-specific models and workflows.

The ultimate vision for Harvey extends beyond specialized professional services to become a general AI super-app for knowledge workers. By starting with high-value, compliance-sensitive use cases in legal, Harvey is building the enterprise-grade infrastructure and security protocols needed to serve broader knowledge work applications. This positions them to capture a significant share of the global knowledge worker productivity market, estimated at over $50 trillion annually. Their partnership with Microsoft Azure provides a scalable distribution channel to reach this broader market.

Multi-model coordination complexity: Harvey's pivot to agentic workflows that chain multiple LLMs creates new operational risks around model coordination, latency management, and cost optimization across different providers. Managing reliability and performance when workflows depend on multiple external AI services introduces additional failure points and makes debugging more complex than single-model deployments.

Custom implementation challenges: Harvey's need to train models on each law firm's proprietary documents and workflows limits scalability. The high-touch deployment model requires significant resources per client and extends time-to-value. This could constrain growth velocity and unit economics, particularly as they attempt to expand beyond elite law firms.

OpenAI dependency: As a company built on GPT-4, Harvey is heavily dependent on OpenAI's infrastructure and pricing. Changes to OpenAI's terms, costs, or competitive stance could materially impact Harvey's margins and competitive position. Their focus on custom implementations makes it difficult to quickly switch to alternative LLM providers if needed.