Harvey, who described the startup as a “lawyer’s assistant,” announced today a $5 million round of funding led by the OpenAI Startup Fund, where OpenAI and its partners are investing in early-stage AI companies to solve key problems. Problems. Also participating in the round was Jeff Dean, Head of Google AI, Google’s AI Research Unit. and Elad Gil, co-founder of Blender Labs, among other angel supporters.
Harvey was founded by Winston Weinberg, formerly of securities and antitrust law firm O’Melveny & Myers, and Gabriel Pereira, formerly a research scientist at DeepMind, Google Brain (another of Google’s AI teams), and Meta AI. Weinberg and Pereira are roommates – Pereira demonstrated Weinberg’s OpenAI GPT-3 text generation system and Weinberg realized it could be used to improve legal workflows.
“Our product provides a natural language interface for legal workflows,” Pereira told TechCrunch in an email interview. “Instead of manually editing legal documents or conducting legal research, Harvey allows lawyers to define the task they want to perform with simple instructions and receive results. To enable this, Harvey uses large language models to understand users’ needs and deliver the right results.”
More concretely, Harvey can answer natural language questions like, “Tell me what the Fourth Circuit distinguishes between an employee and an independent contractor” and “Tell me if this lease violates California law, and if so, rewrite it so it’s no longer a violation.” At first reading, it appears that Harvey can replace lawyers, generate legal arguments, and produce drafts in short order. But Pereira insists this is not the case.
“We want Harvey to serve as an intermediary between the tech and the lawyer, as the natural language interface of the law,” he said. “Harvey makes lawyers more efficient, allowing them to produce higher-quality work and spend more time on the highest-value parts of their work. Harvey provides a unified and intuitive interface for all legal workflows, allowing lawyers to define tasks in plain English rather than using complex and specialized tools for core tasks.” It allows.
It’s powerful stuff in theory. But it’s also full. Given the extreme sensitivity of most legal disputes, lawyers and law firms may be willing to give a device like Harvey access to any case files. There is also the issue of language models’ proclivity to reveal toxicity and muddled facts, which are not particularly well received – if not by perjury – in a court of law.
That’s why Harvey, which is currently in beta, has a disclaimer attached to it: The tool is not intended to provide legal advice to non-lawyers and should be supervised by licensed attorneys.
On the issue of data privacy, Pereira said Harvey takes pains to meet customer compliance needs, de-identifying user data and deleting data after a certain period of time. Users can delete data at any time upon request, he says, and Harvey can take comfort in the fact that it can’t “contaminate” data between customers.
It’s early days. But in the past, Pereira says Harvey is being used by “users across the legal landscape,” from law firms to legal aid organizations.
He faces some competition. Casetext uses AI mainly GPT-3 to find legal cases and assist with general legal research tasks and brief drafting tasks. Additional surgical tools, such as Clarity, use AI to decouple contract revisions. At one point, Startup Augrented was looking for ways to use GPT-3 to summarize legal notices or other resources in plain English to help tenants protect their rights.
For one, Brad Lightcup, OpenAI’s CCO and manager of the OpenAI Startup Fund, believes Harvey’s has been sufficiently identified. It also benefits from its connection with OpenAI. In addition to capital, OpenAI Startup Fund participants also get early access to new OpenAI systems and Azure resources from Microsoft.
“We believe Harvey will transform our legal system, empowering lawyers to efficiently provide high-quality legal services to a wide range of clients,” Lightcap said in an email. “We launched the OpenAI Startup Fund to support companies using powerful AI to advance community-level impact and align Harvey’s vision of how AI can improve access to legal services with our mission to deliver results.”
Harvey has a five-person team, and Pereira expects that number to grow to five to ten employees by the end of the year. He would not answer when asked about revenue figures.