Legal work is labor-intensive and time-consuming, requiring cases to be compiled from vast amounts of evidence. That's why some law firms are piloting AI to streamline certain procedures. According to a 2023 survey by the American Bar Association, 35% of law firms currently use AI tools in their work.
OpenAI-backed Harvey is one of the big winners so far in the burgeoning AI legal tech space, along with startups like Leya and Klarity. But there's room for one more player, say Jerry Zhou and Kyle Lam, co-founders of Supio, an AI platform for personal injury law that emerged from stealth on Tuesday with a $25 million investment led by Sapphire Ventures.
Supio uses generative AI to automate the collection and aggregation of large amounts of data for legal teams. In addition to summarizing information, Zhou says the platform can organize and identify files, or snippets within files, that may be useful for case briefs, drafts, and presentations.
“After attending numerous conferences across the U.S. and meeting hundreds of lawyers, Lam and I decided to focus on personal injury and class action plaintiff law,” says Chou, who is also Supio's CEO. “These are practice areas that require us to collect thousands of documents from multiple sources and analyze the data within to find information.”
Chow and Lam are childhood friends whose career paths have often crossed: They both worked at Microsoft, specifically for its Office 365 organization, before working together again at tax compliance software company Avalara.
The idea for Supio came about after Zhou and Lam left Avalara to pursue building a business that, in Zhou's words, “helps people make sense of complex data and identify key connections within that data.”
“We pursued the legal industry because we knew it was not only document-heavy but also required technological innovation,” Zhou said. “These industries are fields that require collecting thousands of documents from multiple sources and analyzing the data within to find information.”
Civil lawsuits brought on behalf of victims injured in personal injury, mass tort or negligence cases such as the sale of defective products typically revolve around documents such as medical records, police reports, insurance claims, financial statements and consumer complaints. Chou explained that Supio creates a demand letter outlining the legal dispute to be resolved and supporting documents, then allows users to search for evidence through a chatbot-type interface.
This is similar to EvenUp, a startup that uses AI to create legal documents for evaluating personal injury cases, and companies like Lawyaw and Atrium are also applying AI to writing initial complaints.
But Zhou argues that Supio's technological approach is more complex.
“Law is a very complex and nuanced subject, and most creators of work productivity tools don't really understand the legal documents that lawyers ultimately have to produce, which hinders the development of accurate (AI) models,” Zhou said. “At Supio, we run hundreds of models with different capabilities at any given time, trying to understand and classify the documents. We then compare this with the expected work products and gradually improve the results.”
AI like Supio is powerful in theory, but it also comes with risks: Due to the sensitive nature of most legal disputes, lawyers and law firms may be reluctant (or prohibited from) allowing a tool like Supio access to legal documents.
Late last year, the California State Bar issued guidelines instructing legal professionals not to input client information into AI tools that “lack reasonable or adequate security.” (Zhou said Supio stores customer data in its country of origin and has security protocols that comply with privacy regulations like HIPAA and GDPR.)
Another concern with AI legal tech is AI's tendency to perjure itself. Last year, a group of lawyers from Levidow, Levidow & Oberman, PC used OpenAI's AI-powered chatbot, ChatGPT, to prepare a personal injury lawsuit against an airline. The results were disastrous: ChatGPT fabricated citations, misidentified the judge, and referenced airlines that didn't exist. The federal judge presiding over the case ultimately fined the lawyers and their employer $5,000.
Courts are desperately bracing for an increase in inaccurate filings due to the introduction of AI-powered legal research tools.
In November 2023, the U.S. Fifth Circuit Court of Appeals proposed a rule (since repealed) that would require filings of specialized legal documents created with the help of AI to certify that a human has reviewed and approved the documents for accuracy. Earlier that same year, a district court judge in Texas issued an injunction banning the use of generative AI to prepare court filings without human fact-checking.
The risks of AI are so great that a recent survey of more than 300 general counsel and senior legal professionals at large companies found that 25% said they don't think outside counsel should use AI. A separate survey by Thomson Reuters found that one in five law firms has raised warnings about using AI.
Zhou makes the astonishing claim that Supio's AI exhibits “higher than human accuracy” and is “hallucination-free,” meaning it never lies.
“Supio understands that timing and accuracy are key, so we provide flexible, AI-powered software that can organize unstructured data and produce reliable results,” he said.
It's not clear what “human-level” means; Zhou didn't release any test or benchmark results. But it's important to note that just because an AI can accomplish feats like passing the bar exam doesn't mean it has the skills that lawyers gain through experience and education (as the National Board of Bar Examiners asserts). The “no hallucinations” part of Zhou's claim is also not backed up by data, at least not by the data he volunteered.
But some companies believe Supio has a future.
Zhou said Supio currently works with about 30 personal injury and class action law firms and expects that number to reach 100 by the end of the year. Meanwhile, the startup has annual recurring revenue of over $1 million, most of which comes from subscription fees that Supio charges based on the number of cases.
It could be a case of keeping up with the Joneses.
In a survey of legal executives released by LexisNexis, nearly all (90%) said they expect to increase investment in generative AI over the next five years. The same survey found that 43% of firms currently have a dedicated budget for generative AI. Gartner predicts that the appeal of generative AI will drive the legal tech market to $50 billion in value by 2027, nearly double what it was in 2022 ($25.6 million).
Against this dramatic backdrop, Seattle-based Supio is thriving: The company has 27 employees and plans to double its workforce within 12 months.
Having raised a total of $33 million, Supio aims to expand its client base in the near future and eventually “scale up to serve other legal specialties,” Zhou said.
Bonfire Ventures and Foothill Ventures also participated in Supio's latest tranche, Series A, which Zhou said was oversubscribed but declined to disclose the valuation.
