YC’s ROSS Intelligence Leverages IBM’s Watson To Make Sense Of Legal Knowledge

Next Story

Neglect Is Still The Biggest Threat To Data Security

As our body of legal knowledge, court rulings and laws becomes ever more complex, simple keyword searches won’t cut it for surfacing prior precedents and policies.

That’s why the founders behind YC-backed ROSS Intelligence are using IBM’s Watson, an artificial intelligence system capable of natural-language processing, to filter through legal documents.

Instead of searching for documents by keywords, you can ask questions in plain English like “Can an automatic stay be lifted if a plaintiff in another case requests it?” It returns different citations with paired confidence ratings. (Unfortunately, the company didn’t let me test it out personally, so I can’t tell you how good or useful ROSS is or how it compares to Lexis-Nexis.)

“We thought Watson would be perfect because it is good at looking at unstructured data,” said co-founder Andrew Arruda, a trained lawyer who spent several years in legal research in Canada before starting ROSS. “We think this is coming to the perfect market at a perfect time. There’s been a huge shift in the legal industry. Lots of law firms are open to changing their processes because there’s more pressure from the client side to become cost-efficient with billing.”

The famed billable hour, which is how law firms have traditionally charged clients for their labor at rates that can go up to $400 per hour, has been giving way for several years. Simultaneously, lost of the more basic legal paperwork is being outsourced to other countries like India, which has cheaper labor costs and high English fluency rates.

Today, ROSS says that the market size for legal research software is about $8.4 billion per year, given the roughly 1.3 million lawyers in the U.S. and Canada.

“There are thousands of laws are being published each day,” said Arruda, who spent several years in legal research. “But until recently, have our computers have had a very superficial understanding of natural language. ROSS pretty much mimics the human process of reading, identifies patterns in text, and provides contextualized answers with snippets from the document in question.”

The core of ROSS’ corpus of data centers on bankruptcy and insolvency law for the moment, but they’re adding case law and other third-party materials. Arruda’s other co-founders Jimoh Ovbiagele, Akash Venkat, Shuai Wang and Pargles Dall’Oglio have backgrounds in neuroscience and computer science.

The service is free for the moment in a private beta, with a number of undisclosed legal clients.

“People have said we took Watson to law school,” Arruda said.