If you want to understand how advanced technology is driving transformational change in the legal market, instead of looking at how law firms and lawyers operate today, it is far more instructive to consider the legal tech companies that are building next generation tools and products based on AI and machine learning. These companies are themselves the first and best proponents for a new way of doing business, illustrative of the incredible things that can be accomplished with state-of-the-art technology. They are building their own businesses from the bottom up, leveraging AI and machine learning to the max. This is what enables them to run lean and also run circles around their larger entrenched competitors, such as Thomson Reuters and Lexis-Nexis. This is what enables them to routinely bring innovative products to market.
Trellis Research Inc. is a perfect example. Founded as a bootstrap start-up in 2018, the company has single-handedly pioneered the idea of collecting and mining state trial court data to provide valuable insights for litigators and legal teams. In less than 3 years, Trellis developed the technology, collected a trove of data, launched a first-generation product in California, and then rolled out their service in rapid succession in New York, Texas, Florida, Delaware and Illinois. Not too shabby for a start-up that only recently increased their headcount from 6 to 15.
Of course, as with many start-ups, Trellis also has a more complicated and interesting back story. The
idea for the business initially sprang from the brain of Nicole Clark, while she was a litigator at a California law firm, with a practice focused on employment law. From an early stage in her career, Nicole had been struck by the haphazard way in which lawyers prepared motion papers, without ready access to reliable information about the trial court judge who was scheduled to handle the hearing.
Clark’s aha moment came one night while she was writing a motion for summary judgment. Lacking any information about the judge, she felt hamstrung about how to structure her argument. After complaining to a colleague working down the hall, it turned out that the colleague had appeared before the same judge and had a copy of the judge’s prior ruling on the exact same question at issue in the pending motion. “And that was when the light bulb went off for me,” according to Clark. “I realized how crazy it was that lawyers didn’t have ready access to that kind of practical information even though the data was out there. How was it possible that here we were primarily litigating in state trial courts, and yet we were using appellate court resources to do all of our legal research? It just didn’t make any sense.”
It proved to be a long road from Clark’s late-night inspiration to launching Trellis’ first generation product. There were countless obstacles to overcome, largely due to the difficulty of collecting and processing state trial court data – otherwise Thomson Reuters and Lexis would likely have tackled this challenge long ago. Not only is every state different, in most cases there is also variety from county to county, and even from court to court, in terms of the way the data is stored and presented, which is precisely why no one had yet been able to aggregate and systematically analyze it.
But around the same time that Clark had her aha moment, rapid advances were being made in machine learning and natural language processing, for the first time making it possible to tackle problems involving massive amounts of unstructured data. Clark had the good fortune to have several friends who were engineers with expertise in the emerging field. And as it turns out, machine learning was exactly the right technology for the job.
“There is no way that humans alone could go through the enormous volume of data we needed in order to build Trellis,” Clark explains. “I can’t even begin to imagine what a laborious manual process it would have been to tackle this without a machine learning engine.” But with the help of the proper algorithm, the small dedicated team at Trellis was able to accomplish the task with uncanny efficiency.
“Even now,” as Clark explains, “we’re at the infancy of what machine learning and natural language processing will be able to accomplish. This is part of what makes it so exciting for start-ups in the legal tech space these days. Machine learning is also an engine for continuous improvement, which means Trellis today has only achieved a fraction of our potential.”