A new star on the horizon of Silicon Valley has breathed fresh hope into the field of artificial intelligence computing, and it has originated from the most unlikely of places, a union of Harvard Law and Silicon Valley. The story of Evisort is the kind that is only encountered a few times in a lifetime. It's so rare to see experts in two deep professions, law, and computing, that it takes a bit of explaining to understand why this company is such big news.
Legal Research: The Old Way
Contracts are the language of business. Any time two businesses, even if they're sole proprietor consultants, maintain a business relationship, the contracts define the legal boundaries of that working partnership. Companies may accumulate up to thousands of contracts, which means they need somebody to stay up to date on all of them.
If a lawyer at a corporation's legal department gets called upon to assist with renegotiating a provision in a contract with another company, they have to search through this mountain of paperwork. At the current standard, scanned-in text of contracts can be searched, database-style, for legal clauses like "assignment of deliverables," "aggregation of stock," or "release of escrow." But the question they have in mind might be something like "What is standing company policy with this kind of contract?" To find out, the lawyer would have to open each of these documents and read them, at lengths going up to 30 pages.
More likely or ideally, the lawyer would assign this chore to a paralegal team. The process could take days, all because standard software lacks the ability to semantically parse legal language into more relevant contextual categories. Anyone who has had the experience of searching Google for a subject only to find they have to narrow down their search to weed out unrelated hits knows this problem.
Legal Research: The Evisort Way
While semantic textual comprehension of any general-purpose text evades the artificial intelligence field, legal language is a very narrow and exacting subset of language. Very similar to how a programming language uses text to feed into a compiler and receive a functioning utility, legal language is processed within the systems of national and international law to set up the utilities of business.
Evisort is one of the most exciting implementations of deep learning algorithms. Since there are problems that are very complicated to explain to a computer, deep learning is a method where we set up a program to enable it to try any and all methods of solving a given task, without the need to define "how." It's the equivalent of shutting a robot in a maze and telling it "page me when you find the exit," allowing it to roam freely and record trial-and-error paths. In this case, the deep learning AI at the heart of Evisort was fed reams of legal contract data and allowed to attach significance to patterns of terms and clauses, using flexible models of data categorization, until it could predict them on its own.
The end result is that the same lawyer with the same problem can now pull up the terms from 30, or even 300, contracts and not only readjust the relevant data but have it presented in a chart. This now takes seconds.
Fast Answers From An Automated Paralegal
Since Evisort's founding in 2016, it has raised millions in investor capital funding, founded offices in California and Massachusetts, and attracted clientele from Fortune 500 companies to Am Law 100 firms, even to garage-sized tech startups and legal departments of every industry.
The average law professional has to handle dozens of contracts per day, all of which usually require painstaking research. But now with AI that can offset the cognitive load, they can not only make faster decisions but better ones thanks to new ways of visualizing data they didn't even have before. Evisort's AI is just one part of the picture. It's a full office suite that's integrated with common office tools, allowing agile and seamless workflow for legal professionals.
As one lawyer put it when Evisort founder Jerry Ting showed him how the software works, "Why have I wasted ten years of my life doing this?"