Legal Tech Entrepreneur Says ‘Autopilot’ Contract Analysis More Accurate than Human Lawyers

Tel Aviv-based startup Lawgeex, registered in Israel as Legalogic Ltd., is developing an Artificial Intelligence (AI)- based platform that automates the process of reviewing legal contracts.

Lawgeex CEO and co-founder Noory Bechor, a lawyer turned tech entrepreneur, had discussed in an interview with Calcalist the technology and what it means for the future of the legal business.

Why did you decide to start Lawgeex?

I was very frustrated with the inefficiency and the many hours I spent going over simple contracts, and thought there was a need for technological tools that could improve the productivity of legal work.

Are you expecting that Lawgeex will one day replace lawyers?

No. The idea is to replace all the work that could be done by machine. Strategic work involving thought, creativity and expertise we will leave to the lawyers. It is very similar to autopilot that takes over flying the plane so long as it is flying straight ahead, while the pilot performs complex tasks like takeoff and landing.

Lawyers have a reputation for complicating processes and adding on billing hours, while clients strive to simplify and expedite legal processes. How do you negotiate this conflict of interests?

A recently published article discussed Microsoft’s plan to “kill” the per-hour billing model for legal services and move to a new model based on qualitative, not quantitative parameters. Per-hour billing model promotes inefficiency. We want to save time and encourage law firms to compete with higher service quality. A law firm that continues to operate on an hourly billing model is ignoring the reality in which the client wants to get a good service and doesn’t care if it takes one hour or 100 hours. We aim to save time on repetitive, menial work.

How does the technology work?

The contracts that our system analyzes have repetitive legal principles. We use AI technologies such as machine learning and deep learning to teach our system the same legal concepts that interns learn on the job. We don't teach the system to recognize specific terms but instead feed to it a large number of examples and principles. The more of these examples we feed into the system the more accurate it becomes.

The system learns to recognize patterns and laws. For example, if I want to teach it how to recognize a non-compete clause I can upload all the variations available in legal text that refer to non-compete. We have reached a point where the system is more accurate than the average lawyer. We tasked lawyers working with us with analyzing a contract and they reached 85% accuracy while the machine reached 90% accuracy.

How does the system “comment” on a contract?

The client defines important points and the system compares this checklist with the contract and recognizes what is missing and what needs changing. We combine the system’s AI with human quality control to ensure the reports are accurate. Our system is capable of analyzing a contract in one hour while the average time it takes a lawyer to review a contract is measured in days, sometimes weeks.

Who is your target market?

Legal departments of companies that could save work now performed by interns and young lawyers. The idea is to make it possible to sign deals faster. The added value is that time saved allows more resources for lawyers to concentrate on important and higher quality work. We sell primarily in the U.S. to medium and large companies in the retail, insurance and financial sectors. We recently sold our system to White & Case LLP, one of the five largest legal firms in the world.