The Center for Data Innovation spoke with Ilan Admon, chief technology officer of LawGeex, a startup with offices in Tel Aviv and New York that uses AI to review contracts. Admon discussed how LawGeex AI reviews some legal contracts faster and more accurately than human lawyers, how AI can minimize the amount of legalese in contracts, and how AI can reduce the amount of drudge work for lawyers.
Michael McLaughlin: How would a business use LawGeex AI to automate the contract review process
Ilan Admon: LawGeex uses AI to automate the review and approval process of everyday
business contracts. Think non-disclosure agreements, service agreements, and software-as-a-service contracts. Its AI reviews uploaded contracts, approving them if they match a pre-defined legal playbook or escalating them to the legal team if issues are found. LawGeex answers the question “Can I sign this?” within one hour, reducing legal bottlenecks and shortening contract turnaround time.
McLaughlin: Have you detected any consistent risks or trends in legal agreements that could help lawyers write better contracts in the first place?
Admon: One area that can certainly be improved is the use of hard-to-understand legalese, with software helping avoid its overuse either in drafting or review of contracts. Many organizations are looking to cut out the use of legalese and software can help. The LawGeex AI has been exposed to tens of thousands of legal documents. This taught the AI to understand the non-natural language of legalese. We are easily able to identify key concepts even if written in legalese; our technology allows the algorithm to identify concepts even if they were worded in ways never seen before. This makes it easier to understand what you are signing in a contract.
McLaughlin: LawGeex performed an experiment where your AI outperformed 20 lawyers at identifying risks in non-disclosure agreements. Did you expect it to perform so well? What does this mean for the future of NDAs?
Admon: This year, LawGeex revealed that its AI bested 20 top U.S. lawyers for the first time in accurately spotting risks in everyday business contracts. In this case it was non-disclosure agreements. The study, carried out in collaboration with top academics at leading universities, saw the LawGeex AI achieve an accuracy of 94 percent, while the lawyers achieved an average of 85 percent. It took 92 minutes for the lawyer participants to complete all five NDAs compared to only 26 seconds for the LawGeex AI.
We have spent four years training the LawGeex AI on tens of thousands of documents, and a great deal of testing. So the team was more than quietly confident our AI could handle any NDA contract thrown at it. We knew that we would definitely be faster and thought we could at least match the human lawyers for accuracy. However, that said, there is always nervousness going into a competition like this. Importantly, we often say that we are creating a solution for lawyers who do not want to do this sort of drudgework, and these type of everyday “in the way” business contract is not top of mind for lawyers. However this case was different: the 20 lawyers knew their core skills were being tested and that meant the lawyers upped their game and very much focused on the task in hand.
McLaughlin: Besides contracts, the legal profession generates enormous amounts of data. What types of data do you think are being underutilized by the profession and how could AI help?
Admon: There is a long way to go in terms of leveraging all the legal and business information contained in contracts. We offer, for instance, not just fast approval of contracts but also providing entirely new data never before available to lawyers and their organizations on their contract negotiation and approval process. We can provide data on contract approval process—which can take up huge amounts of resources in an organization. This includes data like top missing legal concepts, monthly usage report, and what clauses are holding up deals.
McLaughlin: How prepared are organizations’ legal departments to use AI? What needs to happen for the legal field to adopt AI on a greater scale?
Admon: This is an optimal time for organizations’ legal department to use AI—or in fact any processes that help save them time and money. There are multiple challenges facing organizations that need to be addressed—pressures to improve efficiency, a need for more smart lawyering, and growth of business conscious legal operations. With the advance of AI generally, businesses are looking at how they can implement enhancements including in departments such as legal.