Lexis Unveils New Agreement Analysis Tool

New AI-driven tool takes aim at the most time-consuming pain point in transactional work.

contract

(Image via Getty)

As a fresh-faced summer associate, I remember one of my firm’s senior transactional partners telling the assembled class that practicing law isn’t like anything we’d done before. “Plagiarism is you friend!” he declared. It hit the ears of overachieving students like a lead pipe. No one had ever spoken warmly of the p-word before. But for a global corporate practice sitting on an internal database of prior mega deals, he wanted the team to understand that we weren’t meant to reinvent the wheel, we were meant to draw upon the firm’s amassed wisdom.

This episode stuck with me even though I went into litigation. Drafting agreements reminded me of building an elegant puzzle, scouring time-tested language to find the wording that best fit the deal at hand while getting yelled at by 24-year-old frat bros from Lehman Brothers. But it still required a strong internal database and the time and energy to parse through it all looking for the perfect jigsaw piece.

In July, Lexis+ users will get a new tool aimed at speeding up that process with Agreement Analysis.

Agreement Analysis uses AI to extract and recommend alternate clause language from seven million clauses publicly filed in SEC EDGAR documents, and from Practical Guidance sample clauses, templates and agreements. From there, attorneys can quickly compare alternate language for the most highly negotiated clauses in their document, and leverage drafting guidance and insights from Practical Guidance and benchmarking data points from Market Standards, to effectively customize clauses for a transaction.

Everyone is focused on generative AI as the hot new thing, but extractive AI still has its uses. The robots don’t need to write new clauses, they just need to find the right language that humans already wrote. Then rewrote. Then fought about at 3 in the morning. Then rewrote. Then threatened to walk out over. Then rewrote. Then returned to the original wording.

At launch, the tool focuses on M&A, stock purchase agreements, and asset agreements.

Sponsored

The user interface is straightforward. Drag the agreement over for analysis and the system detects key features and breaks down the draft into its constituent parts.

AA Screenshot Dashboard

From here, the user can go point-by-point through the agreement. For every existing clause, the AI-driven tool will search both the database of nearly 7 million EDGAR filings and Lexis’s internal Practical Guidance to provide alternative wordings that the user can draw from. The algorithm is designed to provide diverse alternatives so the user isn’t getting 13 carbon copies of the same language, allowing the lawyer to see true variations.

AA Screenshot Document Analysis Tab

A filtering process gives the lawyer more control over the suggestions. Limit the search to agreements involving specific firms or have it produce only options deemed “pro-buyer” language.

Sponsored

A different window allows an attorney to recraft the language directly within the Agreement Analysis. Or not an attorney… contract management folks do this work too without being attorneys.

AA Screenshot Compare and Edit

After playing with the language, the user can copy and paste it back into the Word original — no changes can be made directly within Lexis to avoid all the risks that could create.

Plagiarism is still, in fact, a transactional attorney’s friend. This seems like a much more efficient way to find and evaluate the model language that best fits the client’s specific deal than reading 300 old agreements kicking around the document management system hoping to strike gold.


HeadshotJoe Patrice is a senior editor at Above the Law and co-host of Thinking Like A Lawyer. Feel free to email any tips, questions, or comments. Follow him on Twitter if you’re interested in law, politics, and a healthy dose of college sports news. Joe also serves as a Managing Director at RPN Executive Search.

CRM Banner