Francis Scialabba


How lawyers embraced the robots

Long resistant to automation, 2020 was an inflection point for the industry’s adoption of legal tech

· 6 min read

“Does this count as an ‘act of God’?”

That's the question legal teams everywhere were asking when Covid hit the US last March. Clients wanted to search out force majeure clauses in thousands of real estate agreements and other contracts, wondering if a pandemic could render them null and void. Not only were there more docs to review than usual, but also lawyers had to find quick answers to critical, unprecedented issues...all while working from home.

The 100-person team at Luminance, a UK-based AI-for-legal startup, felt like they were at the center of it all.

Founded by University of Cambridge mathematicians in 2015, the company specializes in automated legal doc review and analysis. Although the tech lends itself to a range of legal specialties, Luminance's business largely came from M&A due diligence work in its first four years. All of that changed in 2020, when M&A activity fell sharply and law firms scrambled to use the tech for new tasks in more than 30 different specialties, including property portfolio analysis, contract negotiations, compliance, litigation, and investigations—in order to save money. Overall, business spiked 40% in 2020.

What happened at Luminance is an illustration of a broader shift toward the automation of corporate legal department work. And experts say what’s happening in the legal industry is indicative of the direction many knowledge work industries will go. Automation is often tied to conversations about manufacturing and wage work, but knowledge work automation is no myth.

“Whether it’s the white collar worker or...the laborer, everyone’s going to go through this, what I call ‘automation journey,’ whether they like it or not,” Suneet Dua, US chief product officer at PricewaterhouseCoopers (PwC), said. “The automation train around RPA, robotics, workflows is moving so fast—and there’s low-code, no-code automation. That train is like a speed train. [The] human-skills train is the slowest train out there.”

Stanford University research found that, after cross-referencing more than 16,000 AI-related patents and 800 job descriptions, “knowledge” sectors with highly-paid, well-educated workers may be more susceptible to automation than blue-collar jobs. For example, workers who completed a bachelor’s degree would be exposed to AI at least 5x more than workers with only a high school diploma. For evidence, look no further than the growth of robotic process automation (RPA), which uses neural networks, computer vision, and more to automate rote tasks. UiPath, a leading RPA startup, grew 10x between 2018 and 2021, thanks in part to clients ranging from Google to Equifax.

Voyeur → Actor

In the legal sector, the first stage of automation was centered on the logistics of managing massive amounts of physical documents and data, like converting them into digital files. Now, due to advancement in automation tools and user experience, we’re entering a new stage of automation.

“Now, we’re moving into an environment where unstructured data can be captured and more complex decision-making can be supported through automation,” says Chris Audet, a senior research director at Gartner, who researches in-house legal teams.

Several companies, like Luminance, are tackling this with machine learning (ML) models. For example, Luminance's product is a blend of supervised and unsupervised ML—meaning the model undergoes some training but also uses a learn-as-you-go approach, for “finding unknown unknowns,” Luke Taylor, a subject matter expert for the company, said. Its clients include one-fifth of the world’s largest law firms and all Big Four accounting firms: PwC, Deloitte, KPMG, and Ernst & Young (EY).

The supervised ML part works like this: When a lawyer interacts with one part of the contract (say, a problematic clause) the model can apply that interaction across the entire document pool (e.g., flagging any similar clauses for additional review). The unsupervised ML component, on the other hand, is all about independent pattern recognition. Without learning from a lawyer or using predefined terms, the model can analyze a vast set of documents, find the standards and deviations, and flag anomalies for a lawyer’s eye.

Keep up with the innovative tech transforming business

Tech Brew keeps business leaders up-to-date on the latest innovations, automation advances, policy shifts, and more, so they can make informed decisions about tech.

The product is also meant to be ready right out of the box: After about an hour of setup, Taylor said, a legal team can start analyzing documents for language patterns with no coding or configuration.

This leap forward in tech prowess, and other industries’ widespread adoption of automation, has made legal teams move from a passive to an active stance regarding AI tools, according to Audet.

“[In] the past, automation, AI, advanced analytics were voyeuristic. That was their attitude toward it: ‘I’m kind of curious about it, I want to see how other teams maybe use it and I can evaluate it,’” says Audet. “That’s not the case anymore.”

Audet said 2020 was an inflection point: Legal teams lost staff, either through business contractions or attrition, and to help offset those losses, some earmarked leftover budget dollars for automation solutions.

“Everyone was waiting for the first mover to make a move, and no one was—and now, we’re in a place where everyone is playing a bit of catch-up,” says Audet. “Covid kind of hit the reset button for folks, and what was [merely] of interest before now is actually seen as a core way, and a smart way, to get work done. So they’re asking questions like: ‘Help me build the business case for this?’ ‘Show me what ROI looks like?’ ‘Where has this been done successfully across workflows?’”

According to Audet, in-house legal teams tend to see automation as an opportunity to save cognitive capacity for critical thinking, rather than use up their processing power on high-volume, low-stakes decisions.

Audet added: “For decades, in-house legal teams have been bombarded by business requests that are not high-value...All of that has led them to feel like they’re drinking from a firehose, in managing that volume. ...People are already facing burnout, let alone lawyers [who] are told to review contracts for 75, 80 hours a week.”

Taylor echoed the burnout issue, especially for junior lawyers.

“One of the reasons why I didn't go into a law firm, in the end, is because I was very dissuaded by that grunt work,” Taylor said. “When you are a trainee going through this...and you’re training up to become a lawyer, and you’re looking at just the same contracts over again, it's not really helping you too much.”

Besides saving cognitive capacity, for some legal teams the tech also offers a way to take on more—and often higher-value—work. In Luminance’s case, two global law firm clients working on large document reviews used automation tech instead of temporarily pulling other teams off their current projects. One was able to speed up the review time for 190,000 German employment contracts from 30 weeks to just two.

“You won't hear many lawyers say, ‘I think my job’s going to go away,’” Audet said. “In fact, I haven’t spoken to a single one of them who feels that way. They actually are looking forward to this as a way to focus on the things that they believe they’re actually getting paid for.”

Keep up with the innovative tech transforming business

Tech Brew keeps business leaders up-to-date on the latest innovations, automation advances, policy shifts, and more, so they can make informed decisions about tech.