As AI Hits E-Discovery, Lawyers Go From Big Law to Boutique

Last week, Redgrave LLP, a leading e-discovery and information-law boutique, announced the arrival of three new partners—Robert Keeling, Ray Mangum, and Kristen Knapp—along with seven other lawyers in Washington, D.C. The group came from Sidley Austin, where Keeling founded and co-led the e-discovery practice.

Earlier this week, I interviewed Jonathan Redgrave, who co-founded the firm in 2010, and Keeling about the move. It’s major news in the e-discovery world, where both Keeling and Redgrave are prominent practitioners. We also discussed broader trends in e-discovery and information law—including, of course, the impact of generative artificial intelligence.

I wondered whether the move of Keeling’s team from Sidley to Redgrave reflected a shift in the e-discovery space away from Big Law firms and toward boutiques. After praising Sidley and his former colleagues, Keeling said that yes, “We are seeing increased specialization in e-discovery.”

Here’s how that works at Redgrave. Unlike large, full-service firms, Redgrave doesn’t have a wide range of departments such as antitrust, commercial litigation, or white-collar defense. Instead, Redgrave partners with other firms, which it calls “merits counsel,” and those firms—Big Law firms, midsize firms, or other boutiques—handle the relevant substantive law. Meanwhile, Redgrave handles the e-discovery and information-law aspects of the matter.

This intense focus on e-discovery and information law gives rise to certain advantages for Redgrave. As Jonathan Redgrave told me, “Because we focus on this, we can be better at it.”

“E-discovery is getting more and more complex, with the volume and complexity of data growing exponentially, year after year,” he said. “For firms already covering so many other areas of law, does it make sense for them to also focus on what we do?”

Redgrave works with numerous Am Law 200 firms. Unlike working with another full-service firm, collaborating with Redgrave doesn’t create the same level of risk in terms of losing work to a rival. As Redgrave put it, “We’re not a threat: We’re not going to steal their antitrust or products-liability or white-collar work.”

Sometimes Redgrave is brought into a matter by full-service firms. More frequently, however, it’s hired directly by clients—often those that previously worked with Redgrave on another matter.

Since its founding, Redgrave has expanded along with the size and scope of the e-discovery industry. After the arrival of Keeling’s group, the firm has almost 50 lawyers, plus 18 other legal professionals with expertise and advanced degrees in fields such as software engineering and information science. Redgrave sees their expertise as increasingly valuable in e-discovery, especially as generative AI reshapes the field.

“Gen AI is a transformative technology, and the impact on e-discovery over the next several years will be significant,” Keeling predicted. “We’ve already seen that Gen AI-based document-classification technologies have the potential to outperform traditional machine-learning-based approaches to Technology-Assisted Review (TAR).”

Keeling said that if generative AI is used correctly, it can find a higher percentage of relevant documents and reduce the percentage of irrelevant documents sent to review.

“The main barriers at present are cost and speed—at this point, I don’t think it’s practical for most large-scale reviews,” he said. Although reviewing documents using generative AI tools currently on the market is faster and cheaper than human review, it’s still not as fast and as cheap as TAR, according to Keeling. And if the user of generative AI gets the prompt or instructions wrong, leading to unreliable results, the tool will need to be rerun through the document set—using up additional time and incurring additional costs.

“But we’re also seeing lots of competition and innovation in this space,” Keeling said. “I expect rapid improvement on both of those fronts.”

So generative AI creates opportunities for e-discovery—but it also creates challenges. For example, take deepfakes and other documents that can be generated using AI. How do lawyers and judges authenticate evidence when it’s increasingly difficult to separate the genuine from the fabricated?

“It’s really going to be a hard problem,” Keeling said. “The deepfakes continue to get better and better.”

Keeling doesn’t expect that deepfakes will be a significant issue in a typical discovery response from a corporate defendant in civil litigation, which will largely consist of documents generated in the ordinary course of business. But there are going to be cases where a key piece of evidence is in question and there will be doubts about its authenticity, he said. “You can imagine the need to bring in competing experts to help the court resolve the dispute.”

Privacy is another challenge for lawyers working in e-discovery. A welter of laws and regulations focused on privacy, from the EU General Data Protection Regulation to the California Consumer Privacy Act, makes it more difficult to work with data—and creates potential legal liability for mishandling it, for both law firms and their clients.

“One of the biggest difficulties is that collecting documents in places with strict privacy laws can be more complicated and take longer as a result—for example, when dealing with EU custodians in a US matter,” Keeling explained.

Ensuring that everything is done correctly when handling data, as e-discovery continues to become more complex and challenging, explains why Keeling isn’t worried about generative AI reducing the demand for e-discovery services or the opportunities for lawyers in the field.

“It’s very tempting to make a bold claim that Gen AI will eliminate first-pass responsiveness and privilege review, such as that often performed by contract attorneys today,” Keeling told me. “But if you look historically at other transformative technologies in the legal space, it hasn’t always played out like that,” he said.

The creation of case-law databases eliminated the need for law firm associates to go to the library and scour physical books, Keeling said. “But the ease of searching vast numbers of cases also changed expectations: Associates were suddenly expected to find all the relevant cases, not just those their library had a copy of.”

“It’s not that AI will replace humans. Instead, humans who are efficient with AI will replace those who are not.”