Why Walmart’s Former Head of Discovery Left for an AI Startup

Earlier this month, Aaron Crews departed as Walmart’s global head of e-discovery to join TextIQ — in his own words “a little AI startup” — where he is general counsel and vice president of strategy.

To explain why he is leaving a company that reported $485.8 billion in revenue last year to join one that claims to have millions in revenue, Crews said he believes several changes are imminent in terms of how is data managed.

Right now, at most companies, discovery costs are still rising, he said. But that’s only because data volumes are growing so fast, Crews added.

But by some measures, technology is already helping reduce costs: notwithstanding rising data volumes, costs, on a per document basis, are actually declining, he said. And in the next five years, Crews said he expects companies will adopt new technology to rapidly sort large data sets in ways that will minimize discovery costs.

He’s betting that TextIQ is making the technology that will further reduce costs. Founded in 2014, by two graduates of Columbia University, the company’s main product is called PrivIQ, and aims to help companies locate attorney-client privileged documents from large data sets.

“We built machines that can go through unstructured text and can understand language, people and their relationships,” said Apoorv Agarwal, co-founder and CEO.

Agarwal said his technology looks at relationships between people: It categorizes data based on information flows and is able to locate key documents, such as privileged documents, much more efficiently as a result.

Right now, Crews said most of the best technology for sorting data isn’t really being used for e-discovery. He said TextIQ will have applications for discovery in litigation, and also data review for due diligence, antitrust submissions, compliance and human resources investigations and other data review processes.

He said he expects companies to begin adopting new technology in three to five years, because in the meantime they’ll develop strategies to manage their data: They’ll cull data in real time and preserve less and less.

One of the reasons e-discovery costs are so high is there’s massive amount of garbage you have to sort through … because people keep it,” said Crews. “Businesses are going to get smarter, they’re going to purge it.”

The only part of e-discovery that will remain the same, and be used in other contexts, is the “thought processes” for “how to identify the data you care about: Who has it? where is it? How do I look at it and understand it?” he said.

Gradually, as companies upgrade computer systems, new technology — like TextIQ, he hopes — will replace whatever is currently in place.

“I think smart machines are going to be covering the bulk of the sorting process,” said Crews.

Write to the reporter at gabefriedman@outlook.com.

Write to the editor at csullivan@bloomberglaw.com.