Leave a request
Text IQ chief executive Apoorv Agarwal says his software's job is to spot a needle in a haystack – but a costly one. Make a mistake in discovery during litigation, and a company can face sanctions of tens or hundreds of millions of dollars, he says. "It's a high-stakes needle."
Like other startups, Text IQ's raised funding to solve that problem. Unlike others, Text IQ is profitable. And for its first outside funding, it's taking only about $3 million from top investor Floodgate and a group of veteran legal counsels in a seed round its founders say could be the only money it ever needs.
The tech industry’s interest in legal discovery isn’t new. A quick web search will yield a host of “eDiscovery” options that promise to speed up and cheapen the process, in which a company pays a gaggle of attorneys to look through documents to flag what's privileged or irrelevant. Text IQ says it's different as the only one to really do so through AI.
“There are a lot of companies and a lot of hype around AI,” Agarwal says. “We are trying to attack this problem from a different perspective."
In a super-hot market for AI that's leading some to quip that academics can now raise funding on just a concept, Text IQ started making money instead. The startup is profitable, with January revenue of 10x its burn rate, says cofounder Omar Haroun, and sales expected to be in the millions for the quarter. Customers saved $3 million in legal expenses so far this year, the company says—not including the much higher costs of bad outcomes averted by its use.
The heads-down moneymaking impressed self-described “jaded" investor Mike Maples, a perennial Midas Listerwhose firm Floodgate led the seed investment in Text IQ. “Everyone talks about how they will be tomorrow’s unicorn before they’ve done anything,” Maples says. “These guys had real customers. I liked that they solved the problem without throwing money at it.”
Text IQ took funding at all to gain access to Floodgate's partners and get legal experts skin in the game, its founders say. That includes Dan Cooperman, the former counsel of Apple and Oracle, as well as Randy Milch, former counsel at Verizon.
The origins of Text IQ come out of academia, inspired by PhD research that Agarwal conducted at Columbia University that could scan a 19th century British novel and map out the characters’ relationships. Agarwal's research had been funded by the National Science Foundation and DARPA, the startup investment wing of the Department of Defense. He and Haroun then built the beginnings of their business from some cash from friends and family.
While many eDiscovery services flag messages that come from law firms or include sensitive keywords, Tex tIQ deduces links between parties that aren’t spelled out in the text. That could be 10 medical documents that when combined allow a reader to reasonably infer the identity of the individual involved, says Agarwal, or spotting attorney-client privilege just from the tone a lawyer typically uses.
Testing Text IQ’s claims is simple: clients simply conduct a bake-off, running the software on a series of documents already used by their existing service for a discovery process. When Twitter tried the product, Text IQ caught all the documents its outside reviewer did, but also more privileged documents that would’ve slipped through, says Wendy Riggs, senior manager of eDiscovery and litigation operations at Twitter. “The results were great.”
Text IQ has won every bake-off its conducted so far and already works anonymously with some of the biggest companies in the country, says Haroun, its COO, with contracts running into the hundreds of thousands and millions of dollars.
One difficulty facing Text IQ now: how to add new services without moving too fast and tripping up. Customers like Twitter and investors get excited envisioning how to use Text IQ for more functions than just discovery. At Twitter, for example, Riggs would like Text IQ to work with all the legal team’s document workflow, not just checking for privilege. Floodgate partner Arjun Chopra, who led the investment for the firm, agrees that’s a major opportunity, but says he’s cautioning the startup not to rush. “Part of the challenge they are facing is there is need across the board,” says Chopra. Move too fast, and Text IQ could become players in a few use cases and masters of none.
With more than 70% of its team PhD or masters degree holders in computer science, Haroun says over-selling isn’t wired into Text IQ’s DNA. “A big part of this is, from a technology standpoint, with less than 20 people you can accomplish what companies with 2,000 people do.”
Leave a request