At AI Research Company ROSS, A New Stage of Transparency and Engagement

Was this the same ROSS Intelligence that I’d wondered and speculated about?

For years, the artificial-intelligence legal research company maintained a shroud of secrecy around its product. Sure, they talked about it plenty. They garnered lots of media coverage. And CEO Andrew Arruda was a ubiquitous presence at legal conferences, touting the future of AI in law.

But when it came to seeing the product, that was another story. I’d pestered Arruda for years to let me review it, without success. At a conference of law librarians two years ago, Arruda was called out during a panel for his company’s lack of transparency. Surely somebody must have been seeing ROSS, because law firms were reportedly buying it. But, if so, I had trouble finding them.

Now here I was, on Thursday and Friday last week, sitting in ROSS’s Toronto research and development office, with unfettered access to its entire engineering and design teams. I was given detailed demonstrations of the product’s inner workings. I was invited to sit in on engineering and UX team meetings. I was encouraged to ask any question I wanted, of anyone I wanted to ask it.

And that shroud of secrecy around its product? ROSS had shredded it. In June, without fanfare, ROSS quietly changed its website to offer free trials of its product to anyone. The product that had been treated as if it were a state secret was now open to everyone for a 14-day free trial, without requiring even a credit card.

So what is going on here? The answer to that question starts with a look back at the company’s history.

A Rapid Rise

ROSS began in 2014 at the University of Toronto as a student-built entrant in a cognitive-computing competition staged by IBM to develop applications for its Watson computer – then famous for having won Jeopardy! ROSS’s prototype won the UT competition, earning them a write-up on the front page of The Globe and Mail – which touted ROSS as the future junior associate at Bay Street law firms – and serving as a springboard for the company’s rapid acceleration.

At the time, the oldest of the three founders, CEO Arruda, a University of Saskatchewan law graduate who was then articling at a Toronto law firm, was 25. The youngest, Jimoh Ovbiagele, a computer scientist who is now the company’s CTO, was 21. The third, Pargles Dall’Oglio, also a computer scientist, was a bit younger than Arruda.

ROSS went on to compete in IBM’s national competition in 2015 as the only non-U.S. college team. Although it came in second, its momentum was already established, further fueled by a series of click-bait headlines touting ROSS as the robot lawyer of the future.

Soon, the founders were invited to Silicon Valley to participate in the prestigious Y-Combinator startup incubator. Denton’s NextLaw Labs made ROSS one of its earliest investments. In 2015, they secured $4.3 million in seed funding and then, two years later, another $8.7 million in Series A funding. In 2017, Forbes named the three founders to its “30 Under 30.”

But over the past year, it had seemed to me that ROSS had gone quiet. Not literally, of course. But no longer was its formerly globetrotting CEO a presence at every legal conference. The company put out little in the way of news or announcements. Even the company’s social media presence seemed more restrained.

Then came the call from Arruda, inviting me to Toronto, and resulting in my visit there last week.

A Period of Refocusing

In Toronto, what quickly became clear to me was that the last year had been a period of refocusing and refinement for ROSS. It was a period that reflected, it seemed to me, a new stage of maturity for this still-young company.

The period started almost exactly a year ago with the announcement of the “new ROSS,” which marked the first time the legal research platform included cases from all federal and state courts and all practice areas. ROSS’s first U.S. prototype covered only bankruptcy cases, and as it had added other practice areas, it required users to choose one before submitting a query. With the new ROSS, users no longer had to choose.

While that “new ROSS” marked a turning point, it was not an end point. CTO Ovbiagele said they have worked hard to deliver a better legal research experience. Through the use of natural language processing and machine learning, they have developed a product that turns the research workflow on its head, he said, allowing lawyers to ask complex questions using natural language and get results that are highly on point.

“Legal research platforms have trained lawyers to start with broad results and then narrow them down,” Ovbiagele said. “We say, what if we brought you right to the decision on point, and then you can widen out your research from there.”

In pursuit of that goal, the last year has yielded several significant developments for the company:

• ROSS refined its research product to the point where it was ready to show it to the world and let users decide whether it does – as the ROSS people adamantly maintain – deliver more focused results than other research platforms.
• ROSS brought on Stergios Anastasiadis, the former director of engineering at Shopify and engineering manager at Google, as its head of engineering to lead further refinement and development of the product.
• ROSS refocused its marketing efforts away from larger firms, where it has long targeted its sales, to solo and small firms.
• ROSS introduced transparent, no-obligation monthly pricing, so that users are not tied to long-term subscription contracts. “Pay for the months you need us, then don’t pay for the months you don’t,” Arruda said.

The capstone to all this came last month with the announcement that Jack Newton, cofounder and CEO of practice management company Clio, had joined the ROSS board of directors. Newton is highly regarded in the legal tech world, not just for what he and cofounder Rian Gauvreau accomplished over the last decade in starting and building Clio, but also as a mentor and advisor to other startups.

A More Mature Company

For the three founders once featured in Forbes “30 Under 30,” the year brought one other milestone. CEO Arruda, the oldest of the three, himself turned 30. While symbolically notable, it also underscores the fact that these three once-inexperienced founders have now been at this for five years. They have matured, and so has their company.

In fact, Arruda told me last week, those early years were stressful for them. The founders had a vision to use AI to make legal research more affordable. But they had no experience in what it took to be entrepreneurs or to build a company.

On top of that, because ROSS was one of the earliest companies to focus on AI in legal research, there was no playbook on how to build its product. Although they developed their prototype on Watson, they built their commercial product from scratch. And they did so even as they were already caught up in the glare of media and investor attention.

“It was a huge challenge for us to build it from scratch,” Arruda said. “We were experimenting and innovating under the spotlights, because we had a lot of media attention.”

That explained the secrecy around the product, Arruda told me. They truly thought they were building something unique, and they feared that a competitor would steal it out from under them. So they kept the product close to the vest, until they reached the point where they were confident that it was ready.

Arruda also confirmed my perception that he had scaled back his public speaking over the past year. He did it in part to open opportunities for others in the company to speak, he said. But he also felt the need to focus more of his time and attention on being CEO and managing the day-to-day operations of his growing company, which has a sales and marketing office in San Francisco in addition to the Toronto R&D office.

There is no better way to get the pulse of a company than to move past its executives and marketers and meet with the people who actually design, build and service the product. Over the course of multiple conversations last week, one point became clear to me. Those on the ROSS team truly believe they have created a better legal research tool. And they are not satisfied with stopping there. They have a vision of a product that does not just deliver better results, but that will someday also help lawyers make better sense of those results.

Due largely to AI, legal research is becoming one of the most vital and competitive markets in legal technology. Whereas ROSS was one of the first to tout its use of natural language processing and machine learning, now many others do. Going forward, the biggest challenge facing ROSS will be to stand out in a somewhat crowded field and win market share from long-entrenched rivals.

But in Toronto last week, it seemed clear to me that ROSS has entered a new, more mature phase in its growth, one marked by transparency around its product and pricing and engagement with its customers. My presence there was evidence of that. In the legal research market, that approach makes a lot of sense.

After all, the best measure of a research product is the results it delivers. By lifting the shroud and inviting anyone to try its service for free, ROSS is declaring itself ready for the challenge.

(BTW: Watch for a forthcoming LawNext episode recorded with Arruda and Ovbiagele last week, as well as for my review of the ROSS research platform.)

(Full disclosure: ROSS is reimbursing me for the air and hotel expenses I incurred in traveling from Boston to Toronto.)