An Interview with Basit Mustafa, CEO of Voltaire
Voltaire makes software that aids lawyers in jury analysis. The software has been used in numerous trials, and was recently featured in the ABA Journal.
Jacob Bradshaw: So why don't you start off explaining sort of generally what Voltaire does and how it works?
Basit Mustafa: Voltaire is a tool that helps lawyers make winning decisions… Our first product automates research and analytics of potential jurors during voir dire. Basically, we saw an opportunity to improve the speed and breadth with which research is done, but that's only half the problem when you have a jury pool of a hundred people in front of you. [The problem is] not just the speed which you can research in, but [also] the speed with which you can make meaning of what all that data means. So, Voltaire first not only does the research when you have a jury list in front of you, [but] with it you can [also] automate [matching?] public records, social media, anything that you would normally research in addition to some of our own proprietary data sources, including online petition signatures, etcetera, and then run them through our analytics system that [answers the question] "what does this all mean?" Once we've found all of these data points on the person, what do they holistically all mean together in context, and in the context of the case? You know, someone who might be a second amendment supporter could be good for one side and bad for the other… [W]e begin to make some of those judgments and calculate them with analytics using artificial intelligence.
Jacob: Cool, and so this all happened before voir dire begins? And then [the lawyers] go in armed with all of this knowledge that Voltaire provided?
Basit: Absolutely. In some cases... the jury list isn't even provided until the morning of, or as you're walking into court, so those become obviously even more challenging voir dires, but a fair amount of our customers... [inaudible] in advance, and have sometimes the evening to peruse the results. But sometimes its the other way, sometimes they have ten minutes as they're walking in, so in that case it becomes even more [inaudible] although they don't get as much time to review the source data, they can distribute that top sheet that we give them.
Jacob: Let’s say they've done the questioning... can they plug what they hear in questioning back in and get... updated results based on the new information?
Basit: We actually love it when our customers do that, because it gives us a sharper pencil to work with. Anytime that you have a large city with common names, anytime people can come back and give us more information always helps. And they can also put their own judgments in, absolutely. If they learn new things, or if they have their own evaluations, they can always override the system too, and that's an important thing to note... a computer can always make a human better at what they do, but I don't believe, at least in the law quite yet, that we should wholesale be replacing gut judgments that humans have with computers. Instead, we should be using computers to make it better - the perfect marriage of man and machine if you will.
Jacob: Excellent. Well, you went ahead there and answered one of my later questions, that's good. [My question was going to be] about the future of technology, and whether you think Voltaire might replace lawyers or just help them… it sounds like it's the good one - I'll still have a job when I graduate!
Basit: Well, I think it's definitely going to change it. And I think there's a lot to discuss there too, because, when I say that, I say that is the current state of the art. I am sure the state of the art is going to evolve and continue to get better. I [have seen] some very interesting things happening in legal technology just in the last three or four years, and if that rate of innovation continues... I can definitely... and continue to evolve...
Jacob: So… do you have other products that you are planning to develop… to address new and emerging issues that you think… [will] come up in the legal field, or are you focused right now on [Voltaire]?
Basit: We are pretty focused on this product, but we do have our eye on new applications beyond just voir dire, for our existing technology. Our core technology really focuses on risk-assessment, and specifically risk-assessment when it comes to people, and allowing lay individuals, and when I say lay individuals, I mean non-technical people, so professional lawyers who don't necessarily have... the technical background to do some really sophisticated risk analytics that would otherwise require a PhD in data science. So there are lots of different areas you can apply that to beyond jury selection. You can apply it to depositions, you can apply it to negotiations, you can even apply it to what kind of customer that you, or a client that you decide to work for, and how you work with them. So, there are a few new products that we have in the pipeline for early 2018, that do focus on things that are beyond jury selection, that are focused on risk analytics, especially as it relates to people.
Jacob: Cool. Do you have a lot of competition in the field right now, or do you... feel like you're... leading the charge at the moment?
Basit: I definitely feel like we're leading the charge. We are the first to the market with a jury selection product that uses artificial intelligence and machine learning. In fact, we're still the only one in the market that uses artificial intelligence and machine learning. Our biggest competition honestly I think is, people trying to do this manually. There are a few other software offerings out there, but like I said they don't incorporate artificial intelligence or machine learning in the way, [or] in any way, especially not the way that we do. But there's a second component that involves people who want to just throw people at the problem, and that I would say, that's our main competitor, is putting a bunch of first-year, second-year associates in front of laptops [conducting voir dire research]. Where we think we obviously can help that - we don't necessarily want to do away with that - we just want to make those first and second year associates better.
Jacob: Do you feel that that's sort of where the AI comes in, is that it can sort of work in a more human like way at processing all this information?
Basit: Absolutely. Because there's such a... fire-hose of information and allowing the AI to at the very least do the triage as to "who are the people in the jury pool who need a human eye to look at them" and "who are the people that we can instantly understand as high risk, and this is not even a question" and then you need to focus resources, where a human gut... is the only thing that can really, in the current state of the art, make a practical or good decision. I think that that's absolutely the way it goes, the old 80-20s rule, you know, 80% of the work and [inaudible] done by the computer, and the end 20%, should be done by a human, so we should absolutely let the 80% of work, that should be done by a computer, be done by a computer so the humans who are working on the 20% are more effective and more cost-effective, and more efficient.
Jacob: So you're sort of giving the tools and information... that they need all right there and then the person can use those tools and look at that information and make the final decision?
Basit: Absolutely... and we do... really venture into some advanced artificial intelligence, so... we do attempt to make some decisions... for the human, but in such a high-stakes game, like trial litigation, we think ultimately the human should be in control, so our philosophy and our product values really hinge around "if we can make a decision we're going to show it to you, but we're always going to ask you" and give you the information [for] a human to either override the opinion etcetera... you know... this can't be rise of the bots in court quite yet.
Jacob: I read that... the software is able to do all this because... I guess you licensed IBM's Watson? I didn't realize that was something one could do... how does that work?
Basit: So IBM sells a variety of artificial intelligence tools, so IBM Watson is something that you can, through... their IBM cloud Unix platform, be able to use the Watson super computer, and we do leverage several of their services to provide artificial intelligence and to provide judgement [inaudible] in a jury pool. Anyone with some level of programming skills and a credit card can go to IBM and purchase the right to use the service... You still have to build an entire data model on top of it, so it's not just as easy as having IBM do all the work. The analogy would be "anyone can go buy a hammer, that's built by Stanley... but to build a home you should still have the skills of a carpenter."
Jacob: Yeah. Or... like having a development engine (sic - I meant "environment") or programming language and then having to actually build the software from that.
Jacob: Can you talk about maybe a specific case or client where they used this and it sort of provided them with an edge they wouldn't have thought of otherwise?... I don't know... to what extent you know or are able to talk about that?
Basit: We have some non-disclosure agreements in place here, we're very, very protective of our agreements with our customers, but I'd be more than happy to discuss a couple of them, with actual permission to use them... but I won't use their name... as they've asked... We were working for the plaintiff in a commercial liability case. In this situation, it was a road accident that had involved a commercial vehicle. Our software was employed by the plaintiffs and we were able to find a significant amount of individuals who had expressed tort reform opinions, and even a few of them who themselves had been in accidents, but those hadn't been disclosed, for whatever reason, on the questionnaire... We don't know exactly why those weren't disclosed, if it was an honest mistake, if the question wasn't worded properly on the questionnaire, we can only speculate. But we were able to surface a lot of that data and be able to guide our customer in using their peremptories, in identifying individuals who had anti-tort reform, or who had tort-reform sentiments, and believed in limiting damages.
Jacob: Cool, I just have a couple more questions if there's time.
Basit: Absolutely, it's my pleasure.
Jacob: So, how much would you say, if you compare how long it took beforehand to trawl through all this information... with just a handful of law students that you have on the matter or whatever, versus Voltaire? How much faster do you think Voltaire is?
Basit: I really think there are two ways of looking at this. Number one is that I think we make much more possible. So, to have a [inaudible] comparison, holding all other things equal, I think that we're about 10 times faster than a human doing the same research. And that's just to get the data together. Assembling the data into a single report. Now, the question becomes I think we're even... much faster when you introduce the second step, which is actually having to read through every single social media post, look at every single image - a computer can do that hundreds, thousands times faster than a human can. If I gave a human 3,000 images to look at, and say I need you to do object recognition in these, find all the images that have guns in them, a human is going to take a long time and have a very high error rate in 3,000 images. By the time they get to the tenth image, you know, their eyes are glazed over. A computer can do it in milliseconds. So... in that workload, I would say that we're hundreds of orders of magnitude faster than a human. And in some cases, we do things that humans can't even do.
Jacob: So... this is such a huge idea, and as you were saying you're sort of the first in the market... can you talk about what inspired you to do this... how it all got started and how you put this whole company together?
Basit: Absolutely. So my "aha" moment was when I had friends who are in the law - I actually wanted to be a lawyer myself, but when I was choosing my career path when I was young... when I was eighteen years old the tech boom was going on... and IBM actually dangled a large sum of money in front of them to come work for them, and... financially it was untenable [for me] to take on a whole bunch of debt to go to law school versus achieving my goal of relative financial security as a young person by [taking]... the IBM offer. So fast forward about seven or eight years, I'm still a law geek in the sense that I love the law, and I read about it, and I hear from my friends who are lawyers, practicing attorneys, I'm also on the board of directors of the ACLU of Colorado, and then I'd hear from the attorney's at the ACLU, that we won this case or we lost this case because of this juror. And it was fascinating to me when they would tell me the psychology behind it and the process behind it and their frustrations behind it. And that's where the "aha" moment was, was in the frustrations. And hearing the frustrations of my collective friends and professional network of attorneys, saying jury research is a pain, it's non-specific, its... stressful, and I though "hey I'm a tech guy, and you know, have a decade in the tech industry" I think we can do this better. And I prototyped it, and put in front of some of my friends, had some challenges, had some failures, had some successes, all [through?] beta testing at the end, eventually turned into a company that it is today, by going customer by customer.
Jacob: Alright, well, here's the last question for the podcast, and that's: Do you have any advice for anyone else out there who has, we'll say some ideas that don't compete with yours but affect a completely different area of the law, but are looking maybe [at] how to develop that, and do you have any advice for them on how to succeed as you've done?
Basit: Absolutely. I think one of the most important things, and I alluded to it when I was talking about how I started the company, [is] having the success and failure, and pushing hard enough to fail, specifically in the areas of the customer experience. And I think this is important in any profession, in any industry and any segment, but in the profession of the law, I think it's even more important, that especially technologists understand that the user experience and the customer experience is important. That means delivering the data in the form that it is useful, insightful, and actionable, that is free from jargon, that it follows and respects the workflow of the customer, and that it is something that makes the customer not just happy, but effective. The customer should not have to adapt to the system, the system should adapt to the customer. And again, I said that I think that's true in any industry, but in the law I think it's even more important, because of the stress, what's at stake, and because of the type of individuals who become lawyers. Typically [they] are very intelligent, are typically masters of their domain, especially in the way that they know how to do their work, and I think it's unreasonable to ask them to change their workflow and their patterns simply because the software demands it. So I think that that... would be my number one advice, is, push that limit, and understand what works and what doesn't, and respect the user experience of the customer above all else.