After gathering years of experience in law firm and service provider environments, AI Visionary Josh Kreamer is well versed in the many ways AI can make a legal department a value driver instead of a cost center. He brings this expertise to AstraZeneca and has helped transform the way they do e-discovery.
What is your role at AstraZeneca and how did you get into this field?
I went to Harvard Law School. When I graduated, I went to a big law firm—Steptoe & Johnson—and focused mostly on appellate litigation. I had one case in commercial insurance litigation that became my first exposure to e-discovery.
During that case, I realized pretty quickly that e-discovery was very much the future and I wanted to get involved. I'm very interested in technology and pushing the boundaries of what's possible.
So, from there, I worked at a couple of service providers, another law firm, and then joined AstraZeneca.
I’m the director of e-discovery AstraZeneca. I joined the team in 2018 as a business relationship manager—basically a go-between for IT and legal.
At that time, AstraZeneca did not have an e-discovery team. In fact, we had very few people who really knew e-discovery and the work was largely outsourced. We just weren't all that involved in the process.
Soon after I joined, I looked at our approach and said, “The problem is we don't have enough transparency. We don't have enough control, and we're paying too much for these resources. There's a better way.”
So I put together a business case to create an e-discovery group and I shopped it around the enterprise. Ultimately, finance decided to take us up on it. They could see the strength of the financial case, but they were also sold on my idea that e-discovery is an enterprise process. There's no reason it must sit in either IT or legal, especially because, from a capability perspective, it can reach so much more broadly if it's not pigeonholed within those functions. What we do in e-discovery is intelligently collect, filter, search, and review data. When you look at it that way, it's a much broader set of skills that can be applied to plenty of other uses cases.
That was in 2019. Today, we have four full-time employees, we are hiring four more, and we have six contractors. So we've got a pretty large and growing team and we're actively expanding our capabilities.
After realizing that e-discovery was the future, you spent time working at different firms and service providers before you started at AstraZeneca. Armed with that knowledge, what were your first priorities?
First is a focus on efficiency and, where possible, simplicity. I've long felt that the billable hour is not the client's friend. It can incentivize inefficient ways of working—including more manual work than is necessary, processes that are overly complex, or expectations that exceed real legal obligations. So I wanted to bring in that thirst for efficiency.
Second, I wanted to shift to a world where we had everyone doing what they do best and only that. What the law firm does best is represent us while leading the strategy in litigation, as well as things like motions and trial work.
For me, that doesn't include things like project management, technology selection, or implementation, or having to do the basic, repetitive heavy lifting those tasks entail.
By going instead to people who are more project management oriented and technology minded for those things, we get a better work product that is more affordable.
An example might be using robotic process automation. In that context, you apply different forms of machine learning and AI—for automated categorization, data sorting, assisted review—to eliminate the need for repetitive manual labor. In this way, technology can help us reach the efficiency we're looking for and free up our team to work on what they’re truly good at doing.
You are an early adopter of AI and using machine learning in e-discovery. As you were exploring utilizing AI in your work, what barriers or pushback did you encounter?
Thankfully, internally, I did not face a lot of pushback. There was some concern over whether our outside counsel would act as blockers—a fear that, if we've got technology performing most of the review work for 90 percent of the documents in a matter, we can't then have our outside counsel manually reviewing half of those, too, right? But would they want to anyway?
Ultimately, we found that there was a lot of suspicion among outside counsel over our expectation that the technology be used as much as possible. Specifically, there was suspicion about the quality of the technology. We encountered a lot of exaggeration of the complexity of it, so that it would be framed as: “Rather than engage this highly complex, difficult to understand process that may not be accurate, let's just do it the old way and have attorneys look at the documents.”
When you frame it that way, it suddenly sounds like it's just simpler to have people do the manual work. But you know, I've managed big reviews, and seeing 80 people in a giant rented room all clicking away at computers coding documents—there's nothing simple or efficient about it. Not to mention the consistency issues, which is actually where you run the biggest risk.
I have seen cases and situations where all these people were reviewing documents and, after looking over that output, it becomes clear that they might as well have just been picking codes at random. So, despite the fears, the argument that an AI system might not be as accurate really doesn't hold a lot of weight.
I believe that we have to move forward and try things and, ultimately, not let the perfect be the enemy of the good. Oftentimes, the mindset of attorneys is to plan everything out and think it through until it's perfect. Realistically, though, either perfect never comes or you could have learned a lot more by just jumping in.
I also believe in risk management—we can take smart risks and still be careful, even in legal, and indeed that exercise is part of the function itself. So we have to make sure that that we're being measured, but ultimately, I say let's just jump in and do it. Whether that means conducting a shadow project running in parallel, or trying something new on a low stakes matter where you can easily jump in and do it the old way if it's not working—just jump into action. This is how you lay those suspicions to rest.
Often, in a world that moves as quickly as ours, making a measured but quick decision can be more valuable than making the perfect one after a lot of time is wasted on hesitation.
We all want to make that decision faster so we can move to the next step, and technology can help us answer that urgent call to action. Especially now that the technology is proven to do what it promises, it’s better to just get our hands dirty and get to work.
Where do you think change has to come from? Is it organizations pushing their law firms and vendors to innovate and leverage more cutting edge technology like AI? Or do law firms need to adopt it and bring it to their clients?
It should be less about one trying to convince the other and more about both embracing technology as a starting point toward finding better solutions. On the client side, we don't need to push our law firms to embrace technology so much as we need to tell our law firms that we are embracing technology and, if they want to continue to be our firms, they will join us in doing so.
We just need to decide we're the client, right? A lot of in-house teams are largely staffed by former law firm attorneys, and I say that as a former law firm attorney. There's a feeling that we all have to be hand-in-hand on any major decisions around process. Sometimes that’s the right approach, but other times, if we've got smart risk management people in-house, we know what we’re doing. We can make the decisions ourselves, and if that's how the industry needs to be driven forward, then so be it.
Ultimately, I believe the law firms who focus on enabling that use rather than holding it back are the ones that are going to be successful into the future. Just as the legal function is an enabler to the business—especially in a business like ours, where IP is critical—law firms can be an important enabler to the advancement of technology within corporations.
Of all the wins you've had in your career, what are you the most proud of and why?
What I'm most proud of is being part of an e-discovery team, and leading an e-discovery team, where every member has the same mentality: our job is to make and sell medicine and the way that we contribute to that is by lowering the cost of legal work for the company. If we can reduce the legal budget by a dollar, we can put another dollar into R&D.
This year we've developed a per-unit metric for our e-discovery work, which enables us to measure both forward and backward. So we can now say, if we were still doing things the old way, our efforts this year would have cost $6 million more. And conversely, if we had done it the way we're doing it now in the past, we could have spent $20 million less over the last three years.
This data helps us solve for the volume problem, because we can’t just look at things as if maybe there could be less work to get done. In our space, that isn’t realistic. But if we improve the efficiency per unit across the board, we see huge payoffs anyway—and can’t that without technology. It's been a critical part of our strategy.
What do you do when you’re not working, and how do you decompress?
I actually don't have that many days where I come home and I'm just compressed. I love the work that I do and so I go home feeling energized most days, honestly. It helps to be in my dream job, surrounded by smart people who are looking to invest in good ideas. I have to give credit to the organization, and my luck at having joined a team with a culture that lets me maximize that feeling of fulfillment and free reign.
When I'm not working or spending time with my family, I'm thinking strategically about the future. That may sound lame. But I believe “What have you done for me lately?” is always a valid question, and always one that I should be prepared to ask, or that I should be prepared to be asked.
So I find myself planning the next thing, and the thing after that, and the thing after that.
To me, innovation often is either using an existing process in a unique way, or coming up with a unique process to solve an existing problem. And so I just enjoy looking around and considering the tools we have and how we could think differently about them.