Predictive Coding: Serving Corporate Counsel’s E-Discovery, Compliance And Other IM Needs

Thursday, February 23, 2012 - 18:47

The Editor interviews Craig R. Carpenter, Vice President of Marketing at Recommind Inc.

Editor: Recommind has had a big year. Please bring us up to speed on what’s happened in 2011.

Carpenter: 2011 for Recommind was really a watershed year. We had spent the preceding years building up our infrastructure, getting our name out there and laying the foundation for rapid growth. That growth started hitting in 2011.

We effectively doubled our revenue, growing it by 95 percent over 2010. Having set ourselves up to continue that growth, we expect to maintain a similar rate for some period of time – at least until the law of large numbers catches up to us.

The biggest engine of growth was around predictive coding. Surprisingly, the majority of the growth we saw wasn’t really on the law firm side (which grew nicely), but it was the enormous growth on the corporate side.

Today, corporations are leading the effort to embrace powerful technology and bring it behind the firewall. Outside the firewall, corporations are pushing their outside counsel to aggressively embrace leading-edge technological approaches. One example of robust growth is that our cloud-based business grew roughly 228 percent last year over 2010.  

Editor: What were some of the highlights of LegalTech 2012?

Carpenter:  It is ironic that three years ago, we were the only ones talking about predictive coding. Fast-forward to LegalTech 2012 where there were at least six sessions on predictive coding – and we conducted three of them. There were ten to fifteen vendors who announced something that they were calling predictive coding.

We patented predictive coding and we are the clear market leaders in it. To this day, many people don’t understand what it is and how it is done, but this will change with increased awareness of predictive coding use by our customers.

Editor: Tell us about some of the sessions Recommind conducted at LegalTech.

Carpenter: Our three different sessions on predictive coding comprised half of the predictive-coding sessions held at LegalTech. Of course, this is to be expected given our role as market leaders.

We sponsored the only session that dealt with predictive coding from the corporate counsel’s perspective. The panel included representatives from Google, Nationwide Insurance and Raytheon. Their perspectives were shaped by handling a wide spectrum of cases.   

The one thing that concerned all three of these corporate giants was the need to get to the heart of matters quickly – and this is the reason why predictive coding is so exciting for them.

Editor: What distinguishes these three corporations from others?

Carpenter: They all recognized the perils of the “Frankenstack” approach, which arises from the fact that the e-discovery industry is a cottage industry. Ninety-nine percent of the vendors have approached it over the years from the perspective of "I’m going to have one application that solves this discrete problem." This forces the customer to use another application to solve the next problem and so on. Today, most corporations face a Frankenstack monster of different applications from different vendors that don’t work well together. The applications have different maintenance cycles, are very complex to manage and everything is behind the firewall.

It has reached the point where corporate counsel is faced with a variety of point solutions that are seeking to address different aspects of e-discovery. What companies that we work with – like the three giants in our session at Legal Tech and hundreds more like them -  have come to realize is that Frankenstack is not a scalable approach and doesn’t work. They are looking for something that we call seamless e-discovery. It means using one base platform. Although I may use different tool sets that work nicely with this base platform, I need an end-to-end platform that will work for me from identification all the way through to production. And it's not just that: I need something that works whether it is behind my firewall, in the cloud, at my law firm or some mixture of these.

Seamless e-discovery is the answer, and Recommind is the only company that offers it. It is something that our corporate and law firm clients are using because they all realize that it is the future of e-discovery. It also allows e-discovery to be managed as a business process, just like any other business process.

Recommind offers a process based on predictive coding that goes far beyond e-discovery. It makes possible the online management by software of many management processes whether you are talking about an email system, a billing system or employee reviews. Our clients understand the wide-ranging applications of predictive coding.

Editors: What about the law firm perspective?

Carpenter: A recent Hildebrandt Institute study found that law firms are still under pressure to increase revenue and profitability at a time when corporate counsel are bringing more processes in-house. Another of our LegalTech panel sessions included Fulbright & Jaworski, Wilmer Hale, Morgan Lewis and Jackson Lewis. They agreed that clients wanted them to get to the heart of a matter more quickly.  

The law firms said that clients no longer want to wait until a human review is one-third or half done before the firm figures out what a case is about. So, these firms embrace predictive coding not only for many of the reasons companies do, but also because they can provide better service to their clients by becoming more closely integrated with them by using the same technological backbone as their clients. They can also use it behind the law firm firewall, in the cloud and even on the client’s premises.

Editor: Predictive coding works best if the persons feeding the software are highly skilled. Is that something that law firms can offer?

Carpenter: Absolutely. A law firm is a collection of talented people who offer high-value services and advice to clients.  They can apply very smart legal direction and advice and they just use technology as the work engine. It is similar to a Formula One race car, which is only as good as the driver. You take a good driver and a great car and that driver will make that car sing. And you don’t want great drivers to be distracted by changing the oil or tires; you want them to focus on using their driving skills to win the race. We enable great lawyers to focus on the business of providing high-value legal advice to their clients.

Editor: Besides the e-discovery context, what additional applications does predictive coding have?

Carpenter: The essence of predictive coding is that it applies very powerful machine-learning technology to expensive problems in the legal space that are generated by too much information and an inability to find the right information. This scenario is replicated many times over in corporations worldwide. They are constantly challenged by too much information and an inability to find that information in an effective and timely manner. The only way to address this information explosion challenge is by utilizing sophisticated machine-learning technology that can find exactly what someone is looking for in a way that is keyword agnostic.

The future is with what we call automated predictive information management. In effect, our technology anticipates what folks are going to be looking for and indexes content across an organization in a way that allows them to find exactly what they are looking for very quickly.

Editor: How does Big Data fit into that picture?

Carpenter: Essentially, Big Data is the sheer volume, complexity and variety of information that is being created today – which cannot be dealt with using traditional systems. There are thousands of organizations, vendors, users and scientists focusing on extracting value from this massive volume and variety of information that is being created in a multitude of different places.

The first challenge that Big Data presents is how to find what we are looking for. The challenge is to set up the infrastructure that will allow us to extract value from information flowing in from thousands of different places - in the cloud, behind the firewall, in social media, and on mobile devices, etc.

Looking to the future of information management, Recommind is focused on ways to discover what our information is trying to tell us and what trends are out there.  To be able to extract value from massive amounts of diverse information in a way that shows trends and allows us to see what is happening in a sea of information requires very accurate and sophisticated machine-learning technology.

Editor: Does that mean that it can serve as a compliance tool as well?

Carpenter: Predictive coding technology is smart enough to pick up developments that could lead to compliance failures. This is just one application of its ability to notice trends, think ahead and predict what is likely to happen. How people choose to use it is up to them. There are thousand of different use cases, and what we need to do, and what we are doing, is putting together and deploying systems in conjunction with our customers that address the most relevant and the most pressing of their challenges.

Today the biggest challenge in the legal industry, at least on the litigation side, is how do we automate as much of the early case assessment and review process as possible so that lawyers, like the racecar drivers I mentioned, can focus on the thing they do best – lawyering.

Predictive coding facilitates the conversation that needs to be had quickly, as opposed to waiting with partial information to have that conversation. Think of the advantage of knowing that you have a potential issue on your hands that permits you to nip a problem in the bud.

Editor: A big problem right now is over-preservation.

Carpenter: The efforts made at the Duke Conference and all the other efforts that are being made to deal with over-preservation from a rules perspective are admirable. Lawyers for Civil Justice deserves much credit for its efforts. Tom Hill of General Electric and others submitted excellent testimony to the House Judiciary Committee in support of rules changes. The size of the problem can also be effectively addressed by using predictive coding technology to winnow down the number of litigation holds by eliminating those where the possibility of litigation has passed. 

Editor: Big Data was recently covered in the New York Times. Is this the next hot topic for 2012?

Carpenter: It is somewhat inaccurate to say that it is the next hot topic for 2012. Big Data is a continuing problem that has been on people's radar screens for some time. It isn’t something that is going to be of interest for a year or two and then go away. Predictive coding provides a tool that can automate the life cycle of this information.

Please email the interviewee at with questions about this interview.