Companies are legally obligated to perform "e-discovery" if they are part of litigation or a regulatory investigation. The e-discovery process includes finding potentially relevant data, having attorneys review the information, and then producing the relevant information to the opposing party or government. Data volumes have increased to the point that it is common to have matters with data sets in the range of multiple terabytes. Companies can't NOT review the data, so they need to balance cost without sacrificing quality and inadvertently producing privileged information to the other side. To help address this problem, software vendors have started to introduce machine learning – a combination of statistics and computer science algorithms - to their document review applications. Sometimes called computer assisted review, technology assisted review or predictive coding – these additions are intended to help legal teams accurately classify every document in the matter. But lawyers, in general, are a non-technical audience. Wading into the complexities of advanced math can quickly become more problematic than reviewing millions of documents. More comfortable with clear decision points than ambiguity these users need software which can help them see beyond the raw numbers and into what the numbers mean in terms of hours and dollars to support the review project. They also need a solution which they can explain and defend in front of a judge and jury. The FTI Technology team developed a predictive coding application with a user interface that attorneys can confidently use, understand and explain.

FTI Technology developed an interactive, machine learning visualization, which combines key data and decision points in ways that allow legal teams to quickly identify a correct path through the project. This innovative layering of the data allows for rapid insight and even more rapid decision making. Nobody has done this in this industry (e-discovery) – build an enterprise-grade data visualization on top of something that ultimately has to be defensible to the court, but behaves like a consumer product. By layering this data, we’ve taken days off the task, and by consumerizing the experience, we’ve taken time off the training as well.