Daegis: designing a more cost-efficient process for litigation
November 16, 2012
Daegis is an e-discovery services provider that offers a comprehensive suite of solutions for the discovery phase of litigation, including managed document review — by far the most costly component of the multi-billion dollar e-discovery industry. During document review, Daegis manages a team of attorneys who read through millions of documents (mostly e-mail) to identify the handful of evidence relevant to the litigation — a perfect example of a “needle in a haystack” type of problem. With the explosion of big data, this “haystack” is also growing, significantly increasing the cost and time of performing the document review process. Datascope Analytics (DsA) worked with Daegis to understand how data can be used as a resource to lower costs and decrease the time required for review.
To begin, DsA collaborated with Daegis to design a prototype system that would automate much of their managed review process by predicting the relevancy of documents to any particular case. We like to think of this "technology-assisted review" as the Robocop of document review that takes advantage of the unique strengths of humans and computers; humans excel at making difficult decisions about relevance on ambiguous documents, and computers excel at quickly culling through documents that are obviously not relevant. The prototype demonstrated that Daegis could reduce the number of manually-reviewed documents by 70-90% while maintaining or improving on their best-of-market accuracy standards.
The prototype revealed design challenges that were hard to foresee. First, to avoid the costs associated with switching technologies (e.g., training seminars, becoming efficient with new technology), the technology needed to be designed in a way that did not force Daegis to change their highly-developed review workflow. Second, in order for Daegis’ customers to feel comfortable adopting the new technology, the system needed to be designed in a way that would simultaneously make predictions and measure the accuracy of the predictions. Finally, the interface to the technology needed to be designed in a way that would allow attorneys to defend the technology-assisted process in court. User feedback from the prototypes made it clear that without that final component, the technology would not be adopted.
After seeing the results of the prototype, Daegis contracted DsA to integrate the new technology into their existing review platform. It was important that the integrated "technology assisted review" product be understandable and usable by both the experts at Daegis on one extreme and their clients on the other extreme. With both extremes in mind, DsA designed the dashboard (a screenshot of which can be seen here) to be educational and to emphasize the transparency of the process. The dashboard teaches Daegis' clients about the technology behind the system, and clearly illustrates the current status of the review. In fact, the design of the system takes transparency, a common goal in dashboard design, to an extreme: the information must be the material that a legal team uses to defend their process in the judge’s quarters.
As it turns out, there are several competitors with technology-assisted review products in the e-discovery market. While it will take time to learn which companies’ products have the most sophisticated analytics (we believe that ours does!), it is already clear that the transparency and educational component of our design makes the Daegis solution the top choice for those customers that have concern about adopting a radical new technology. The resulting product, Acumen, has recently been publicly announced and is currently in use by Daegis’ clients. For further details about the project, please contact us.
Contributors to “Daegis: designing a more cost-efficient process for litigation”
Wondering why there are multiple contributors? At DsA, we work in teams. Even on blog posts, we often work together or ask for others to take a look at the post before we post it. When we do that, the pictures of those that wrote the post are larger than those that edited the post.