Overview

Conducting internal investigations pose their own unique challenges compared to other types of legal reviews. Cost constraints and timelines are common challenges, but for an internal investigation there is uncertainty if documents that support, or deny, the subject matter being investigated even exist at all. For a global corporation, other factors like foreign language, local laws, and regulations can further complicate the investigation process. These all add additional costs and time to document analysis and review.

The Challenge

A multi-billion dollar global semiconductor corporation had to conduct an internal investigation.  The original document set that needed to be reviewed contained 60,000 documents, mostly in Japanese. They requested help from their Legal Service Provider, Level Legal. There was an aggressive two week time limit to review the documents and report any findings. Level Legal needed automated tools to figure out if documents existed to substantiate the investigation in order to escalate accordingly, or close out the investigation.

The Solution

Level Legal utilized CaseAssist, Casepoint’s AI and Advanced Analytics suite when conducting the internal investigation document review. With assistance from a Casepoint Client Success Manager, Level Legal set up their workflows for an Active Learning based review. This included steps for accuracy verification and guidelines for model stability based on F-measure scores.

In just the first full day running the Active Learning based review for both English and Japanese terms, the feature culled the initial review set by 50%. Level Legal was then able to perform a review on over 600 documents. This allowed level Legal to quickly provide their corporate client with a better estimate for the total cost of review and confirm that they could meet the timeline right away.  

Continuing to iterate the active learning searches, Level Legal was eventually able to cull the relevant data set to under 1,800 total documents, representing an overall 97% cull rate. Level Legal then presented their findings for the investigation to the corporation ahead of an already tight two week time frame while avoiding high costs typically involved with foreign language document review. 

The Impact

Using Casepoint’s CaseAssist Active Learning tools to conduct the internal investigation, the project was completed on time and $175,000 – $200,000 under initial eDiscovery cost estimates. There were several ways technology generated savings for this internal investigation client:

  • Casepoint’s CaseAssist Active Learning is built-in to the standard eDiscovery platform, ready for use, with no additional costs
  • The ability of the AI to conduct and run searches on multiple languages (in this case English and Japanese) eliminated the cost of translation and the costs of finding local council to weigh in on localization (if words used were typical for the industry, geography, social, and cultural context)
  • The 97% cull rate identifying key documents reduced the scope of the project from 60,000 documents to 1,800 relevant documents and 50 highly relevant documents. Level Legal needed fewer days to complete the review and their seasoned review team completed the review within a week

Casepoint’s Advanced Analytics features, also known as CaseAssist, saved our team a tremendous amount of review time, and saved our client close to $200,000 dollars in review costs. We plan to use CaseAssist Active Learning in as many cases as we can. It brings indisputable value to corporations that are already looking for ways to trim legal spend and leverage artificial intelligence-based technology.

Director, Client Services at Level Legal

Greg Moreman

Director, Client Services at Level Legal