Our communication analytics visualizations were used to illustrate a six-month pattern of IP theft by one insider. The employee had managed to create 19 separate email accounts to capture one-off actions where he sent stolen company secrets & IP. Our visualizations and analytics showed the insider as both the sender and the recipient, and also allowed the company to track every document so that they could understand the extent of the theft. This took 20 minutes to achieve once the data was in the system. The company was able to avoid significant damages and trade secret infringement by quickly identifying the thief and acting on the information discovered.
A Financial Services Company needed to archive email for FINRA regulatory compliance. We set up their system to send all email messages directly to the Orca AI Archive, and upon ingestion our machine learning scanned and classified all the content. Orca’s language modelling was tuned for FINRA compliance with sets of known hot words and phrases. This was supplemented by additional inhouse tuning which greatly improved overall compliance accuracy, while reducing their daily workload. The clients tuning of the machine learning system reduced their rates of false positives by over 20% within the first 60 days of the system being installed.
An importer of wood products was being investigated by the government. The merchant provided us with copies of the same data the USDA had just seized and were confident there were no issues. Our communications analytics quickly detected a pattern of fraud being committed by an external Chinese supplier and one internal sales person, leading to Lacey Act violations. The case settled in short order after our system had discovered who was involved. These violations were occurring right under their noses and were detected in one hour. Our proactive technology saved them the PR nightmare as well as what could have been a huge fine.
1st Court Approved ML
We provided the first machine learning (ML) automated categorization technology in the market to be approved in a US Court (April 2012, Global Aerospace Case). Using a conventional legal email review process, the volume of email would have required months of time from a warehouse full of attorneys. Instead, using our technology, one subject matter expert trained our machine learning system over a 17-hour period and was able to eliminate over 82% of the unimportant documents, reducing the data to a small collection of highly- relevant documents. This enabled the legal team to focus on what was relevant. Our technology helped set the legal precedent for the use of machine learning in the legal industry.
Orca AI provided the core analytics technology to enable multiple plaintiffs to win a $2.2BN settlement in a faulty medical devices matter against a Fortune 500 company. Orca AI's technology enabled the uncovering of key fact and communication patterns, highlighted significant data gaps, and identified incorrectly redacted information core to the case. This multi-district litigation was a collaborative effort between 70 legal professionals across 35 plaintiff law firms using our technology in a highly coordinated and integrated legal battle. We helped non-medical lawyers to understand the medical and non-medical terms through our conceptual link analysis and clustering technologies.
Human Resources Audit
While performing an HR Compliance Audit at a major energy company, our machine-learning language filters detected problems. Looking at the latest three months of email data from 12,000 employees, our analytics system was tuned to look for HR issues and rapidly identified multiple HR violations including evidence of violence and weapons in the workplace. It also detected a pattern of predatory sexual behavior between a supervisor and of his five direct reports. In total, 340 emails were found on the first day and sent to in-house counsel for review and action. This led to the “data driven dismissal” of this sexual predator. It also moved the company from a reactive to proactive mind set for HR compliance.