How We've Been Applied

Inspired by the remarkable way our brain handles information, Saffron operates under the premise that impactful data solutions aren't one-size-fits-all. Your unique business deserves a unique solution. Our platform and products are designed to reveal what actually matters to you - constantly learning on the fly to help your business interact with your customers and the world in a way that's personal and meaningful.

Unique knowledge. Better decisions. Minimized risk.

Issue and Defect Resolution

Faster to Market, Saving Millions
One of the world's largest complex hardware manufacturers' business units, responsible for the post-product validation process, was looking to speed up its time to market and the time it took to identify and resolve process issues. Industry-wide, typically around 50% of the cost of development is attributable to validation and testing, making this a high value and high priority program. Saffron helped accelerate the verification and test process without changing the existing workflow or the existing Product Lifecycle Management or Application Lifecycle Management systems. By unifying data across multiple systems at the entity level, Saffron enabled the rapid discovery of similar issues and defects in testing centers around the globe, shaving 35% off time to market for new product releases. Additionally, Saffron helped shaved undisclosed millions off the cost of testing each new product, with as many as 30 products being launched annually.

High Tech Manufacturing Case Study
See the Infographic
two low poly objects1_0055
Manufacturing

Keeping Global Aeronautics Flying

Predictive Maintenance
Building and maintaining planes can be a complex task as commercial aircraft tend to be in service 25 to 30 years. It becomes even more challenging when aircraft go out of production but are utilized beyond their expected lifetime. Manufacturers dealing with out-of-production aircraft often experience challenges providing their customers with spare parts, which are typically not in stock due to a number of planes in service, and the time they have been out of production. Saffron handled this challenge using it's associative memory and reasoning layer. The manufacturer was able to identify, manage and anticipate spare part problems more accurately and efficiently than ever before - bringing the average request time down from 4 hours to 5 minutes and saving hundreds of millions of dollars.

Main
Read the case study

Keeping Energy Safe

Anticipating Adverse Events
Operating and maintaining a safe nuclear power plant starts with documenting literally everything. The hazard from a spilled cup of coffee, a pump or valve starting to leak, a gauge not registering properly, catastrophic equipment failure or leaking radiation are all part of the mix. The respective operators within all areas of the plant use the Corrective Action Program report (CAP) to render the information. Given the breadth and depth of the events this report covers, tens of thousands of CAP reports are generated every year - by each nuclear power plant. Our client, a service company managing US plant operations, needed a way to automate the classification process to increase its efficiency and to extract information from years of reported events. Out of the box, Saffron was 92% accurate, enabling immediate correction of previously misclassified records within the data.

Adverse Events
Read the case study
two low poly objects1_0055
two low poly objects1_0055

Preventing Fraud, Waste and Abuse

Detecting Fraud Rings
The insurance industry estimates that 10 to 15 cents of every dollar paid in premiums goes to paying fraudulent claims, totaling more than $1.3 billion each year. In fact, nearly $80 billion in fraudulent claims are made annually in the United States. The speed and sophistication at which new fraud techniques are emerging create major challenges for today’s fraud detection systems to keep up; it involves building new models, creating new dictionaries, and performing new training. Relevant information to detect and prevent these crimes is more spread out, or “siloed,” than ever; such data are found not only on the open web and in insurance databases, but also in social media. A leading insurer contacted Saffron with a challenge: reduce its $6.3 billion expenditures on auto insurance claims, which is its largest annual expense. In less than a month, Saffron examined 113,000 claims from 1 year in 1 state and found 3 potential fraud rings, saving hundreds of thousands of dollars.

Fraud Case Study
Read the case study

Redefining Medical Diagnosis

Personalized Medicine
The echocardiogram is one of the most prescribed diagnostic tests in cardiology. It provides an incredibly complex and multidimensional data set, impossible to process entirely at the brain level. When physicians use echocardiograms to diagnose a patient, they commonly use only seven attributes among the thousands of attributes available in the data to make a diagnosis - reducing the potential accuracy. See how Saffron was able to use all the data available to differentiate three presenting heart conditions at a 90% accuracy rate.




Dr. Partho Sengupta
Read the case study
two low poly objects1_0055

Defense Case Study

Slowing Cocaine Trafficking with Cognitive Computing
In 2007, JIATFS supported approximately 57% of all cocaine disruptions worldwide. However, as the traffickers applied increasingly sophisticated methods to mitigate the task force’s efforts, coupled with the decline of detection and monitoring assets provided by governmental agencies and departments within the US and abroad, this figure fell to 38% in 2013. Ingesting large volumes of data from interagency and international partners is extraordinarily difficult. In the four previous decades prior to the creation of the task force, Congress assigned other US departments and agencies the detection and monitoring mission and all failed due in large part to the inability to unify the data sets. Intelligence-driven operations require unified data sets to produce actionable targets.

Defense Case Study
Read the case study

Latest Tweets