One of Pecan's early adopters has developed a payment processing platform that serves millions of businesses around the globe. The company's goal is to ensure that customers have access to the most advanced, reliable, and transparent payment services on the market.
The payment platform is constantly battling fraud attempts and striving to maintain the highest system reliability for its customers. The challenge is daunting in light of the 1 million transactions processed each day, a significant number of which are illegal fraud attempts. The company was looking for a solution that will minimize financial losses and prevent negative effects on its brand reputation.
Pecan was established in 2016 by PhDs in the field of computational neuroscience, AI and mathematics. Pecan offers an innovative Deep Learning platform as a service (PaaS). This includesstate-of-the-art, automated and general-purpose AI technologies that achieve unprecedented accuracy in a variety of use cases.
Existing solutions in the market know how to deal with only one of the many problems that enterprises must address, but each solution costs a great deal of money - tens to hundreds of thousands of dollars - and it takes many months to build and implement it.Moreover, the customer is required to pay the amount in advance and has no idea what solution they’ll receive and how effective it’ll be.
The Pecan platform can cope simultaneously with various and diversified challenges by relying on deep neural networks and artificial intelligence. The system connects to the raw data of the customer,and, based on the dynamics of the networks, locates and understands the analytical context. This enables predictions and deep insights based on machine learning derived from the data. Using these predictions and insights, the customer can take business, legal and marketing measures that solve their problems and maximize potential and opportunities.
Because the Pecan platform is based on deep-learningand artificial intelligence, the process of generating business insights from the data is done automatically without human contact and therefore without"contamination" of the data. The process is shorter than the alternatives in the market, up to a few days to build the model and apply it to the data. Moreover, the combination of the use of raw data and the most advanced artificial intelligence technology has given rise to extremely accurate models that are valid and reliable even for exceptional events that cause other systems to crash.
Pecan created a working model for the payment processing platform within a week of access to the raw data (compared to 5-7months required for alternative solutions). The cost of the project was estimated at $20,000 to $40,000 per (compared to $200,000 to $400,000 per year in alternative solutions). Out of a total 6,259 frauds identified by the system, 4,079 cases and actions taken by the company prevented 3,511 attempts. The additional revenue generated by the system is estimated at $1.45 million. The solution added 156% to the 56%reliability rate of the company's system.
The accuracy of Pecan's prediction reached 81% true positives and 4% false positives
CEO of the payment processing platform:
"Reliability and safety are the no. 1 concern of financial transactions platforms; no compromises are accepted in such a sensitive market. For the first time, we believe that we can fully manage and overcome the vast majority of criminal attempts on our systems. Pecan’s innovative and exceptional technologies provide powerful tools to predict attempts and take actions to mitigate them. Within a week of Pecan beginning the project, we have experienced a complete turnaround in our capabilities to confront financial crimes and thus have a better foundation for our futuregrowth".