Major features

Built for 24/7 operation and 99.999% availability

Safer Payments is usually implemented as a cluster of three instances. Each instance is independently able to run the full production volume and together they automatically replicate all data. This "triple redundancy" ensures that operations continue even if an instance, a server, or an entire data center fails. It also allows for software and hardware maintenance without any disruption of production.

Horizontal and vertical scalability

Using a clustered architecture and parallel programming techniques, multi-core commodity servers are utilized most efficiently. Even low end Intel Xeon-cores deliver 1000+ transactions per second. Our customers operate up to 12,000 transactions per second sustained peak volume with Safer Payments (financial and non-financial transactions).

Full configurability

All aspects of Safer Payments are fully configurable from its web user interface. This includes payment instruments, data streams, the model data dictionary, and the actual detection models. This enables ultra-short implementation times (3 to 6 months typical), since Safer Payments is configured according to the existing environment and not the other way around.

Fully accessibility API and microservices

Used as microservice, all of Safer Payments functions are accessible from the outside via https enabling any other system or programming environment to access and customize any of the Safer Payments functions. This allows your data scientists to bring their preferred machine learning modeling and runtime frameworks, including open source, to Safer Payments.

Product Flyer

Read the brochure

Customer case studies

Case study image

Protecting an entire country from payment fraud

STET, Paris
Read the case study
Case study image

Multi-channel fraud prevention at a payment switch

Indue, Brisbane
Read the case study
Case study image

Protecting e-commerce from fraud

Borgun, Reykjavik
Read the case study

How customers use it

  • Real-time payments fraud prevention

    Real-time payments fraud prevention

    Problem

    Real-time payments are low margin, high risk transactions with few describing data elements. Criminals structure payments and 'smurf' them through mule accounts to disguise the flow of money.

    Solution

    We have developed "peer profiling" that profiles each originator and beneficiary of a payment as both sender and recipient of payments in the past. This enables Safer Payments to reconstruct the true flow of monies in real-time.

  • Model factory

    Model factory

    Problem

    No one machine learning or artificial intelligence technique can detect all types of fraud. It has been shown in data science, and it is the experience of our customers and our specialists, that only a combination of techniques is efficient.

    Solution

    To enable the combination of the nest data science techniques, we have created the "model factory" concept. It allows Safer Payments users to pick the combination of techniques best suited to solve their fraud problem.

  • Online and Mobile Banking

    Online and Mobile Banking

    Problem

    Most account-to-account transactions will migrate to immediate payments in the next decade. Since the digital channel is the major initiation point for these transactions, it must be hardened to maximum protection from fraudulent manipulation.

    Solution

    Safer Payments utilizes data from the actual online/mobile banking session. A built-in device identification capability, with device behavior profiling and a device reputation database as well as biometric parameter recognition hardens this channel.

  • Utilize open source data science

    Utilize open source data science

    Problem

    When participating in the game-changing innovation that derives from the open source data science movement, you want to be entirely free in your choice of technology and modelling tools.

    Solution

    Safer Payments is the 'most open' data science platform for payment fraud prevention currently available in the market. You can export and import models in portable PMML format and ingest models or feature extractions as Python code.

  • Multi-tenancy and multi-channel configuration

    Multi-tenancy and multi-channel configuration

    Problem

    Many or our customers are processing multiple portfolios of the same payment instrument and/or different channels.

    Solution

    Safer Payments provides hierarchical multi-tenancy. This allows to efficiently manage hundreds of different tenants, each with their own data and customization. Inheritance allows to also maintain "central" models and configurations.

Technical details

Software requirements

Requires on any bare Linux operating system. Red Hat Enterprise Linux recommended. Installs on OS or as Docker container. Real-time database, application server, and replication layer all fully embedded in product and must not be provided.

    Hardware requirements

    Runs on commodity servers (x86). Physical or virtual. Recommend three servers for triple redundancy. Sizing depends on data volumes and retention periods.

      Technical specifications

      To deliver performance of this magnitude, Safer Payments utilizes massive parallel computing—critical computations scale linearly with the number of CPU cores available. IBM Safer Payments is written in C/C++, the programming language of choice for any application requiring massive performance.

        See a complete list of technical specifications