Plumbing Meets Human Reasoning

To make faster payments a reality, decades-old core systems have to be streamlined. Here’s an approach that relies on artificial intelligence to avoid the system overhauls that have been tried in the past.

The Federal Reserve’s faster-payments initiative has created a catalyst to improve core systems to respond to the demands of real-time authorization and settlement of transactions, and to provide the recipient with irrevocable access to funds upon receipt.

Mundane as they might be, core systems function as the plumbing of the payments infrastructure. These systems support the daily banking functions related to deposit and loan transaction processing, including eligibility determination and access to general-ledger and back-office servicing. To say that the systems are legacy is an understatement. Most baby boomers were at the early stages of their careers when these systems were first implemented.

The risks associated with making any changes to these inflexible, inefficient, and outdated core systems often outweigh the benefits. The transaction-processing and database functions are inexorably tied together, making it extremely difficult integrate these legacy systems with new applications and products not only to create product interdependencies, but also to support complex risk-management frameworks and the associated information reporting. Furthermore, these systems do not support multidimensional views of the customer or cross-channel access.

Next-Generation Core Systems

The movement in the late 1990s and early 2000s to completely replace core systems with enterprise payment switches failed. A complete system overhaul proved to be too much to tackle. The next-generation core systems were destined to be updated incrementally, particularly focusing on improved access to data sets across multiple systems and channels.

Yet, faster payment requires system enhancements to support three important functions: (1) faster access to data and the ability to retrieve information from multiple sources; (2) quicker decisioning processes, facilitated through enhanced analysis capabilities; and (3) a higher degree of accuracy, increasing the level of confidence in the quality of the information, thereby reducing risk.

How do we move to the next-generation core system to support this functionality without a complete replacement of legacy systems? The answer is to design core systems to separate the presentation layer (i.e., channels) and customer data management from the transaction processing, settlement, and accounting functions.

These modifications will be facilitated using open application programming interfaces (APIs) and incorporating a middle layer. New technologies can then be applied in targeted application areas to further remediate the legacy-system shortcomings.

AI As a Strategy

Here’s where artificial intelligence (AI) is key. AI uses human reasoning as a guide to provide better services or create better products. While APIs provide access to internal and external data sets that have been historically siloed, AI marries these data sets for enhanced onboarding and risk profiling.

AI can not only enhance the authorization function but also provide a deeper understanding and multidimensional view of the customer and his/her behavior. AI will be part of any core system enhancements, providing a better understanding of what the customer is doing.

The use of AI is emerging in three areas relating to security and compliance, namely: identity management, access management, and know your customer (KYC). Here’s a look at all three:

Identity Management

Identity management is the process of verifying and authenticating the identities of individuals and corporations. It requires access to robust data sets and incorporates the use of both static and dynamic biometric data as the basis for user authentication. As part of the identity-management process, AI analyzes user behavioral patterns and activities and detects anomalies automatically.

Identities are registered as part of the onboarding process and maintained in identity directories. Identity management is not limited to consumers but includes businesses as well. A business-to-business directory from NACHA, the governing body for the nation’s automated clearing house network, provides an example of how this might be structured using blockchain technology.

Access Management

Access management goes hand-in-hand with identity management. It enforces policies and access rights to all of the data sets. AI will be embedded into future products’ enabling systems to learn about the user for access management. User names and passwords with algorithms will be replaced by smart systems that verify identity. Access management will move from being user-controlled to machine-controlled, thereby improving overall security and enabling regulatory compliance.

Know Your Customer (KYC)

KYC is all about truly knowing the identity of a customer or company and parsing those data sets into meaningful buckets so that due diligence can be performed. KYC entails the comprehensive collection of personal information about a customer or business as part of the onboarding process and ongoing management of customer relationships. It must be a dynamic process, tracking the activity of customers throughout the life-cycle of the customer relationship.

AI enhances the KYC data-collection process, particularly in helping to identify high-risk customers. AI takes the information available from disparate internal and external sources, puts all this information together, and makes sense of the data. AI automation parses multiple data sets, creating associations with many different types of data instead of just a single piece of information.

These data sets include account data, payment components, customer history, credit-score profiles, static and dynamic behavioral patterns, and even societal data from social-media sources. Although these data sets are constantly evolving, AI can help with both their movement and activity. AI can also be used to evaluate the accuracy of the data sets.

AI automation partitions raw data flows by tagging, analyzing, and delivering enhanced data. AI complies all this information and builds a customer risk profile, including a credit risk score. AI will be instrumental in supporting real-time data verification of KYC—an essential requirement for faster payments.

The development of accurate risk scores supports faster and more accurate real-time decisioning. For example, AI enables real-time anti-money-laundering background checks and performs sophisticated verification services simultaneously.

AI is used to create clear definitions of policies and procedures beginning at enrollment or product introduction. Human analysts can use this enriched data to examine relationships and draw conclusions about the customer or business risk profile. Ultimately, these people are in control and they will decide what needs to be hardened in the proof-of-concept to move to real-time decisioning at the point of transaction.

What’s Your AI Strategy?

Industry specialists can help organizations develop a strategy to identity the low-hanging fruit and to update, modernize, and enhance customer information sets so they work more efficiently with the faster payments real-time authorization requirements. The reference architecture (see diagram) suggests a point of departure for getting started.

We need to change our current thinking. Most financial institutions think they are in the business of moving money, but their business is really about monetizing information the same way Google, Facebook, Amazon, and Netflix do.

These behemoth information companies treat client information as an asset, creating new products and services based on customer behavior and patterns, and innovating based on an understanding of how, when, where, why, and what the client is doing.

A comprehensive approach to AI used in combination with native APIs and business rules and logic will securitize next-generation core and digital platforms, resulting in safer, more efficient, and predictable payments and thereby supporting the faster-payments initiative.

—Bo Berg is a digital-transformation, blockchain, and AI expert. Reach him at bo@edweb.com. Maria Arminio, co-author, is president and chief executive at Avenue B Consulting, Redondo Beach, Calif. Reach her at maria.arminio@­avenuebconsulting.com.