American Express has long been one of the world’s most well-respected and innovative companies. From being one of the first companies to purchase its own computer back in 1961 to being an early adopter of online banking, American Express has a proven track record of embracing cutting-edge technology. In addition, as Amex offered superior fraud protection and a slew of luxurious membership rewards, it became the preferred credit card provider to the wealthy, jet set elite. 

However, in between the rise of fintech and the growing digital sophistication of its competitors, American Express has struggled to maintain its brand status and profit margins. In fact, as the COVID-19 crisis continuous to impact the travel and leisure sector, American Express's revenues have fallen short of expectations. 

To help turn the tide, Amex is looking towards artificial intelligence (AI) to optimize what it does best: guarantee secure transactions and deliver exceptional customer experiences. 

ML-Powered Fraud Detection 

Cyber crime is not only on the rise, it's more financially ruinous than ever. In 2020, the average business cost of a cyberattack was $3.86 million. By the end of 2021, Cyberattacks are projected to cost the global economy upward of $6 trillion, 2X as much as in 2015. 

One major target of cyber criminals are credit card companies such as AMEX. Not only do fraudulent charges cost U.S. consumers roughly $11 billion per year, they’re also surging in frequency year after year.  

To better identify and mitigate incidence of credit card fraud, American Express is now deploying deep-learning-based models optimized with NVIDIA TensorRT and running on NVIDIA Triton Inference Server.  Using recurrent neural networks (RNNs) enhanced with long short-term memory networks (LSTMs) these tools are able to identify and flag anomalies in tens of millions of daily transactions in real time.

Leveraging this enhanced, real-time fraud detection system, American Express was able to not only adhere to its two-millisecond latency requirement, but deliver a 50x improvement over CPU-based configurations. In addition, the GPU-accelerated LSTM deep neural network combined with its long-standing gradient boosting machine (GBM) model (used for regression and classification) improved fraud detection accuracy by up to 6% in specific segments.

Years in the making, Amex’s newest ML model for fraud detection, Gen X,  was developed on billions of observations and executes a sequence of more than 1,000 decision trees. As it automates over 8 billion decisions, ingests data from over $1T in transactions, and generates decisions in mere milliseconds, it might even be the largest commercial ML model in the world. One thing that is for certain is that, thanks to GEN X and its predecessors, Amex has maintained the lowest fraud rates in the credit card industry (half that of its competitors).

Accelerating Commercial Credit

Large financial services organizations and credit card companies are almost always at the bleeding-edge of digital technology adoption. As the need for speed and precision in this industry is paramount, one might assume that a tech-savvy company like American Express had long ago automated 100% of its processes. 

However, you would be wrong. Until recently, the commercial card underwriting process (whereby credit for small business approved or denied) at American Express was still manual throughout much of Europe. Collecting the initial background financial information alone could take up to 30 days and then a human reviewer would analyze each document one-at-time. As a result, the end-to-end process could take weeks if not months. 

As the COVID-19 crisis took hold, American Express realized that these manual processes simply wouldn’t work. Not only did lock-down orders make it more difficult to collect and manage the relevant financial documents, the survival of many small businesses depended on securing additional lines of credit ASAP. 

To accelerate the commercial credit onboarding process, American Express leveraged AI and machine learning (ML) to automate that document analysis process, enabling the underwriter to focus on working with the customer. 

As VP and Chief Credit Officer, Global Commercial Payments, EMEA, Jill Zucker Sheckman explained to diginomica, “Machine learning absolutely has helped us extend credit, especially right now when businesses are still trying to recover and need fast decisions to do that. So then instead of waiting 30 days to find out how much credit you have as a business owner, you now get a more or less instant, and accurate, decision.

Overall, use of machine learning at Amex has also led to a 20% to 30% improvement in our risk models, plus an overall reduction in cycle time - so better and quicker decisions. We think that is extremely helpful for both the SMEs and the large businesses we serve as we all go forward to recovery.”

She also added, “once we had people sitting in our credit operations team taking in data or documents, but now the customer can self-service in some cases, or a salesperson might be able to assist them and tell them they can call our credit operations team, ask for a line increase, and can get one potentially in a matter of seconds.”

Create Exceptional Customer Experiences

Later this year, Amex will be releasing it’s new app-based, contextual and predictive search capability. Trained on an natural language processing (NLP) model initially designed for the company’s customer service chatbots, the feature will “understand” various scenarios and, if all goes right, predict what customers need before they type anything at all.

For example, if a person opens the app at the airport, the tool will surmise that the person is trying to find the lounge. Or, if a person opens the app after seeing suspicious charges, the tool will ask if they’d like to contest the charge right away.

As for the impetus for this project, Josh Pizzaro, the company’s director of AI, recently explained to VentureBeat, “we started building the model because if you think about where the world was, it was in a place where we would ask our card members how they’re feeling and what they wanted. And now today, in the machine learning era, we just need to know, and we do know based on the data that we have. And so we look across the different services that we provide and try to reduce the burden on the customer, and in this case, search and present things in that contextual and fast way so they get what they want faster. Because ultimately, great customer experience is about speed.”