We recently sat down with Amyn Dhala, Vice President Global Product Management AI Express, Mastercard. Amyn discussed the key differentiators that make AI Express so unique and valuable, as well as some of its most impactful use cases. Amyn also outlined the widespread AI adoption obstacles he’s witnessed across enterprises.
Unlike the majority of AI solutions currently on the market, our AI Express product enables organizations to quickly build their AI use cases. Within just six to eight weeks, a custom AI model is developed based on the customer organization’s pressing or emergent business needs. Leveraging state-of-the-art Mastercard AI technology, AI Express solves critical operational objectives, offers optimal performance and delivers an actionable deployment blueprint. AI Express empowers organizations to have a hands-on experience with their custom AI implementation and relevant datasets.
See Amyn Dhala speaking about AI Express in our AI Innovators video series.
Walk us through the cycle of implementing AI Express. What can customers expect?
First, we identify the customer’s use case for AI Express. Then, we execute. I can’t overstate the quality and amount of collaboration we have with customers during an AI Express implementation. We really leverage their business expertise and previous data analytics insight, and combine that with MasterCard’s AI technology so we can provide the best possible results. We’re constantly working with our customers’ data scientists and business leads to identify the potential business challenges, use cases and/or candidates for AI Express.
For example, we spend a lot of time analyzing the customer’s available data for the problem at hand. Once the customer has extracted the data, anonymized it and transferred it to Mastercard, our data scientists begin aggregating that data, enriching it and running it through our AI modules. From there, we come up with modeling iterations that we share with the customer to get their feedback.
This iterative form of model development is really important. It’s highly agile, yet also customized. In fact, we always provide a detailed deployment overview, so the customer knows exactly what it takes to deploy their model into production. This all takes place within six to eight weeks!
What types of organizations can use AI Express?
Any organization that’s interested in investing in AI is a candidate for AI Express. It helps to already have a healthy amount of data to extract intelligence from as this can lead to the creation of more personalized experiences for customers and improved profitability. For instance, a lot of financial institutions like issuers, acquirers or global merchants find value in AI Express.
What sets AI Express apart from the huge pool of other AI tools?
The models we develop offer exceptional performance. They’re being used in mission critical industries where we’re scoring tens, if not hundreds, of billions of events every year. And we’re doing that in near real-time and deploying these models at scale, too.
Also, our technology can ingest data from any source or format. Our customers find this really valuable, and even more helpful is that we can leverage any insights organizations have derived from prior exercises. We can then combine that intelligence with our AI Express technology to come up with a model output that’s truly superior and able to immediately meet the organization’s most pressing needs.
Lastly, AI Express doesn’t require organizations to purchase any hardware or software. All they need is to provide historical data relevant to their specific use case as well as data analytics resources to help provide feedback on the models we develop for them.
What are some of the most popular use cases you’re seeing for AI?
We’re seeing three to four major use cases across our customers. The first is for marketing optimization or customer attrition. Organizations are always trying to predict when their customers are going to churn, so by taking preemptive corrective action with AI they can actually execute on this critical business need.
Product recommendations is another popular use case, or even experience recommendations (i.e. using AI to provide the most personalized experience that’s most relevant to a specific customer at a specific point in time).
Credit delinquency (i.e. predicting which customers are likely to become a credit risk) is another major use case for our AI. Small improvements to our model can literally result in tens of millions of dollars in terms of benefit, if not more.
AI Express can also drastically reduce the number of false positive alerts, which improves the customer experience and in turn helps decrease customer attrition. As a result, transaction fraud or acquired fraud is another popular use case for our AI technology, in addition to anti-money laundering.
What’s kept more organizations from adopting AI?
I think there are two key reasons why organizations haven’t adopted AI. 1) They’re often not sure where to start, and 2) they don’t know how to build a business case for AI.
The good news is, we can help with both of those challenges. In terms of not knowing where to begin, we work with our customers to determine a range of use cases they can potentially focus on. We help them clarify the business value and outline deployment data considerations so they can actually prioritize their tasks and develop a clear roadmap. In terms of not knowing how to build a business case for AI, AI Express is designed for exactly that. AI Express models provide the increase in revenue and/or decrease in losses as applicable for particular use cases, allowing organizations to objectively make the case for adopting AI.
When it comes to AI, what’s the key ingredient for success?
Collaboration is paramount. With AI Express, we work really hard to leverage our customers’ domain expertise, existing data analytics and previous models. The crux is in bringing that information together with our AI technology to create a superior model that directly relates to the core business [challenge] an organization is trying to solve.
Even after AI Express reaches the production stage, we’re constantly asking, “What changes can be made?” “How can we work together to make an even bigger impact from a business perspective?” When it comes time to deploy, we can actually capture those highly specific needs and goals to ensure the models being deployed are as impactful to organizations as possible.