The financial services industry has always been among the first sectors to embrace leading-edge information technology, so it should not come as a surprise that artificial intelligence (AI) is no exception. After all, trading firms have been taking advantage of various algorithms to programmatically trade stocks in matter of microseconds for years now.
But the rise of AI does present the financial services sector with some unique challenges. Financial services firms have made massive investments in data centers to run some of the most latency-sensitive applications in the world. And yet, for AI to be accurate, the models that get constructed need access to massive amounts of data that is costly to store locally. Many financial services firms are now trying to figure out how to have the best of both worlds by continuing to process high-performance applications locally, while taking advantage of public clouds to store massive amounts of data. What makes that especially challenging is synchronizing data running in a local production environment with copies of that data running in the cloud in a way that ensures all the regulations concerning protecting sensitive customer data are met.
AI models accessing data in the cloud need to be able to surface recommendations in a way that is timely. After all, investment advice that shows up after a transaction is completed isn’t going to be all that useful. To achieve that goal, many financial services firms are at the forefront of connecting their local applications to various cloud services over private high-speed wide area networks (WANs). It may take a while for the AI applications to be broadly applied across the FinTech sector, but it’s already clear an AI arms race is underway.
For example, AI Powered Equity ETF has created an exchange-traded fund that uses AI to choose long-term stock holdings, which is one of the first instances where algorithms are being used to select what stocks to invest in. Most of those calculations run on the IBM Watson supercomputer that is accessed as a service provided by IBM. Now it’s only a matter of time before investment firms use AI models to try to outsmart one another versus relying solely on the knowledge of human traders.
In the meantime, the need to master AI and other emerging technologies served as a catalyst this month for the launch of the Fintech Open Source Foundation (FINOS). Formerly known as the Symphony Software Foundation, FINOS has 30 initial launch members, including BNY Mellon, Citi, Credit Suisse, Goldman Sachs, HSBC, JP Morgan, Morgan Stanley, UBS and Well Fargo. FINOS is already home for 64 projects, 82 open source repositories and counts over 300 contributors. Vendors that have joined FINOS include Red Hat and NodeSource.
Customer-facing applications will be written in Node.js in a way that enables them to access backend COBOL and Java applications as well as various AI models built using open source frameworks such as TensorFlow, Caffe and a host of other that will be used to build AI applications hosted mainly in the cloud, says DeMeo. Those AI models will be accessed via well-defined application programming interfaces (API) that will make it simpler for the average developer to embed them with their applications. That approach should enable financial services firms to build new applications faster versus simply reinventing existing ones, says DeMeo.
“It’s all a part of being agile,” says DeMeo.
The unprecedented level of cooperation occurring across financial services is expected to continue for the foreseeable future. Faced with a raft of startup competitors leveraging the latest IT technologies, the old financial guard is now moving aggressively to turn those rivals into partners. But a World FinTech Report 2018 from Capgemini and LinkedIn finds that 70 percent of FinTech executives said their top challenges to collaborating with traditional financial firms is a lack of agility. That issue is getting compounded by the rate of change occurring within IT, says Bill Sullivan, global head of financial services market intelligence at Capgemini Financial Services.
‘The pace of change is occurring faster than anticipated,” says Sullivan.
That rate of change explains why many financial services firms to increase their reliance on managed service providers (MSPs) even when they already have large internal IT staffs. The complexity associated with building, testing and managing modern applications in addition to managing all the data they need to access is exacerbating an already existing skills shortage, says Joni Williamson, director of FinServ Solutions for Rackspace.
Rackspace this week announced it has achieved Amazon Web Services (AWS) Financial Services Competency status, which AWS awards to IT service providers that have demonstrated mastery of deploying and managing financial applications on its public cloud. Most financial services firms have a lot of work to be done modernizing processes before they can take advantage of AI models, including moving large amounts of data in the public cloud, notes Williamson.
Couple that with the rise of DevOps, microservices, and soon blockchain databases and it’s apparent that even the most well-heeled financial services firms need additional IT help, says Williamson. By relying more on Rackspace to manage IT infrastructure, existing resources can be diverted to applications that enable financial services firms to compete more effectively, adds Williamson.
“We tell them they should focus on their own secret sauce,” says Williamson.
Much of that secret sauce is clearly going to be embedded within a multitude of AI models. Not everyone may fully appreciate the impact those AI models will have not only on the business, but also on how IT within a financial services firm is managed. But no matter how that transformation plays out, it’s almost certain that other industries will soon be copying whatever playbook for building and managing AI-infused applications that the financial services sector winds up writing.