Editor's note: Channel partners, from global systems integrators to regional managed service providers, are beginning to find more work in artificial intelligence. Much of that work is tied to vertical-market-specific use cases. In this feature, we examine AI applications in finance and the opportunities for partners.
Financial enterprises now work with more information than ever before. However, they often have trouble using that data for a competitive advantage. Enter AI and machine learning -- tools that automate data analysis. Here's an overview of some currently prominent applications for AI in financial services.
Financial services companies need to protect valuable assets, so fraud detection is one area where AI use is common. In fact, companies spend more than $1 billion on such products, according to Ronald Schmelzer, managing partner at Cognilytica, an AI market research firm in Washington, D.C.
Financial services organizations started by relying on the software to monitor large transactions, such as corporate stock trades, and to guard against problems, such as money laundering. But they have expanded fraud detection's reach into the consumer market, where they use it to monitor banking and credit card transactions.
In the customer service field, much of the early AI work focused on back-office procedures to provide clients easier access to their account information. But there's a shift underway.
"Recently, AI has begun having an impact on the front office," said Robert Huntsman, global head of data science at Synechron, a financial services consulting firm with headquarters in New York, 8,000 employees scattered across the globe and revenue of more than $500 million.
For instance, financial services companies rely on the technology to develop their marketing plans.
"Increasingly, financial services companies are focused on hyper-personalization," said Kathleen Walch, managing partner of Cognilytica. Such programs focus on each individual's needs, rather than the needs of various groups of individuals.
Data deluge fuels AI applications in finance
With the advent of smaller devices and the growing power of data center servers, financial firms find themselves tinkering with vast amounts of information.
"We have been working with one financial services company that is trying to integrate 60,000 database management systems," Synechron's Huntsman said.
When taking on such work, traditional human-computer interactions take too long and cost too much, so organizations are turning to AI and machine learning for help. With the former, a developer writes software that automates steps in the process of reaching a deduction based on data. With machine learning, the machine draws such conclusions without a developer explicitly instructing it how to get there.
AI can help financial services firms get new customers in the system, as well as fill the sales funnel. IT solutions providers are beginning to target this area.
Synechron's Wealth Tech Accelerator program, unveiled in March 2018, employs optical character recognition (OCR) and natural language processing (NLP) to ease client onboarding in the wealth management space. OCR and NLP are used to scan a new customer's ID and extract data to automatically populate a wealth management firm's customer onboarding forms.
Customer service chatbots
Chatbots, meanwhile, are another class of customer-service-oriented AI applications in finance.
"Many financial services organizations are deploying chatbots, which have varying levels of sophistication," noted Matt Bienfang, managing director and global industry lead at Insight Enterprises, which has provided consulting services to enterprises since 1988 and has more than 6,600 employees situated in 20 countries across the globe.
A study published in February by Juniper Research said banks' use of chatbots will generate operational cost savings of $7.3 billion worldwide by 2023. That figure compares with a savings of $209 million in 2019, according to the market research firm.
AI in financial services: The competitive landscape
AI market growth has attracted a bevy of competitors. In addition to other channel companies, partners compete with the service organizations of platform suppliers, such as AWS, Google and IBM.
Staffing up to compete can prove a challenge, however.
"Finding personnel who know computer programming, AI data modeling and the financial services industry is quite difficult," Insight Enterprises' Bienfang explained. "You might find someone with one or two needed skills, but seldom find anyone with all of them."
Competition for such talent is intense. Financial service data scientists command high salaries, starting in the six figures and sometimes zooming past $250,000. Partners need to provide these employees with not only a lot of money, but also a stimulating and flexible workplace.
Data collection and routing
Matt Bienfangmanaging director and global industry lead, Insight Enterprises
Finance is a paper- and form-intensive business. Financial services firms spend a lot of time collecting information for auditing and compliance reports. Small armies of employees spend their days collecting, examining, consolidating and routing customer transaction information. Robotic process automation (RPA) automates that workflow.
RPA products may not have AI at their core, but some offerings integrate with intelligent OCR and machine learning technology. Global revenue for RPA tools will reach $4.1 billion in 2026, according to Acumen Research and Consulting.
Financial services employees continuously use information to make financial decisions: Should they approve a new business loan? Should they sell a stock? AI and machine learning algorithms analyze data and make recommendations -- and increasingly decisions -- about such transactions.
AI in financial services: Opportunities and challenges
The various AI initiatives offer tremendous opportunities for partners.
"AI projects often run several months and are seven-figure engagements," Synechron's Huntsman said. "In cases where they migrate from a legacy to a new system, projects often run for several years."
But deploying AI in financial services is not easy. Building a data model entails the right mix of art and science. Financial services companies want numbers to illustrate business connections -- the relationship between advertising on a social media platform and new sales leads, for example. But drawing such deductions is subjective. Regularly, AI models are developed and tested, but do not generate the expected results.
In some cases, the data itself is the problem. If the underlying data is inaccurate, then the model is useless. In many cases, data was collected, categorized and stored for one reason and is now being used for another.
"First, a corporation needs to understand its own data integrity issues," Insight Enterprises' Bienfang said. "With AI or machine learning applications, garbage in, garbage out."
Additional reporting by John Moore.