- Key insight: Advanced AI in financial services is no longer a question of if, but where.
- What's at stake: Deployments abound, but questions of data provenance, resiliency and accuracy still remain.
- Forward look: Many banks are thinking about or putting in place data warehouses that will help them ease the challenges of implementation and integration.
Generative AI has reached the phase of adoption in financial services where tech leaders in companies are focused on adding new use cases and working out kinks like data access and governance.
This is reflected in the conversation at Amazon Web Services' Financial Services Symposium Thursday in New York City. Bankers here are speaking of several practical, everyday uses for the technology, and the challenge of dealing with data in silos comes up frequently.
Here are a few examples.
US Bank: Not making customers repeat themselves
U.S. Bank is using AI to do something that sounds simple, but that few large banks can actually do: take care of customers' queries and complaints without switching them to different departments and making them repeat themselves.
Andy Bingenheimer, executive vice president and corporate CIO at the $692 billion-asset Minneapolis bank, described deploying Amazon Connect in the bank's contact centers at an AWS event in New York on Thursday.
U.S. Bank has 13 million consumer clients and 1.4 million business clients. Its call centers handle 15 million calls and its 70,000 customer service agents talk through 65 million minutes of customer conversations monthly.
Amazon Connect provides context for what the customer is asking or complaining about, bringing any earlier interactions to the customer service rep's screen, whether they were in other phone calls, chat sessions, emails or text messages.
"We know that our client base is changing," Bingenheimer said. "The expectation around customers having a relationship with a single bank for the entirety of their financial life is changing. We know now that customers have choice and tend to spread their relationships across multiple companies. We're competing with that last great experience that our customer had, regardless of whether it was with a bank or another company."
U.S. Bank's goal was to provide consistency across different communication channels, making sure any handoffs are seamless and that customers don't have to repeat themselves.
"It was a challenging project," Bingenheimer said.
U.S. Bank's
With Amazon Connect in place, a customer might check the mobile app, see a transaction he doesn't recognize and click a link to connect with a live agent.
"This situation is pretty stressful, so having an agent who is there and available to quickly reassure the customer, walk them through a process," is critical, Bingenheimer said. The customer can later check the status of the disputed transaction through the mobile app or website.
In the future, U.S. Bank wants to provide support for multiple languages to the system. It wants to become more proactive with personalized recommendations.
The bank announced Thursday an expanded collaboration with AWS. Under the agreement, U.S. Bank will migrate hundreds of mission-critical banking applications to AWS. The multiyear initiative will modernize U.S. Bank's payment processing systems and wealth management platforms.
CoBank: Creating digital twins
To wring knowledge out of senior employees who are planning to retire, CoBank in Denver plans to use AI to build digital twins of them.
CoBank, which has $330 billion of assets, has significant exposure to agricultural lending and commercial and industrial lending, including funding data centers and other forms of utility infrastructure. Four months ago, the bank launched an innovation lab and an internal incubator called Forge. Employees have submitted 75 AI use cases.
One use case the bank has deployed is a knowledge assistant in Amazon Bedrock that can consume and analyze data from hundreds of leasing and other documents stored in Microsoft Sharepoint. The agents can consume large amounts of information and provide quick summaries. Now the bank is moving into AI agents that can summarize information and take action.
Another use case Alp Basol, head of strategic innovation and AI Center of Excellence at CoBank, is thinking about is using AI to cope with the intellectual loss as people retire.
"What happens if I turn around to one of my retiring folks and say, for the next six months, AI is going to do interviews with you and look at workflows and I'm going to create a digital twin of a mentor that's leading the business," Basol said.
Rocket: Building a 360-degree view of customers
Garima Sharma, vice president of data engineering at Rocket, said the online mortgage provider has been working to improve customer experience for its digital mortgages and get to "more AI connected customer interactions throughout the [customer] journey.
"Every step we took forward, the amount of data generated by customer activity grew exponentially, and with that came a realization that we required a common understanding of our customer across the entire business value chain," she said.
Her team built a unified data foundation using Amazon Redshift and Snowflake.
"It was a decision to unify everything, bring data in one place, process it once and make it usable for everyone across the business," Sharma said. "Data was no longer fragmented. Teams were moving faster. They had access to connected customer insights to personalize, to apply AI or to automate."
After Rocket
The data foundation helped make this possible. During the integration of data from Mr. Cooper customers, 40,000 leads were onboarded within nine days, and the first client closed in just three days, she said.
"Unified customer understanding is far more important than isolated AI initiatives," Sharma said. "Why? Because AI will be fueled by that shared context of your business entities."
