Using AI to Your Advantage in Banking and Finance

It's easy to make mistakes when working with numbers on your own; for instance, transposing digits or misplacing a decimal point. But what if your bank made these mistakes? An error of this magnitude is unacceptable and will cause the bank to lose credibility with its clients. Accuracy is paramount in all industries, especially in banking and finance, which deal with numbers and sensitive issues such as client funds.

That’s why so many banks and financial companies are finding that automation is a perfect fit for them—and not just for crunching numbers. Take the issue of extracting data attributes related to corporate action announcements. That’s the use case that Frank Chen, Head of Intelligent Automation and Project Management Office (PMO) at J.P. Morgan, discussed during the UiPath Artificial Intelligence (AI) Summit 2022.

With UiPath AI Center and Document Understanding, we've been able to use an out-of-the-box, self-training machine learning model—without relying on a data scientist. That's been a game changer for us.

Frank Chen, Head of Intelligent Automation and PMO, J.P. Morgan

Chen noted that automating that process has always been a tough challenge because there are no industry standards that mandate a consistent document format. As a result, traditional optical character recognition (OCR) tools can’t effectively extract the required data. Formats and content are always changing. Historically, if a company had a template to help automate the process, any change or deviation from the template could break the automation, requiring a costly and time-consuming fix.

J.P. Morgan’s solution was a three-pronged approach built on multiple UiPath products. The first step was to use UiPath Studio to create a robot to visit the exchange website and pull the relevant action notices. Next, the action notice would then pass through UiPath Document Understanding to extract the relevant attributes and populate a template with the required information. That process relied on a UiPath AI Center data model. And according to Chen, it was this component that was a “game changer” for J.P. Morgan. 

A non-technical user from the operations team built a template and trained the out-of-the-box model to extract the right data attributes—without the help of a data scientist. Previously, the process required a lengthy collaboration with a data scientist to create a workable model. Every time changes were needed, the team had to reengage the data scientist and start the process all over. 

The third element of the approach involved UiPath Action Center. 

If new to Action Center, here's a quick video:

J.P. Morgan wanted to formalize a review process that would give users and business developers confidence that the models they built were performing as needed. And if something wasn’t right, they wanted the ability to retrain the model and continually improve its output. 

Three things for banks starting with automation

So, what advice does Chen have for organizations that are just starting on their automation journey?

1. Start small

It’s best to have a solid understanding of automation’s capabilities and make sure a tool works as it should before you expand your scope. To gain automation advocates and champions within the organization, you’ll need a successful proof of concept project.

2. Understand what—and why—you want to automate

Automation is an effective solution in many cases. However, as with any business decision, research multiple options to find the one that works best in your situation.

3. Reuse, repurpose, and scale

The initial use case you create may have a wider application. Other teams and departments may face similar challenges. Expand automation’s value by leveraging and reusing existing automation components and involving other stakeholders.

What's next?

J.P. Morgan’s application of AI and automation to incorporate action announcements is just their start. They’re beginning to explore how to put AI to work on use cases with more unstructured data, such as client instructions and private market documents. 

Amit Kumar and Nitin Purwar from UiPath closed out the AI Summit session with a discussion of some of the factors driving wider adoption of AI and automation in banking. 

Factors driving wider adoption of AI and automation in banking

First, banking is a document-intensive industry with no overarching format standards. Banks handle a vast amount of semi-structured and unstructured data, and the related processes can be very labor-intensive. Automation and AI can take over a growing share of that workload. 

Also, the industry has already automated many of its repetitive, rules-based tasks—in other words, the “low-hanging fruit.” Now companies are looking for ways to automate variable processes that mix rules-based tasks with more fluid, unstructured data. That requires advanced capabilities for machine learning, predictions, and other problem-solving abilities that mirror human cognitive skills.

Finally, every company now wants to be more customer centric.  And to deliver exceptional experiences, banks need to monitor and respond to unstructured data from a wide variety of channels, including email and social media. Technologies such as machine learning, UiPath AI Center, and UiPath Document Understanding can help banks better manage the unstructured data around the customer experience and deliver better, more responsive service.

Read more examples of the growing impact of AI and automation in banking and finance. And there’s still time to access all of the sessions from the UiPath AI Summit 2022 (available on demand).