Executives are excited about artificial intelligence. According to Venture Beat, 90 percent of C-suite executives say that AI is the next technological revolution. Financial services firms have already embraced AI for risk assessment, customer care and cognitive digitization, but Forbes notes that 51 percent of companies cite cost reduction as the primary benefit. How do financial firms leverage AI in banking to create significant cost takeout?

The data dilemma

Big data poses big challenges for financial firms. Unstructured data — also called dark data — is an even bigger challenge, as it outpaces its more organized counterpart in size and scope. As Financial Director notes, many firms rely on unstructured data spread across sources — Excel spreadsheets, end-user computing tools, physical documents, legacy software systems — and financial enterprises often find themselves using time- and cost-consuming physical processes to bridge this data divide.

Consider companies dealing with client onboarding experiences, such as those in bank loan documents, HR services and mortgage processing. They need to help clients leverage their dark data by creating reliable templates for critical documents. Many of these clients provide physical boxes full of hard-copy documents, which are then manually scanned to create consistent templates and facilitate digitization. Other client data is structured but frequently changing, which is also challenging and time consuming to manage. The variety of the data formats and the complex nature of the semi-structured data impedes quick turnarounds; onboardings can take weeks — sometimes months — to complete.

IBM’s Watson for Finance, along with other products and open source capabilities, can offer these companies an alternative: AI-driven cognitive and machine learning tools capable of ingesting scanned documents and intelligently classifying them on demand.

Cost takeout: The big three

The shift to AI helps address three key areas in cost takeout: cycle time, defect rate and solution spend.

  • Cycle time: Time spent cleaning, digitizing and onboarding document templates is time wasted. By automating the digitization process, organizations gain significant cycle time reductions, which in turn reduce total spending. With enough data to work with, AI tools can help financial institutions shift from a Blockbuster-type digitization process — one that relies on companies knowing where to look or risk wandering aimlessly — to one that resembles a combination of Netflix and TurboTax: intelligent suggestions accompanied by requests for more information to narrow search parameters. The result? Vastly improved cycle times and the ability to redeploy human capital to other projects.
  • Defect rate: Errors are costly. They’re also unavoidable — remember, to err is human. As noted by recent survey data compiled by Netwrix, human error is the leading cause of financial data breaches. Automation in banking systems provides a way to significantly reduce error rates without increasing complexity or cost. For example, mortgage income validation via mobile devices can be automated with a 90 to 95 percent accuracy rate, according to IBM estimates, while document digitization for financial firms can reach 90+ percent accuracy. What’s more, AI solutions are capable of handling unstructured data and rapidly changing structured data sources without a commensurate increase in error rates.
  • Solution spend: Spend on document digitization is substantial. The traditional onboarding process can run a single large department hundreds of millions of dollars, based on data IBM has compiled. By deploying AI tools that can be 80 percent automated and are capable of 90 percent accuracy, firms can reduce this onboarding process to six simple steps that focus on data validation rather than physical presentation and scanning. That’d save approximately 30 to 40 percent of the original onboarding spend. Such a switch would also lead to reduced error rates and better use of employee effort.

Market forces

The market for AI in banking is expanding as organizations leverage open-source and proprietary solutions to meet evolving expectations, access dark data and — most importantly — improve cost takeout. In the near term, this means leveraging best-of-breed solutions to reduce cycle time, decrease defect rate and cut solution spend. In the long term, it means using AI-driven automation and cognitive services across the financial industry, allowing banks, mortgage processors, insurance companies and more to streamline internal document onboarding and collaborative interoperability.

 

Learn more about IBM’s big data services and AI consulting services.

Was this article helpful?
YesNo

More from Automation

Enhancing observability with chaos engineering: Steadybit integration with Instana

3 min read - In today's dynamic software landscape, maintaining high performance and reliability is crucial for businesses. Achieving this requires effective observability and two powerful tools to accomplish this are Steadybit and Instana®. The seamless integration of Steadybit with Instana unlocks proactive reliability engineering techniques and a comprehensive solution for optimizing and managing your applications. Steadybit chaos engineering platform Steadybit is a resilience testing platform that is designed to proactively identify weaknesses and potential failures in distributed systems. It empowers organizations to build…

What is managed DNS, anyway?

3 min read - Managed DNS is where a third-party hosts and optimizes your DNS resolution architecture to provide the fastest, most secure, most reliable experience. Perhaps the easiest way to explain it is by looking at the opposite scenario: what if you don’t have a managed DNS service in place? Every query in the Domain Name System (DNS) follows the same logic to resolve IP addresses. If the DNS records for sites aren’t found in the local cache, DNS will progressively query a series of…

The difference between ALIAS and CNAME and when to use them

3 min read - The chief difference between a CNAME record and an ALIAS record is not in the result—both point to another DNS record—but in how they resolve the target DNS record when queried. As a result of this difference, one is safe to use at the zone apex (for example, naked domain such as example.com), while the other is not. Let’s start with the CNAME record type. It simply points a DNS name, like www.example.com, at another DNS name, like lb.example.net.  This tells the resolver to look…

IBM Newsletters

Get our newsletters and topic updates that deliver the latest thought leadership and insights on emerging trends.
Subscribe now More newsletters