Smaller Banks are Falling Behind on the GenAI Curve

Don’t Hesitate to Use Existing Solutions

Many bank employees are already using tech platforms, like Google and Microsoft, that have invested heavily in GenAI tools. Merely having employees experiment with these solutions that are at their fingertips can be a great first step in the AI journey. This strategy allows smaller financial institutions to begin their adoption of AI without having to make immediate, significant investments.

Having an employee base that’s grown familiar with AI and can use the technology to streamline the more labor-intensive banking processes is at least as impactful as unveiling a flashy, customer-facing AI tool. This will cost significantly less, too.

Start with simple use cases and readily available solutions, and build up from there.

At Grasshopper Bank we’ve found that agility, not scale, is the real differentiator.

Read more: Don’t Let Data Paranoia Hamper Your Bank’s Use of GenAI

An Example of Using AI to Solve Problems in the Trenches

There are dozens of back-office tasks across banks that can benefit from incorporating GenAI. Getting started with a small handful of processes to test AI with can lead to lasting efficiency gains.

At Grasshopper, our lending teams identified that repetitive tasks, like following up on missing documents, can be automated with AI. This process formerly took two to three hours per loan. Now it takes two to three minutes — creating better results for our team and our clients.

Our lending team has found several additional use cases for AI beyond automating follow ups, particularly with auto loans. Using AI to automate document collection and begin the internal sorting process has become a big timesaver. This has allowed the auto team to operate more efficiently.

Here’s how AI can improve auto lending: We receive many auto loan applications on weekends because consumers typically shop for cars on Fridays and Saturdays. In the past, those applications would sit until our loan operations team was back in the office on Monday to review and verify them.

GenAI does the grunt work. Recently, we’ve introduced GenAI to the know-your-customer process, using the technology to verify driver’s license information and other personal details.

This moves the application process along much faster — there are fewer weekend holdups, the loan operations team can perform quality control more quickly, and ultimately, the borrower benefits from a smoother, more efficient path from application to decision.

Critically, there are still human checkpoints throughout the auto loan review process — nothing can truly replace human decision-making. But using GenAI to reduce time-intensive tasks like sorting through driver’s license details allows our team to process a much higher volume of applications.

Read more: Five Real-World AI Applications That Will Boost Your Bank’s Operations

How to Find the Right Balance of AI and Human

Being practical also means assessing comfort levels with GenAI and finding ways to use it accordingly.

For example, a bank might be uncomfortable involving GenAI in credit decisioning, which could scare them off from using the technology at all.

Banks shouldn’t let discomfort with particular applications of AI dissuade them from using it. Instead, find a level of involvement that feels comfortable and makes sense. In this example, a bank could use AI to help collect data for loan reviews, instead of involving it in the actual decision-making.

When thinking about where to start, consider tasks that require minimal thought and a simple decision-making process. Take extracting KYC and anti-money laundering information. AI can pull such data from PDFs and scanned documents much faster than humans can. Other such tasks include standardizing and cross-checking vendor and client data, and sorting contracts and invoices.

Experimenting with automating these routine tasks builds comfort with AI with minimal risk.

Scaling Up with AI Requires a Deliberative Process

Implementing generative AI requires constant feedback. Banks should assess how much time a task takes to perform manually and compare that to how long it takes with the help of automation.

Processes that show significant time savings not only confirm that resources are being well-spent, but can also spark consideration for additional use cases.

These feedback loops also help identify areas that will not benefit from introducing generative AI. For example, at Grasshopper, through internal conversations, we’ve decided not to use GenAI in the credit decisioning process.

Why? Weighing whether or not to extend a loan or line of credit requires consistent, repeatable logic. Our teams have found that generative AI isn’t truly deterministic in this way — it might reach different decisions in very similar scenarios.

Read more: