By Ray Birch
SAN FRANCISCO—The day is rapidly approaching when credit union staff won’t just be working closely with each other, they’ll be handing off work to a new set of “employees”—AI “agents.”
In fact, it’s already happening on a small scale, according to Justin Wickett, CEO and co-founder of Informed.IQ. His company, which provides AI verification solutions for financial institutions and which has partnered with Origence, offers these additional “staff” to credit unions and other organizations to drive efficiencies in lending, especially when it comes to reviewing and analyzing big documents and even making decisions.
“There is an opportunity to delegate to these AI agents and take your staff and team and leverage them where they're going to be able to create the most value for the members,” Wickett told CUToday.info. “I think we're going to be working in a world where every organization, not just credit unions, has both their own personnel that they're managing, as well as these AI agents that are performing these specialized tasks. That's going to drive incredible efficiencies; it will help you do more with less and to scale the credit union’s operations without having to constantly add more headcount.”
‘Not That Far Off’
Wickett acknowledged referring to artificial intelligence as an “agent” makes the technology sound almost human, as if it’s taking a seat in an office or working remotely.
“But we are really not that far off from that, as the AI agent emulates human decision making,” Wickett stated. “So, while the AI agent is not a person, the premise here is that you can count on the AI agent to make decisions that are part of your workflow. And in the world of auto lending, oftentimes our lenders delegate to our Informed.IQ AI agents for decisions—such as whether or not they can onboard in a compliant manner. Can they pay their dealer while ensuring the retail installment sales contract is properly itemized? Has the member signed all the appropriate disclosures? Was the loan was originated in a compliant manner? (You can) really think of the AI agent as emulating human decision making.”
‘Swimming’ in Documents
For now, what Informed.IQ knows is credit unions and large consumer lenders are “swimming” in documents and unstructured data on which they're rendering decisions, Wickett told CUToday.info.
“They are regulated entities that need to make sure that when a consumer is applying for credit that their income really is as represented. That a car dealer is indeed adhering by the applicable laws, and ensuring that a gap contract or some ancillary product that's being sold to the consumer is done so with the right disclosure,” said Wickett. “We see a tremendous opportunity in the industry to leverage AI to automate a lot of the workflows that would otherwise fall to staff within the credit union.”
Justin Wickett
What AI Does Very Well
What AI does incredibly well is read text off of documents, Wickett said, before noting it can also do more.
“Oftentimes, lenders think about optical character recognition and I think that's the start of the process, for sure. But AI enables numbers and texts to be associated with key information that a credit union needs, say, a membership form,” Wickett explained. “For example, the applicant’s name and address, date of birth… AI is able to associate the numbers and the text on these documents very accurately, which really frees staff up at the credit union. Rather than having to spend days looking over these mountains of paperwork that car dealers or members are having to submit, they can delegate to the AI agent and do what they do best, which is adding that human touch.”
All of the media attention around one new deployment of artificial intelligence--ChatGPT, which hit 100 million users in just two months--has opened people’s eyes to greater uses for AI, Wickett reminded.
“As I said, in this new world you're not just leveraging your peers, your teammates at the credit union, but you're also delegating to AI agents,” Wickett said. “These agents are specialized in particular tasks to help you do more with less, and to be able to scale the credit union’s operations without having to constantly add more headcount.”
What Must be Recognized
According to Wickett, it's critical for credit unions to recognize the types of decisions that can be delegated to AI solutions, including knowing what kind of decisions the technology can make with accuracy, having the appropriate controls in place, and understanding which decisions are not yet suitable for AI.
“I think it's really important to partner with organizations that have a lot of experience with AI to harness AI in this kind of capacity,” said Wickett. “These companies have assessed the AI accuracy from a model risk management standpoint, and those organizations make for great partners.”
Just the First Wave
Meanwhile, Wickett said the chatbot-type AI interactions with which many consumers and CU leaders are already familiar are just the first wave, Wickett forecast.
“What we're seeing now is not just a chatbot; it's a bot that is capable of reading the information off of title documents and promissory notes and retail installment sales contracts and then navigating some of the historical data—what other credit unions and other lenders have originated—navigating state laws, rendering a decision, and backing up the decision with data,” he explained. “For example, saying. ‘Hey, the paystub contained within this loan jacket is actually very similar to a paystub that was flagged by another credit union as fraud recently’.”
The AI could also cite that a disclosure was missing from the loan jacket, and the state law requires the member must have been presented with this type of disclosure, he added.
“So, we should be following up with this type of disclosure, the agent would say,” Wickett said. “It’s not just a chatbot anymore, it's the next generation of that, being able to emulate a lot of the human decision-making.”
Wickett said his company practices what he preaches, noting employees of Informed.IQ have delegated some of their work to AI agents.
“It helps us to be more efficient. It makes a lot of sense for staff at credit unions to view these AI agents as partners,” Wickett said.
How Does the Interaction Work?
How does a credit union employee interact with the agent?
“How it works today at some of the large lenders we serve (is) credit union staff can receive a 100-page PDF from a car dealer, for example. That car dealer wants to get paid for an auto loan they just originated and the staff at the at the FI send that PDF to the agent and then ask it questions, via their computer, such as are there any missing disclosures, missing signatures?” Wickett explained. “Are there any issues with this loan jacket? How does how does this loan jacket compare to other loan jackets that other lenders have received from this dealer. The AI agent researches and provides back that kind of information. The whole process takes 30 seconds to a minute.
“You are a credit union loan officer sitting in front of your loan origination system, and the fields in your loan origination system are being automatically prefilled, and any kind of issues are getting automatically flagged with the help of the agent,” continued Wickett. “That lets you be a lot more a lot more tactical with your time and more focused on ultimately resolving any loan defects, rather than having to review the entire loan jacket from scratch.”
A credit union does not need 10 different AI agents, Wickett stated, “Unless you are trying to run 10 different types of functions.”
“What we're seeing more and more of in the industry is a single AI agent that's capable of making multiple decisions,” he predicted.
Will AI Take Away Jobs?
All of that brings the discussion back to a point of concern for many, including those in white collar potistions: will these AI “agents” be taking the jobs of human beings?
“This really drives more efficiency than it does threaten jobs,” said Wickett. “But it definitely is going to change the way that folks work. It is changing the work dynamic. What it does best is free people up in the organization to do what they actually enjoy the most and what they do best—which is the actual human-to-human interaction.”
More in this series: