Artificial intelligence could turn pay later from an offer at checkout into advice about whether financing the purchase makes sense in the first place. Doing that requires more than finding the lowest monthly payment. An agent needs enough context to judge financing against available cash, existing obligations and the full cost of borrowing.
For Splitit CEO Nandan Sheth, the premise begins with who benefits from the recommendation.
“The agent is working for the consumer,” Sheth told PYMNTS CEO Karen Webster.
Sheth’s own shopping experience offers an example. While looking for a watch for his wife, his AI returned the product he requested at several prices, then suggested an unfamiliar boutique brand based on what it knew about him and his family. He bought the alternative, which he said was roughly 10% to 15% cheaper.
Financing adds a different set of calculations into the mix. Joint research from Splitit and PYMNTS Intelligence cited by Webster found that 61% of consumers would take an AI recommendation for credit or pay later, while only 2% would let the agent decide on its own.
Sheth expects recommendations to come first. “The starting point will be ‘surface me the offer, I’ll make the decision,’” he said. “The context that the AI provides allows the consumer to make a better decision.”
Context could make an $80 monthly payment look quite different. Sheth described an agent that could consider the consumer’s liquidity and other payment plans while also showing the monthly payment and total financing cost.
Splitit has tested how additional financing information affects conversion. Sheth said one version presented an $80 monthly payment, the APR and total cost. The other added that there was no prepayment penalty and explained that the consumer could make the lower payments for three months, then pay off the remaining balance and incur only a portion of the interest that would have been paid over the full six-month or 12-month term.
Sheth said the version providing that additional context produced a conversion rate about two to 2.5 times as high as the version that stopped at the payment, APR and total cost.
Financial guidance also doesn’t necessarily end at checkout. Webster raised the example of a consumer who needs the lower payment initially but has enough liquidity several months later to repay the balance. She described the potential as “a dynamic relationship between agent and consumer,” informed by the timing of money coming in and going out.
Better Advice Requires More Financial Context
We’d love to be your preferred source for news.
Please add us to your preferred sources list so our news, data and interviews show up in your feed. Thanks!
Building an intertwined relationship between the user and the AI agent requires information far beyond shopping history.
Bank balances, credit card statements, revolving balances and even brokerage statements were among the inputs Sheth identified. An agent wouldn’t necessarily need all of them, he said. Just some of that data, provided securely, could improve decisions involving pay later or a personal loan. Sheth said greater access could help with decisions about investing, saving and spending, while acknowledging his own reluctance to provide the necessary information.
“I have a fear of a breach, and I have a fear of that data being misused,” he said.
Localization is one approach Sheth raised as a way to safeguard data. He described a personalized large language model (LLM) that performs much of its processing locally and reaches outside for specific tasks, limiting how much sensitive information needs to move outside the consumer’s environment.
“When will consumers be comfortable letting agents take the wheel?” Webster asked.
Sheth said the answer depends on the purchase, merchant and transaction size. He sees lower-value, routine transactions as more likely candidates for automated purchasing.
Groceries serve as a prime test case. A consumer could set a $300 monthly budget, provide information about typical purchases and discretionary items, and ask the agent to find the best deal. With a defined spending limit, Sheth said he could see the agent completing the payment.
Sheth said an AI agent with sufficient context could recommend immediate payment for one purchase and pay later for another based on the economics of the transaction.
He connected that possibility to the industry’s longstanding focus on top-of-wallet status. A consumer may have a preferred card or payment method, but an agent evaluating each purchase has another set of information to consider, including financing cost, available liquidity and other benefits attached to the transaction.
Watch the full interview with Nandan Sheth to learn more about:
- What merchants are learning about making their products discoverable to consumers using AI.
- Why Sheth believes an LLM’s accumulated memory may influence where consumers seek AI assistance.
- Where he sees unanswered questions about an agent acting as a consumer’s proxy when accepting credit.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
Add as Preferred Source
