Industry technology provider Polly is broadening the use of agentic artificial intelligence across its mortgage capital markets platform in a move that CEO Adam Carmel said addresses a key pain-point that home lenders consider when deciding whether to change vendors.

A significant reduction in set-up time for mortgage companies adopting product and pricing engine-related technology is one upside of the new technology, said Carmel. He said that it can significantly compress the average time it takes in the industry to get a new PPE in place.

"The industry takes many months. Our agentic implementation is four to six weeks," he said.

The current technology puts implementation on a track where the platform could be production-ready in as few as 24 days, according to a company press release. Generally, Polly can roll out new offerings to existing customers even more quickly than new ones, according to Carmel. 

While the result of the patented, agentic AI capability is faster processing, it took some time to develop.

What it took to get there

"We've been working on this for a very long time, and it's accelerated here in the last six to nine months," said Carmel. 

The process took time in part because there were risks to consider and address.

"The challenge with AI, particularly in the capital markets segment, is you cannot have something that's probabilistic, it really needs to be entirely deterministic. You can't have things be wrong," Carmel said. 

"We've been very focused and very intentional on being on the frontier of the technology and leveraging it to its fullest extent, while also ensuring that the inputs and outputs are entirely deterministic. If they're not, then you put a human in the loop, and that's the critical thing," he added.

Polly's agentic AI saves time by handling basic configuration while implementation consultants focus on edge cases, according to the company's release.

"It's allowed for our teams to be highly consultative, evaluate internal workflows and make recommendations as part of the implementation process because now we've got that bandwidth," Carmel said.

General AI advice for the industry

Carmel advises mortgage companies considering AI use in any form decide what their goal in implementing it would be, which may be return on investment. 

Whatever the goal is, he advised doing some research into the technology's nuances to get a sense of the particular automation involved would realistically get them there.

"The industry has historically been pretty slow to adopt new technology, and particularly now with the pace of things, it's getting hit in every direction by everyone claiming they've got something 'AI,'" Carmel said. "In many cases it's just not. It's a rules engine. That's fine, but like that's not really going to drive ROI."

He recommends single-family mortgage companies considering AI implementation be aware not only of whether it can fulfill their goal for it, but whether achieving this requires adjustments to their operations and staffing that they're willing to make.

"A lender may say, 'I want to drive my costs down.' Since so much of their cost structure is labor, they may then need to be willing to affect that change," he said. "There's an opportunity there but it has an impact, and it's tough."