Figure Technology Solutions, Inc., a technology platform that powers a more efficient and liquid marketplace for financial products, announced its first major release developed on OpenAI’s GPT. The tool, a “stare and compare” killer, aims to reduce costs, manual work, and bias while increasing speed and customer satisfaction across loan origination and purchasing processes to benefit lending partners and consumers.
The company set out to use AI to tackle lending’s biggest, costliest, most time-consuming headaches first, starting with document reviews. To date, loan processors and originators have largely leaned on a literal “stare and compare” method of reviewing documents – meaning staring at reams of side-to-side loan documents and data and comparing the information to ensure accuracy – often resulting in human errors and biases, loan processing bottlenecks and customer dissatisfaction.
To create its OpenAI-powered “stare and compare” killer, Figure developed a four-phased process. First, it extracted twelve areas of data points that were traditionally key to the “stare and compare” process. The team then built an AI-driven model including more than six years of origination data, 168K+ total documents, 1.7M+ pages and 2M+ attributes. From there, a “Curation team” of customer-facing employees provided feedback based on hours of daily interactions with applicants and the technology; this feedback enabled model fine-tuning for accuracy and speed improvements.
Also Read: Grid Dynamics Expands with JUXT, Boosting BFS Expertise
Initial results include:
- A 93% improvement in manual upfront document review labor, which reduces the high costs and unnecessary bottlenecks typically associated with processing loans;
- When applied to efficiency savings garnered by Figure’s recently launched AI-powered customer service chatbot, Figure expects to remove more than $8 million in costs;
- An NPS score of 80 – a two point increase from an already exceptional 78; and
- A 100% YoY increase in 5-star reviews mentioning speed.
“We’ve yet to find a lender who doesn’t want to reduce bias, lower costs, and eliminate manual work or isn’t eager to improve accuracy and customer and employee satisfaction. Now, thanks to AI, a technology leader like Figure is even more well positioned to help lenders solve for these big pain points,” said Nicole Beaulieu, Chief Technology Officer at Figure. “Our growing generative AI investment is a prime example of how we’re building a technology platform at the forefront of modernizing the mortgage ecosystem to reduce costs and provide value to lenders and consumers alike.”
“We’re now embedded in more than 120 lending partners, and we hold ourselves to the highest standard of technology to keep raising the bar for them, with a series of innovations that exceed their expectations. We consider our AI our lending partners’ AI, so we designed it with them in mind – for their own funnels, processes, and customers,” said Michael Tannenbaum, Chief Executive Officer of Figure. “Bigger picture, by lowering loan processing costs, we’re lowering our product costs and thereby housing costs broadly.”
Today’s launch comes on the heels of three other actions Figure has taken to create a turnkey technology solution:
- Launch of an AI-powered chatbot. The custom-tuned large language model (LLM), powered with Google Gemini, has already resulted in reduced customer service transaction time, while improving both employee and customer experiences.
- A move to a data-based home valuation vs. the traditional appraiser model that almost all lenders still use.
- Automation that assists in faster income identification and verification vs human underwriters.
Source: Businesswire