How GPT-4 Is Transforming Mortgage Lending at Fintech Startup Figure

Lending startup Figure has announced the launch of a new AI tool designed to streamline the home lending process. Powered by GPT-4, the tool is aimed at reducing errors in lending documents, marking a significant step in the company’s push toward AI integration.

Figure’s Unique Approach to Home Equity Lines of Credit

Founded in 2018, Figure specializes in helping consumers secure home equity lines of credit. By offering an all-online process, the company has reduced what is typically a 45-day process down to just 5 days. More than half of Figure’s business is B2B, partnering with companies like solar panel loan provider GoodLeap.

With over $1.5 billion raised and a valuation of approximately $3 billion, according to PitchBook, Figure is now expanding into AI under the leadership of new CEO Michael Tannenbaum, former COO of Brex.

CEO Michael Tannenbaum on AI’s Impact in Fintech

Tannenbaum, who joined Figure after his tenure at Brex, views AI as a transformative force in fintech. The company’s new AI tool addresses the “stare and compare” aspect of the lending process. For example, it verifies that a property-level description—a crucial asset identifier—remains consistent across multiple legal documents, a task that once required manually reviewing over 60 pages. With AI, the verification process is faster and more efficient, drastically cutting down manual labor.

“AI is taking costs out of complex processes,” Tannenbaum explained, emphasizing how automation can reduce time and effort in the mortgage industry.

Ensuring Privacy and Flexibility with AI Models

Given the sensitive personal information involved in loan applications, Figure worked closely with OpenAI to ensure privacy protection and to prevent customer data from being used to train AI models. The company has prioritized building model-agnostic systems, which allows them to quickly pivot to the best-performing AI models as they become available.

“We’re constantly testing and evaluating different models as they come out,” said Ruben Padron, Chief Data Officer. Their systems "offer us a lot of flexibility to allow us to quickly and dynamically pivot to whatever vendor is offering the highest performance.”

The Future of AI in Lending

Looking ahead, Figure plans to continue integrating AI into the lending process, with a focus on reducing errors, minimizing bias, and lowering costs. As Padron pointed out, “We’re really trying to lower the cost, eliminate the manual work, reduce the bias. It’s very much a journey, not a destination.”