Rocket launches patented AI-driven tech platform

7 months ago 11

Rocket Mortgage’s bet on artificial intelligence (AI) has resulted in a patented platform called Rocket Logic, which can automatically process nearly 90% of data points extracted from documents in a month and reduce by 25% the number of times a professional interacts with a mortgage loan year over year, the company claims. 

The launch is part of the decision of its parent firm, Rocket Companies, to put AI at the center of its business strategy. It’s supported by Varun Krishna, a veteran in the financial technology world and former executive at Intuit and PayPal, who was appointed as the company’s new CEO in July 2023, replacing Bob Walters. 

Under Krishna’s leadership, Rocket piloted an AI virtual assistant with 325 mortgage bankers for outbound client calls in the fourth quarter of 2023. The assistant automatically transcribes, summarizes, and populates hundreds of crucial application fields hands-free in real-time, enabling bankers to be more productive. 

In February, the company expanded its board of directors from eight to nine, onboarding the AI and fintech expert Alastair (Alex) Rampell to provide insights as the lender pursues its mission of AI-powered homeownership. Rampell co-founded several fintech firms, including Affirm

In the meantime, Rocket developed its patented platform, which the lender claims relies on more than 10 petabytes of proprietary data, 50 million annual call transcripts, deep learning, and generative AI systems. The lender says the technology allows the company to close loans 2.5 times faster than the industry average. 

“By leveraging data and advanced AI, we are streamlining the loan origination process from application to closing, helping our clients home with speed and certainty,” Josh Zook, chief technology officer for Rocket Mortgage, said in a prepared statement. 

Rocket Logic scans and identifies the files when the client uploads its first document to the lender’s portal, securing the correct documentation. Then, the system’s computer vision models extract pertinent information and process the documents instantly. It lets clients know quickly if they are qualified for a mortgage. 

However, Rocket says, “The system is built on deep learning algorithms to increase efficiency and accuracy while simultaneously re-enforced by underwriter oversight to continuously improve the process.”  

The platform automatically identifies nearly 70% of the more than 1.5 million documents received monthly. In February, it also processed about 90% of the 4.3 million data points extracted from documents such as W-2s and bank statements, saving 4,000 hours of manual work, according to the company’s estimates. 

Regarding loan origination, Rocket said its platform “helped shave several days off the amount of time it takes Rocket’s clients to close on a home purchase, decreasing turn times by 25% from August 2022 to February 2024.”

Additional products, including from external partners, are expected to be added to the platform in 2024.

Rocket executives have expressed confidence that investment in AI would accelerate the company’s profitability. In 2023, the firm lost money, with a GAAP net loss of $390 million.

Brian Brown, the company’s chief financial officer, told analysts in February that increased productivity thought AI would bring faster turn times and better client experience. With AI, Rocket can get loans off the balance sheet faster, lowering financing costs. 

“So, we completely believe that these will translate into financial metrics and, frankly speaking, they already have, through some of the investments in 2023,” Brown told analysts. “We believe we can keep our fixed costs relatively flat (with AI) while originating significantly higher volumes.”

Article From: www.housingwire.com
Read Entire Article



Note:

We invite you to explore our website, engage with our content, and become part of our community. Thank you for trusting us as your go-to destination for news that matters.

Certain articles, images, or other media on this website may be sourced from external contributors, agencies, or organizations. In such cases, we make every effort to provide proper attribution, acknowledging the original source of the content.

If you believe that your copyrighted work has been used on our site in a way that constitutes copyright infringement, please contact us promptly. We are committed to addressing and rectifying any such instances

To remove this article:
Removal Request