A 24/7 Digital MLO

November 8, 2024: A Large Language Model (LLMs) is a type of AI system that predicts and generates human-like text based on the input it receives. It processes and analyzes patterns in language to provide coherent responses, such as answering questions and can even engage in complex conversations.

So far, there hasn’t been much profit from this rapidly accelerating technology, despite the billions invested by big tech. Although that might soon change.  A potentially profitable use of LLMs could be to engage with homeowners on their mortgage needs in human-like conversations.

Is it possible that mortgage loan officers could eventually be replaced by an LLM platform? Estimates for when this will happen range from “the next few years” to “possibly never.”

To fully replace a mortgage loan officer, an LLM would need to be integrated into a comprehensive mortgage origination system that can do several things:

– Provide the client with all necessary information upfront and answer questions about mortgage products.

– Collect, verify, and analyze income and asset documents, credit scores, liabilities, purchase contracts, deeds, etc.

– Advise clients on the best mortgage products based on their financial profiles and personal preferences.

– Comply with federal, state, and local regulations and manage all required disclosure documentation

-Explain, as needed, all loan terms, approval conditions, rate and lock-in options and closing costs.

– Provide additional support, as needed, throughout the application process by addressing client concerns and assisting with problem resolution.

– Ensure data security and privacy to protect sensitive client information.

Assuming all this can be accomplished, the ultimate  challenge for an LLM would be to simulate empathy and build trust through personalized interactions with the client. Achieving all of this would require not only advanced AI capabilities but also robust backend systems, regulatory oversight, and human-in-the-loop processes to handle exceptions.

Is this possible? Maybe in the next few years. Maybe never.