Key facts
- Mortgage AI is evolving beyond simple task automation to handle complex reasoning.
- Generative AI, like ChatGPT, performs simple tasks but requires user context.
- Agentic AI can perform rule-based tasks but has brittle guardrails.
- Enterprise General Intelligence (EGI) can reason across entire loan scenarios, including appraisals and titles.
- Human oversight remains essential for final underwriting decisions and monitoring AI performance.
- AI is expected to increase mortgage productivity by allowing professionals to manage more loans via AI assistants.
Mike Romano, Director of Sales at JazzX AI, explained that artificial intelligence in the mortgage industry is progressing beyond basic automation to tackle more complex tasks, including reasoning across entire loan applications. In a conversation with HousingWire, Romano differentiated between various AI technologies, starting with generative AI, which he described as similar to ChatGPT, capable of summarizing information or performing simple tasks but requiring explicit user direction.
He then discussed agentic AI, which can execute specific tasks based on predefined rules but struggles with unpredictable scenarios common in mortgage lending, such as changes in documentation that could alter a loan's outcome. Romano introduced enterprise general intelligence (EGI) as the next evolutionary step, envisioning AI coworkers that can think and reason across all aspects of a loan, from income and assets to appraisals and titles, and collaborate with other assistants to perform significant portions of a human's work.