In talking to individuals who attended the recent technology conference I was somewhat surprised that many brought up the fact that the industry was not only ready, but looking for the opportunities to run their companies using robot or “BOT” technology.
While the idea that companies would be run completely by some type of “bot”, whether intellectually, or physically is still far from reality, this conference seemed to be giving signals that we are headed that way. In reviewing various summaries of the conference, it was readily apparent that those involved in the technological side of the mortgage industry are looking toward the implementation of joining current technologies to make this happen. In other words, having progressed through data consistency, compliance issues, rule-based artificial intelligence and OCR opportunities, mortgage technologists are looking forward to combining them into the ability for the technologies alone to conduct functions that are currently being completed through an interface with company personnel. While this “BOT” approach, is already in place in parts of the industry, the mortgage application process comes to mind, it has not yet reached the potential envisioned in the early 1990’s when the first automated underwriting tools were developed.
While no one, even those who believe strongly in the potential of AI, think the industry is on the brink of replacing humans with a machine, there are many areas where this approach can provide significant benefits. Many of these areas have been within the scope of industry visionaries for years. One that I am most familiar with is the one I generated for automated quality control.
In 1995 I co-authored an article in Mortgage Banking magazine about the changing role of QC in the new area of automated underwriting systems. Prominently featured in that article was the idea of pre-funding QC which could be built off the systems created for underwriting automation. Even though it took a catastrophic collapse of the mortgage industry for anyone to recognize the value, lenders are now required to conduct such a review. However, it has not played out the way I envisioned it in that article where automation would provide the function.
Despite the rejection of the concept I plowed ahead with visions of automated quality control and the potential value it had to the industry. In my mind the existing quality control function would be replaced by a rule-based engine that would test each step of the process at it occurred. This program would then culminate in a score that would identify performance risk due to the lender’s mistakes or variances from guidelines. This score, and the underlying data, could be used to identify process weaknesses as well as give investors an accurate risk of the loans they were buying.
Eventually I did create this program based on a risk model that I developed and which has a patent pending. Unfortunately, I tried to commercialize it in the early 2000’s when the only concern was the avarice opportunities in the industry. The program has been sitting waiting for the right time and place. By staying the course I may finally see my vision a reality.