As AI-powered tools continue to move into the mortgage business, lenders should learn how to ask questions to make new technologies work.
Prompt engineering is among the latest fad and companies are spending a substantial amount of money recruiting individuals who know how to ask the new AI tools quality questions. It turns out that AI is a lot like real people in this regard, where you ask poor questions and you’ll receive mediocre answers.
This is a very positive development as it has underscored how important good information is when trying to effectively connect a human team with a technology platform. Technology buyers are beginning to understand why their technology partners ask a wealth of questions before they begin a new initiative.
And, in truth, the new AI tools are changing the way technologists think, but these changes began long before ChatGPT entered the scene.
Asking better questions has always been the path to refined information. And having more information has always been the key to smoother and more successful technology implementations.
As AI-powered tools continue to move into the mortgage business, we’re seeing our prospect’s thinking patterns evolve.
In the past, the most important questions for new users typically included, “How big is this change you’re asking us to embrace.” No one likes change, and every executive knows the more pieces involved in the change, the lower your overall adoption rate will typically be.
In the past, technology buyers have looked for a narrow delta between their current operations and their operations post implementation. For many reasons, this was a practical approach.
But ChatGPT has disrupted this mindset. Unlike past technologies that took years to gain adoption, OpenAI’s new tool won its first million users in days, not months or years. And this, despite the fact that the delta between what yesterday’s search engines could deliver and what ChatGPT returned was not narrow.
Overnight, people realized that technology could change their lives by an order of magnitude and be just as easy as the old way. Well, almost as easy.
It is just as easy to type a question into the old Google interface as it is to type it into Google’s new Bard AI tool. The difference is that a poorly crafted question will deliver millions of website hits in the old search engine but provide only a marginally useful answer in Bard.
The same is true for today’s technology implementations. Take the new AI-powered document management technology built into MCP, for example. Lenders can now rely on an LOS that can confidently identify virtually any document a borrower provides, index it, and scrape the important data into the database while also store the digital document in the right place.
But like all the functionality built into today’s mortgage origination technology, it only works where the lender lets it do so.
And that means that new questions need to be asked prior to implementation, questions that we haven’t had to ask before. When our team knows exactly what the lender wants to do and where they want to gain efficiency, our tools can be configured to do so.
Not having those answers means that you’ll still have very powerful technology, just not see the results out of it that you would like. It’s like prompt engineering, but made easier because we know how to guide our partners into asking the right questions and providing better answers. Reach out today to find out what AI-powered functionality is available to you now.
By Jim Rosen, EVP, Services at Mortgage Cadence
Follow us on LinkedIn to be notified when our next article is released.