Mortgage brokers are slowly embracing AI tools like ChatGPT

Dustan Woodhouse may have been a C+ student in high school way back in the 1980s, but he knows a lot about writing. He penned four books on the mortgage business within a decade, has over 400 podcast episodes online, and maintains a lively LinkedIn presence.

ChatGPT still blew him away. While the Vancouver-based broker says he can outperform a less experienced writer, the generative AI tool now allows such writers to hold their own. “ChatGPT is the game-changer,” Woodhouse says. “It just absolutely levels the playing field.”

AI is creeping into every industry imaginable, from oil and gas to fintech, and the mortgage industry is no exception. While most publicly available AIs like ChatGPT or Bard can’t cold-call a customer, assure a worried customer, or sign off on a deal, some Canadian brokers are toying with AI as a way to cut down on a variety of tasks.

Some brokers say they’re saving time crafting emails, content plans, and product research to focus on customers. But the industry, overall, is relatively slow-moving when it comes to AI. Woodhouse estimates maybe 5% of the brokers he’s surveyed say they are using it, largely due to a potent mix of overwork, stressful market conditions, and some very real concerns about the ethics and reliability of AI.

Always be ChatGPT-ing

Nolan Smith, owner of Oceanvale Mortgage & Finance, is one who didn’t have his head in the sand on AI. He says he was messing around with ChatGPT before it really blew up. A friend of his in real estate had introduced his manager to the generative AI tool, and Smith began using it to help write blog posts and boost search engine optimization.

With the help of Chris Johnstone, an online marketing coach for Canadian mortgage brokers and President of Connection Incorporated, Smith is bringing in ChatGPT to help him figure out the best keywords to include in blog posts on his website.

But it also acts as a general research tool for mortgage marketing. “If I don’t understand a software, or a way into lead funnelling, or landing pages, I’ll just ask ChatGPT about that,” he says.

Chat GPT chatbot

Unlike Google, which already allows you to pull up answers to these questions, ChatGPT sifts through every result possible and compiles it all into one handy AI-written precis—a perfect tool for a busy broker. As Woodhouse points out, so much of what a broker like him does on a daily basis is fix miscommunications or misunderstandings between other brokers and clients.

“Arguably, mortgage brokering is about communicating complex financial terms, topics, and policies in simpler language,” he explains. Underwriters and brokers speak their own dialects, Woodhouse jokes, and ChatGPT can act as a helpful translator. He’s also used it to help him put together a couple of scripts for his podcast (all of which are disclosed beforehand).

“The number one spot for AI in a mortgage broker’s world is, as I see it right now, with written communications,” Woodhouse says, “whether that’s a blog post or a script for video content.” But he has other uses for it, too. ChatGPT helps Woodhouse, who describes himself as ‘rather verbose,’ proofread long-winded emails to complicated questions from clients.

And ChatGPT can do even more than tidy up a lengthy email these days.

Johnstone says brokers can save themselves about an hour a day by letting AI generate email responses to clients—and even adopt different tones for different situations. “At that point, you’ve got 100% unique communication with everybody that you communicate with, even though it’s being done by AI,” he says.

Learn to code

While much of the spotlight around AI in the mortgage industry is focused on scripts, blogs, or emails, AI is also a potent research tool. It can help brokers dig up all of the details a customer might need to make a more informed choice about mortgage products.

Geoff Willis, president and CEO of Newton, says regulators like the Financial Services Regulatory Authority of Ontario (FSRA) are keen on brokers getting customers more involved in the process.

AI-fuelled mortgage research, he says, “will probably accelerate the ability to serve up that content.” While programs like ChatGPT do contain knowledge gaps for events after 2021, consumers can use it to look up general mortgage-related terms and regulations, rather than bugging their broker.

And AI involvement in mortgage research doesn’t stop there. In a couple of cases, proprietary AI programs have been used to facilitate mortgage transactions. Celligence, a U.S. fintech, announced in May that it completed an end-to-end home purchase driven by an AI system known as MORGAN.

According to a statement from Pavan Agarwal, Celligence’s CEO, MORGAN got a conditional approval for an unnamed buyer in just 30 minutes, sent him daily real estate listings according to his criteria, and determined an offer price. It also sent proof of funds to his listing agent, created offers and counteroffers, and worked alongside the real estate agent to close the deal.

Smith happens to know Celligence’s founder. “When I learned that,” Johnstone says, referring to the May transaction, “I knew that AI was going to change our industry.”

While some brokers don’t think an AI can eliminate their profession, the idea of companies using proprietary AI to move in on the mortgage business isn’t entirely far-fetched.

AI not yet on the radar for some

Trust is one of the biggest stumbling blocks for AI adoption at the moment. AI isn’t perfect. It creates legal cases that don’t exist, violates its own ethical standards against giving weapons-making advice, and may be committing rampant copyright infringement.

In the mortgage business, trust is everything. Getting a frustrated, nervous client to sign off on a mortgage—likely the most high-stakes financial deal of their lives—requires a great deal of it. Woodhouse suspects the relatively low adoption rate (or simply the rate at which brokers say they’re using AI) may be due to a general distrust of AI-generated content.

But despite pro-AI brokers like Smith who insist brokers will quickly be left behind if they don’t get up to speed on tools like ChatGPT, there may be another reason why the industry is so slow to adopt AI—it is way too busy.

Between Canada’s record-high home prices, a lack of interest rate reductions in the near future, and the mortgage stress test, Woodhouse says it is perhaps the toughest time in history to be a mortgage broker.

Unless someone can show brokers an AI tool that helps them find more highly qualified clients capable of passing today’s strict financing requirements, Woodhouse says AI won’t get their attention. “The majority of brokers in our industry right now are in survival mode,” he says.

And AI can’t push a client who’s just short of the stress test over the line. Nor can it magically lower house prices, interest rates, or down payments. But it may act as a way for brokers to find the clients that can put themselves over the line, and close a commission.

As Woodhouse puts it, “I don’t think anyone’s quite unravelled that puzzle just yet.”

Watch for Part 2 where we’ll dive into the applications of AI in mortgage originations.

Leave a Comment