Artificial intelligence is becoming more common in the workplace, but many mortgage professionals still struggle with knowing which tools to use and how to get started.
“People are eager to use these new tools, but the hardest part is knowing where to start,” said Tom Hall, founder of BluMortgage. However, he added that once people start experimenting with it and trying new things, they can learn and become proficient very quickly.
Here’s a guide to which AI tools are best suited for different tasks and how to determine the right subscription level for your needs.
AI can support a wide range of functions across many areas of the mortgage industry, including:
- Sales and marketing activities: AI can create social media posts and write blog content.
- Customer service: It can assist with replying to emails, overcoming rate or product objections, refining submission notes to lenders, and communicating with referral partners.
- Brainstorming and research: Mortgage professionals can use AI as a research assistant or sounding board to work through problems and stay current on market developments.
- Office tasks: AI can support tasks like income verification by reviewing pay stubs and flagging gaps or patterns a person might overlook.
- Customer relationship and database management: With AI running in the background, it can identify opportunities within a broker’s database. For example, it may flag clients who are eligible to refinance at a lower rate without penalties. “It almost generates business on autopilot,” said Reuven Gorsht, co-founder and CEO of Deeded, a technology-enabled platform for closing real estate and mortgage deals.
With so many possible uses, different AI models are better suited to different tasks. Each one comes with its own pricing tiers, but Gorsht says a higher price doesn’t always mean better performance—just as a $1,000 set of golf clubs won’t help much if you’re still learning the game.
He recommends starting with a basic version and only upgrading if your needs go beyond what’s included. With AI evolving so quickly, it may just be a matter of waiting a few weeks for new features to appear.
It’s also worth noting that AI is generally most effective at making tasks you already know how to do more efficient, rather than helping with things you’re unfamiliar with. “There’s a big misnomer that AI is a silver bullet and it will help me do stuff that I don’t do well,” Gorsht said. “That’s generally a bad idea because it amplifies the bad.”
But when it’s used to support processes you already understand, here’s how Gorsht says each AI tool performs best:
ChatGPT
Gorsht says ChatGPT is best for a range of tasks, including writing emails, crafting responses, and doing light research. “It’s a good all-rounder,” he said. The paid version, available for $20 a month, includes features like unlimited image generation and advanced voice capabilities.
Claude
Claude is more powerful when it comes to deep research, complex problem-solving, and in-depth analysis. Gorsht said it’s especially well suited for office tasks and database management. There’s a free version, as well as a pro tier for about $27 a month and a max tier for around $136 a month.
Perplexity
Similar to Claude, Perplexity excels at analysis and coding for back-end office tasks and managing client databases. It can also generate interactive dashboards to help mortgage professionals understand trends and solve problems. A free version is available, with a professional plan priced at $20 per month.
Gemini
Google’s Gemini is strong in analytics and effective at generating images and videos. Gorsht described it as a “Swiss army knife”—versatile across many tasks, though not necessarily the best at any one thing. The paid version is available for $27 a month.
For those new to AI, the idea of incorporating it into daily work can feel overwhelming. But for early adopters like Hall, the key is simply to start small and build from there.
Hall has been using the paid version of ChatGPT for the past two years. He likes that it includes voice interaction, and says he often chats with it while driving.
For example, on a recent drive to work, he says he had a conversation with ChatGPT about macroeconomic conditions in the U.S. and Canada, and what they might signal for the Bank of Canada’s next rate decision.
For Hall, ChatGPT is a valuable tool for brainstorming and working through ideas. When it comes to capturing and organizing conversations with others, he turns to tools like Fathom and Otter to transcribe and summarize meetings.
However, Hall says he doesn’t just use ChatGPT to speed up parts of his work, he uses it to support nearly every aspect of his job.
“It’s this big debate right now: is it going to replace us or is it going to augment?” Hall said. “I’m very much so—right now, at least—in the augment camp.”
He said AI can help complete routine tasks with far less upfront effort. But he emphasized the importance of providing clear input from the start, and taking responsibility for reviewing the output.
“The whole middle part is where AI can come in and make that process much more efficient,” he said.
Gorsht echoes this sentiment, noting that if mortgage professionals don’t provide quality input, they shouldn’t expect quality results.
“Garbage in, garbage out,” he said.
One of the main criticisms of AI is that it can make mistakes, from generating inaccurate information to offering misleading suggestions. But Hall says that shouldn’t stop mortgage professionals from exploring its potential.
“Don’t just write it off because it made a mistake,” Hall says. “Understand that it’s one of the limitations.”
He notes that AI can sometimes generate misleading answers, either to give you what it thinks you want or to appear more confident than it should. That’s why, he says, it’s essential to proofread the output and question anything that seems off, especially if the source of the information isn’t clear.
“You still have to know what you’re doing,” Gorsht added. “Think about it as a tool. So whatever outcome the tool produces always needs to be managed by an individual, and when the tool does not produce what you expect it to, you can always course correct.”
Because AI is built for public use, it raises questions about privacy and regulatory compliance.
While the industry already follows established privacy and confidentiality rules, those frameworks were created before AI entered the picture, and professionals hold varying views on how best to protect client information in this new context.
Gorsht’s rule of thumb is to always anonymize data when inputting it into a large language model. For example, if you’re asking AI to draft an email, it’s best to leave out the recipient’s name or other personal details.
“I wouldn’t want to put anything that I wouldn’t want to put, say, out in the internet,” he said. “Especially with tools that are free, they’re monetizing it in some way, shape or form. So if you’re not paying for it, you’re likely the product.”
Hall, however, takes a different view. He believes it’s fine to include someone’s name in an AI prompt, just as you would when using other everyday tools like Gmail.
“It’s a double standard when it relates to these tools they’re already using where they’re putting all sorts of personal information,” Hall said. He emphasizes that while it’s important to be mindful of privacy, concerns around it shouldn’t prevent mortgage professionals from exploring the benefits of AI.
“I don’t think that AI is going to replace us,” Hall said. “I think the mortgage professionals that use AI are going to replace the ones who don’t.”
Visited 998 times, 998 visit(s) today
AI AI in mortgages artificial intelligence BluMortgage ChatGPT claude deeded mortgage broker technology perplexity technology tom hall
Last modified: July 7, 2025
