One year after introducing Mia at UWM LIVE!, the Mia AI-powered LO assistant has evolved from a borrower communication tool into a broader engagement platform supporting millions of conversations. Driven largely by broker feedback, Mia now includes multilingual capabilities, including Spanish, expanded outreach functions and new capabilities designed to help brokers strengthen relationships with borrowers and referral partners.
In this conversation, Jason Bressler, Chief Technology Officer at United Wholesale Mortgage (UWM), discusses what the company learned from Mia’s first year in the market, how broker adoption exceeded expectations and why AI-powered communication is becoming an increasingly important part of the mortgage experience.
HousingWire: Mia was introduced as an AI-powered LO assistant designed to help brokers manage borrower communication more efficiently. What has surprised you most about how brokers have adopted and used it?
Jason Bressler: The biggest surprise was how quickly adoption changed. Early on, there was a lot of hesitation. Brokers and loan officers didn’t want Mia communicating with their leads, borrowers or past clients without knowing exactly how it would perform.
We saw a decent amount of immediate opt-outs because people simply didn’t know what to expect. But within three to four months, that changed dramatically. Brokers started seeing real results. Our CEO, Mat Ishbia, highlighted success stories in sales meetings, shared testimonials and played recordings of Mia’s conversations. Some loan officers were generating multiple new deals each week through Mia’s outreach efforts.
Once brokers saw the business impact, adoption accelerated quickly. Today, the response is overwhelmingly positive. Brokers want more functionality, more customization and more ways to integrate Mia into their businesses. From a technology standpoint, that feedback loop is invaluable because it helps us continually improve the product based on real-world usage.
HW: Mia has now supported nearly three million calls and more than 80,000 closed loans. What have you learned about what brokers and borrowers expect from AI-powered communication?
JB: At the beginning, there really weren’t many expectations because this was such a new concept, especially at the scale we launched it. We gave roughly 50,000 loan officers access to their own Mia phone number and the ability to have AI handle inbound and outbound communication.
Initially, there was a lot of curiosity and caution. Over time, however, we learned that brokers want Mia to have increasingly detailed and natural conversations. They want her to understand UWM products, know information about their business, understand licensing limitations and communicate as an extension of their team.
As adoption has grown, expectations have grown as well. Brokers increasingly want Mia to speak on their behalf and handle more complex conversations. The goal is to make those interactions as informative, personalized and comprehensive as possible.
HW: UWM launched Mia On Demand at UWM LIVE!, which now includes call options for listing-agent outreach, pre-qualification follow-up and mortgage reviews. Why were those the next use cases to prioritize?
JB: Those enhancements came directly from broker feedback. The easiest way to describe it is that many people hesitate to initiate conversations. Whether it’s reaching out to a potential referral partner or making a cold call, there’s always some level of discomfort.
Mia eliminates that barrier.
She can make those initial outreach calls, start conversations and create opportunities that many people might otherwise avoid. That opens the door for brokers to develop new relationships and create more business opportunities.
The new use cases reflect areas where brokers wanted help starting and maintaining conversations. We listened to that feedback and built solutions around it.
HW: Spanish-language support was one of the most requested enhancements. How important is multilingual communication to the future of AI borrower engagement?
JB: It’s extremely important. We have a large Hispanic broker community, and many brokers told us they were hesitant to use Mia with certain borrowers due to language limitations. Once we introduced Spanish functionality, Mia’s AI adoption increased significantly.
Today, if a borrower begins speaking Spanish, Mia automatically recognizes it and switches languages. We’re also adding functionality through Brand 360, UWM’s marketing portal for clients, that will allow loan officers to designate specific contacts for Spanish communication from the beginning of the conversation.
What’s exciting is that this extends beyond Spanish. Mia can currently support roughly 32 languages. If a borrower begins speaking Korean or another supported language, Mia can automatically transition into that language as well.
That creates a much more accessible experience for borrowers while allowing brokers to engage a broader audience.
HW: As Mia continues to evolve, how has your perspective on AI’s role in mortgage lending changed over the past year?
JB: I think Mia’s role will continue to expand significantly. She already has the capability to handle many parts of the mortgage conversation process, and over time, she’ll be able to assist with nearly every aspect of the loan application journey, except for actually quoting rates.
The future is about moving beyond simple voice AI and creating increasingly sophisticated, conversational interactions, making Mia more knowledgeable, more responsive and more capable of handling a much wider range of borrower and broker questions.
The next major evolution for the industry will likely be AI-powered call centers. Many voice AI providers can have conversations, but what’s much harder is responding intelligently to objections, adjusting in real time and handling complex interactions at scale.
That requires extensive infrastructure, training, scripting and ongoing refinement. We’ve invested heavily in building that foundation. As the technology matures, I believe the industry will see a new generation of AI capable of handling much more advanced customer engagement and support functions.
HW: UWM LIVE! also introduced Refi ’86. How do you see AI and pricing strategy working together to help brokers capture refinance and retention opportunities?
JB: Some of the most impactful AI applications in mortgage lending won’t necessarily be customer-facing. AI-driven data modeling and machine learning will increasingly power dynamic pricing strategies. Lenders will be able to evaluate not only borrower characteristics, but also broker performance patterns, geographic factors, income profiles and numerous other variables to identify the best product and pricing options almost instantly.
The ability to deliver highly personalized recommendations quickly will become a major competitive advantage. Much of that technology already exists today. The question is how aggressively lenders and brokers choose to invest in it and integrate it into their workflows.
Ultimately, technology and pricing strategy will continue working together to help brokers identify opportunities, improve borrower retention and deliver more tailored mortgage solutions.
