The mortgage industry does not need another reason to feel behind on AI. It needs a clearer view of what is worth changing. PhoenixTeam CTO Tela Mathias joined host Zeb Lowe on HousingWire’s Power House podcast to discuss moving beyond AI FOMO and toward practical changes in how mortgage work gets done. The opportunity is not simply to automate today’s steps. It is to reconsider the process around the outcome borrowers and lenders actually need.
Rethink the Process, Not Just the Task
Making an individual task faster can be useful. But if the file still waits at every handoff, the overall experience may not improve much. An AI strategy that preserves every existing step and dependency can leave the larger sources of delay untouched.
We need to ask different questions: Which activities genuinely depend on one another? Which can happen in parallel? Where are people repeating work rather than applying judgment? Those questions shift the focus from adding AI to a workflow to designing a better workflow.
Give Autonomy Clear Boundaries
Controlled autonomy is not permission for a system to do anything it can. It means being deliberate about what it may do independently, where a person must review the result, and when an exception needs to be escalated.
In mortgage, those boundaries belong in the design from the start. A useful implementation makes responsibility clear and gives people a way to intervene. Governance is part of making the solution usable, not a separate exercise after the technology is built.
Start with Something You Can Deliver
The pressure to demonstrate AI progress can encourage activity without a clear result. A growing list of experiments does not, by itself, tell us whether the business or borrower experience is improving.
Start with a defined use case that can be delivered and evaluated. Establish what success means before implementation. Depending on the problem, that might mean less waiting, fewer repeated requests, or less manual rework. These are examples of outcomes to measure, not claims that deploying AI will automatically produce them.
The point is to learn from work that reaches the people who need it, then use that evidence to decide what comes next.
Protect the Human Moments
A mortgage is not just a file moving through a process. For the borrower, it can be an unfamiliar and consequential experience. Faster processing should not come at the expense of having someone who can explain a decision, understand a concern, or help navigate an exception.
Removing repetitive work creates an opportunity to give those interactions more attention. That also means preparing the people doing the work for changing responsibilities. Process redesign needs to account for their expertise and judgment, not just the tasks a system can perform.
Listen to the Conversation
Listen to Beyond AI FOMO: Rethinking the mortgage process on HousingWire’s Power House podcast, featuring Tela Mathias and Zeb Lowe. HousingWire published the episode on November 19, 2026.
Ready to examine where AI can make a practical difference in your mortgage operations? Explore [our AI services](/services/ai) or [connect with PhoenixTeam](/lets-talk).


