Insights from a conversation with Neil Armstrong, Chief Revenue Officer at Indecomm, on the California MBA’s Connect Podcast:
The mortgage industry has spent years talking about automation and AI. But if you walk into the average lender’s operation today, you’ll still find underwriters manually keying in data for 28 to 30 minutes per file and getting it wrong.
That gap between what technology promises and what’s actually happening on the ground is real, and it’s costing lenders more than they realize. The problem isn’t a lack of tools. It’s a lack of clarity about process, about cost, and about where automation actually belongs.
The problem isn’t a lack of tools. It’s a lack of clarity about process, about cost, and about where automation actually belongs.
The Shift That's Actually Happening
For the past 18 to 24 months, something has changed in how lenders approach their own operations and it’s not just budget pressure forcing their hand.
The trigger is margin compression. With rates keeping volume low and costs stubbornly high, lenders can no longer afford to manage their business at a 30,000-foot view. They’re going function by function, task by task, asking a question that sounds simple but turns out to be revelatory: what does it actually cost us to do this one thing?
And, this operational assessment is an important process to go through that yields critical insights. The format is low-tech: get a team in a room, put the process on a whiteboard, and start assigning a cost to every function. What happens next almost always surprises someone in the room teams that have worked together for years discover they don’t actually agree on how their own process works.
That kind of visibility finally seeing your own process clearly is what creates real momentum for change.
Where Automation Is Actually Delivering
When lenders have an accurate picture of their process, the results of automation can be dramatic. Data from six years of Indecomm client deployments shows a 60% lift in underwriter productivity across various Genius AI implementations. Recently, that lift has been most prominent at the intersection of document intelligence and automated underwriting.
Here is where our clients are getting a lift:
Document indexing, classification, and data extraction. Moving from legacy OCR which often delivers only 78% first-pass accuracy to machine learning and AI-based, intelligent document processing is the foundational step. This is automation 101. If you get this right, downstream automation and AI will be more accurate and effective as it will be grounded in accurate data.
Third-party services ordering & review. Once fraud reports, flood certifications, appraisals, and title documents come back, those review workflows are highly automatable. A title review that takes 38 minutes manually can be brought down to 3. An appraisal review that takes 40 minutes can be compressed to 6.
QC & Underwriting Checklists/ Validation: It sounds unglamorous, but the numbers are hard to argue with: between 88 to 100% of pre-close, pre-underwriting, and QC validation checklists are automatable. The ROI is fast and the deployment is clean. If a lender has never automated anything before, this is where to start.
Breaking Down Silos From the Inside Out
When most of a workflow gets automated, the traditional job roles built around it start to become arbitrary. Those roles such as the loan officer, processor, underwriter, and closer, were designed around human limitations: people need to specialize, hand off work, and check each other’s work. Automation doesn’t have those limitations, so the old divisions stop making logical sense. The real question isn’t just “how much money can we save?” but “do we need to reorganize entirely?”
Take the processor role as an example. When routine review tasks are lifted off their plate, that freed time can be redirected toward the borrower meaning more touchpoints, better communication, stronger relationships. The result is improved sales, higher retention, and lower cost from the same headcount. That’s what reorganizing actually looks like in practice: not fewer people, but people doing fundamentally different work.
The Real Reason Automation Deployments Fail
Where do lenders get stuck? Data quality, process gaps, system juggling, and leadership alignment all play a role, but the single biggest reason implementations stall is change management. It can’t be an afterthought; it has to be built into the plan from day one. When one lender deployed automated underwriting to only half their team because the other half didn’t trust it yet, the results made the case on their own. The group using it doubled productivity and cut errors. The hesitant half came around quickly once they saw the numbers.
The second failure mode is trying to automate a broken process. Automation doesn’t fix dysfunction, it amplifies it. The sequence matters: diagnose first, fix what’s broken, then automate.
Both of these failures ultimately trace back to the same root cause: lack of leadership commitment. When a top 10 lender engaged Indecomm, the CEO personally identified the problem, set a deadline, and committed to a phased rollout. Today that operation runs like a well-orchestrated machine because leadership treated it as a strategic priority, not an IT project. There is no successful deployment that doesn’t start at the top.
Technology's Place: Behind the Scenes
With all the momentum around AI, it’s easy to start imagining a future where technology is everywhere in the borrower experience. The more disciplined position is more deliberate than that.
People are the biggest asset in any lending organization. Technology’s role is to make them more effective, not to replace the relationship. The digital assistant handles the data. The human handles the borrower.
Mortgage is not a transaction business. For most borrowers, this is the biggest financial decision of their life. Automating borrower communication in a way that removes the human from that experience is a mistake.
What automation should do is free up the humans to be more present. Give a processor three hours back in their day and let them spend it with clients. Route a servicing call to a live agent the moment someone really needs help. Give an LO richer data so they walk into every conversation better prepared.
Technology behind the scenes. People out front. That’s the balance that works.
What This Means for Lenders Right Now
The market isn’t going to wait. The lenders positioning themselves well are the ones doing the operational work now during the slow period, when they have the time and the pressure to be honest with themselves about what’s actually working.
The playbook is straightforward, even if the execution isn’t easy:
1. Assess before you automate. Map your process function by function. Assign costs. Find out where you’re actually losing time and money.
2. Fix process before deploying tech. Automation amplifies what’s already there good or bad.
3. Start with a win. Checklists, document classification, review functions pick something with fast ROI and clean deployment.
4. Lead from the top. Without executive alignment, change management fails and implementations stall.
5. Keep humans at the center. Technology is the assistant. Your people are the asset.
The lenders who treat this moment as an operational reset not just a cost-cutting exercise are the ones building something sustainable. When rates do drop and volume returns, the question won’t be whether you can hire fast enough. It’ll be whether you built a machine that can scale.
