Flexibility by design

Why automated underwriting can scale non-agency lending if the engine bends to the lender's credit box - not the other way around


Non-agency lending is difficult for the same reason it is valuable: the products are not supposed to look identical. A bank can define one credit box, a credit union another, and a private lender something entirely different. The guidelines are often the strategy. For Derek Long, COO of LoanPASS, that means the technology used to automate those decisions cannot begin with a rigid model and ask lenders to squeeze themselves into it.


Long has been thinking about that problem for more than two decades. He entered mortgage banking in 1992, moved quickly into secondary marketing and product development, and helped create an electronic guideline system his company called the Loan Library. The idea grew into a separate technology business serving roughly 150 mortgage bankers and eventually a pricing engine. Before the 2007 market disruption changed the course of that company, his team had started work on an automated underwriting engine.


We don't hard-code things. We build flexibility into the engine to begin with.

- Derek Long


When Long met Mike Lewis and Bill Roy in 2020, the LoanPASS vision sounded familiar: build a configurable automated underwriting engine for products that did not have the equivalent of DU or LP. He joined the company and is now its COO. The through-line in his thinking has remained consistent. The engine has to be flexible enough to carry the lender's own policy, detailed enough to express real risk strategy, and deterministic enough to return the same answer when the same facts are presented.


The unfinished idea

Long's first technology venture started with a practical problem. Mortgage professionals needed a better way to find and use product guidelines, so he helped move that information from paper and scattered references into an electronic library. Loan officers responded to the speed and accessibility, and the concept expanded from one lender into a technology company serving many.


Pricing followed because pricing and eligibility live close to guideline logic. Automated underwriting was the next natural step. Long says the earlier company was working toward that capability when 2007 intervened and the business was eventually sold to AllRegs. The idea did not disappear; it simply remained unfinished.


A tool like LoanPASS can let them be very granular and more detailed and expand the credit box, safely.

- Derek Long


That history helps explain why LoanPASS appealed to him in 2020. The company was starting from the same premise Long had been circling for years: lenders with proprietary or portfolio products needed a way to turn their own rules into repeatable decisions. For Long, joining LoanPASS was a chance to pick up a thread that had been waiting for the technology and the market to catch up.


Don't code yourself into a corner

Long describes configurability as an architectural discipline, not a feature added at the end. From the beginning, the team asked how the engine could handle very different products without being rewritten every time. His shorthand is simple: do not hard-code yourself into a corner.


That meant designing with reverse mortgages, second liens, jumbo, non-QM and even non-mortgage credit in mind. The products do not share one credit policy, but they can share an approach to expressing policy: defined data, defined calculations, defined eligibility rules and defined conditions. The engine has to be able to change those components without changing its identity.


A platform intended to follow the lender's credit box rather than force the credit box to follow the software.

- Derek Long


The result, in Long's description, is a platform intended to follow the lender's credit box rather than force the credit box to follow the software. That distinction is especially important in portfolio lending, where differentiation often lives inside the details that one institution is willing to finance and another is not.


Pricing was the gateway

LoanPASS went to market first with product and pricing. Long calls the PPE a gateway project because many of the components needed for automated underwriting were already present: eligibility logic, rules, calculations and the ability to configure a lender's product structure. Pricing could be commercialized sooner while the team continued building toward the larger underwriting objective.


The work on pricing also created a useful design constraint. If the AUS was eventually going to use the same policy framework, the pricing engine could not be built as an isolated tool. Decisions about flexibility, rule writing and data had to anticipate a future in which price and credit would be evaluated together.


Long says one of the last milestones before he felt comfortable calling the product an AUS was the ability to bring a credit report into the decision. Once that input could be evaluated alongside the loan application and configured rules, the engine had crossed from product eligibility into a more complete credit decisioning workflow.



The combined model matters because a locked loan can change. Long says the real power appears when pricing and automated underwriting can be rerun together, reducing the risk that a credit decision and the economics attached to that decision drift apart as a file moves through the process.


Consistency without simplicity

Long sees the first operational benefit as consistency. Non-agency underwriting can spend a great deal of time getting to a no, and two people can interpret the same policy differently. A rules engine can return an answer quickly and apply the configured logic the same way across users and channels.


But consistency does not require the lender to simplify the credit box. Long argues that manual environments often encourage credit managers to keep policies easy enough for people to remember. Technology changes that constraint. The engine can carry more detail than a person should be expected to memorize and can evaluate combinations of factors with greater granularity.


That creates a different kind of efficiency. Loan officers can get earlier direction, processors and underwriters can work from the same decision logic, and post-closing or third-party review teams can see how the file was evaluated. The goal is not only speed at one underwriting desk; it is consistency across the life of the loan.


The credit box stays yours

Long is explicit that LoanPASS is not intended to be an off-the-shelf underwriting policy. Each lender defines its own guidelines, and the engine is configured around them. A bank, credit union or private lender can express the distinctions that make its products competitive rather than conforming to a generic model simply because the technology requires it.


That is why Long calls the implementation bespoke. The lender can change the engine as policies change, add products, tighten or expand criteria and continue refining the decision structure. The technology is meant to be fluid because the business it represents is fluid.


LoanPASS is not intended to be an off-the-shelf underwriting policy

- Derek Long


He also sees the same architecture extending beyond mortgage. If a depository proves the model with a mortgage portfolio, Long sees no conceptual reason the institution could not use a configurable decision engine for business lending, aircraft finance or other products where it controls the policy. The specific rules would change; the framework would not.


Enhanced connection rather than replacement

Flexibility does not mean trying to become every application in the stack. Long says LoanPASS is not an LOS, POS or CRM. Its job is narrower: take the lender's decision logic, evaluate the loan, return eligibility and conditions, and make that structured decision available to the surrounding systems.


That leaves room for connectivity. Long expects LOS platforms, point-of-sale tools, third-party review firms and other vendors to connect to a decision engine rather than be replaced by it. He describes LoanPASS as the smart brain that can sit inside a larger ecosystem and give those systems a consistent answer to work from.


We see our AUS as the smart brain that will be connected to lots of technology.

- Derek Long


he model is also selective about who benefits most. Long points to banks, credit unions, correspondents and private lenders that own or meaningfully control their guidelines. An independent mortgage banker selling entirely to conduit investors may have less need for a custom engine because the credit policy ultimately belongs to someone else.


Deterministic where it matters

The flexibility of the platform makes Long's view on AI especially important. He does not believe lenders should rely on a probabilistic model for the final underwriting decision. In credit, he argues, the same facts should produce the same result, and the decision path should be explainable after the fact.


That does not put AI outside the system. Long says AI can make configuration work faster by digesting large amounts of guideline information and helping convert it into the rules, calculations and data structures the engine needs. It can also help gather data that the deterministic engine will evaluate.


Long sees another role emerging around policy analysis. LoanPASS has discussed a 2027 roadmap in which AI could help compare guidelines and recommend areas a lender may want to review in pursuit of more volume, lower defaults or another defined objective. The lender would still own the policy. AI would help surface possibilities rather than approve the loan.


Control builds trust

Long connects trust to what LoanPASS calls its Q-City values: quality, collaboration, integrity, transformation and innovation. His practical version of that promise is equally direct. Do not oversell. Do not present vaporware. Build what the client can test and understand.


Client control is part of that philosophy. Long says some customers manage their own versions of LoanPASS after implementation, to the point that the company may not be involved in day-to-day changes. That shifts trust away from a black box vendor relationship and toward a tool the lender can inspect, test and govern itself.


We don't oversell. We don't provide vaporware.

- Derek Long


Long estimates a working implementation can be configured from a lender's guidelines in roughly two to three months, after which the depth of the build becomes a balancing act. The more data the lender wants to collect and the more detail it wants to encode, the more fields, rules and conditions can be added.


The underlying governance question remains with the institution. LoanPASS can make policy executable, but it does not decide what the policy ought to be. Control over the credit box is preserved precisely because the lender is expected to define, test and maintain the rules it wants the engine to apply.


Risk can be more granular

Long's favorite examples are the kinds of layered rules that are difficult to manage manually. Instead of a flat rule that rejects every loan below a particular FICO, a lender could evaluate that score alongside LTV, DTI, reserves and other factors, then approve or decline based on the combination it actually considers acceptable.


That granularity can also be connected to pricing. Long notes that a lender can create LLPAs for the additional risk it chooses to allow, making risk strategy and yield strategy part of the same configured framework. In that model, expanding the credit box does not mean becoming less disciplined; it means describing the discipline in more detail.


Five elements, one decision

Long's view of the future extends beyond the loan application and credit report. He expects the engine to incorporate more of the core evidence underwriters evaluate, including appraisal information as well as verified income and assets. The goal is not to recreate every verification tool, but to receive the relevant outputs and apply the lender's rules to them.


An appraisal, for example, could populate defined fields that trigger property rules and conditions. Income and asset verification providers could continue doing what they do well while sending final data points into the decision engine. The architecture remains connected rather than monolithic.


Long describes the eventual target as all five elements of underwriting: the lending guidelines themselves, income, assets, property and credit. If those elements are evaluated within the same configurable framework, the lender has a much more complete view of why the loan fits and what must still be satisfied.


The real power is when you combine it with the pricing.

- Derek Long


Combined with pricing and lock management, Long believes that structure can support a wide range of lending products without stripping away their differences. That is the paradox at the center of his vision: the platform can become more universal by refusing to make the credit policy universal.


A twenty-year loop closes

Long's career began with a guideline library and moved toward pricing because lenders needed their rules to become usable technology. More than twenty years ago, he was already trying to take the next step into automated underwriting. LoanPASS has given him another opportunity to finish that progression.


The lesson he carried forward is not that every lender needs the same answer. It is that every lender needs a reliable way to express its own answer. A configurable rules engine can make a complex policy repeatable without making it generic.


If LoanPASS succeeds on Long's terms, the technology will not erase the distinctions among non-agency lenders. It will make those distinctions easier to execute, easier to change and easier to connect to the rest of the lending stack.


We can mimic DU and LP, but we want to go beyond that.

- Derek Long


That is flexibility by design: a decision engine broad enough to serve many kinds of lending, while remaining specific enough to still sound like the lender that built the rules.


ABOUT THE AUTHOR

Laura Kay Sheely is a two-time Amazon best-selling author with 25 years of experience researching and writing about home finance and the technology companies shaping how consumers finance, buy and own homes. Through The 52 Academy Press, she explores the people, ideas and innovations changing the mortgage industry.

Keep up with industry insights and LoanPASS updates.

Subscribe for updates, insights, and other great content.

or

Keep up with industry insights and LoanPASS updates.

Subscribe for updates, insights, and other great content.

or

Keep up with industry insights and LoanPASS updates.

Subscribe for updates, insights, and other great content.

or

Keep up with industry insights and LoanPASS updates.

Subscribe for updates, insights, and other great content.

or