Consistency with every touch

Why Tess Anson believes automated underwriting should make non-QM lending faster and more consistent without removing the underwriter from the decisions that still require human judgment.


Tess Anson, VP Guideline Management, LoanPASS, approaches automated underwriting from the seat most technology companies are trying to change: the underwriter's. She spent more than 20 years as a senior underwriter during a mortgage career that has stretched beyond three decades, and she has also worked in processing, closing, shipping, insuring and post-closing.


That experience makes her case for automation notably practical. She knows where a file slows down, where guidelines get interpreted differently and where repeated manual work consumes time that could be spent on the pieces of a loan that actually need judgment.


For Anson, the promise of LoanPASS is not an automated replacement for the underwriter. It is a consistent layer that can apply the lender's defined guidelines the same way at each touch, return an answer quickly and let people manage the exceptions and context that a rules engine cannot own by itself.


The promise of LoanPASS is not an automated replacement for the underwriter. It is a consistent layer that can apply the lender's defined guidelines the same way at each touch.

- Tess Anson


Built for the manual gap

Anson says the vision for LoanPASS's automated underwriting engine has been present from the beginning, with a particular focus on non-QM products. Those loans have traditionally required more manual underwriting, which increases the amount of labor around the file and can lengthen the process.


The problem is not simply that manual underwriting takes time. Different underwriters can have different points of view, and the same guideline can be interpreted differently depending on who touches the file and when.


That is why Anson pairs speed with consistency. A faster answer is useful, but a repeatable answer is what changes the operating model. The engine applies the same configured requirements each time rather than relying on memory or personal interpretation for routine rules.


A faster answer is useful, but a repeatable answer is what changes the operating model.

- Tess Anson


Speed is only half the goal

Anson calls process time the immediate lift. When loan data is entered, the engine can return a result quickly instead of waiting for a person to work through the routine eligibility and guideline checks one by one.


But she repeatedly comes back to reliability. A decision that arrives quickly but cannot be trusted creates more work later. The goal is to make the result both faster and more consistent so every part of the loan process is working from the same underlying policy.


That consistency matters across departments. Anson sees the AUS being used by loan officers, brokers, processors and operations teams before the file reaches the underwriter, giving each group a common view of what the loan requires.


That consistency matters across departments.

- Tess Anson


The same guideline, every time

The most direct value of a rules engine is that the same input can be evaluated against the same guideline every time. Anson contrasts that with a manual environment where a person can miss a requirement or apply a slightly different interpretation.


That does not mean every file becomes simple. It means the simple, repeatable parts are handled consistently, which helps isolate the part that actually deserves expert review.


For an underwriter, that is a meaningful shift in workload. Instead of spending time rebuilding the policy from scratch, the underwriter can begin with a structured result and focus on what is unusual, incomplete or genuinely judgmental about the file.



The workflow becomes a shared sequence rather than a series of individual interpretations: data goes in, the configured guidelines are applied, the result and conditions come back, and a knowledgeable person can review what still needs human attention.


Do not replace the underwriter

Anson is explicit that automation should not be used to remove the underwriter. Each file can still contain facts and circumstances that require context, and lending remains a process where human contact matters.


Her goal is efficiency, not elimination. The AUS should help operations teams manage their time better by handling repeatable work earlier and more consistently.


That creates room for expertise. The underwriter can spend more time resolving the part of the file that is not reducible to a calculation, explaining a condition or evaluating information that falls outside the standard path.


Trust grows through use

For credit managers implementing an AUS, Anson recommends using it broadly enough to learn from it. Run scenarios, add products, integrate more of the workflow and compare the engine's response with the policy the institution expects it to enforce.


Her point is not that the system becomes correct through repetition on its own. The engine remains based on defined rules. The value of use is that teams gain confidence in how those rules are configured, uncover gaps and become more familiar with the kinds of decisions the system can handle consistently.


Adoption therefore becomes part of governance. The lender learns where the engine is strong, where configuration needs to change and where people should remain directly involved.


The lender learns where the engine is strong, where configuration needs to change and where people should remain directly involved.

- Tess Anson


AI is not making the call

Anson is careful to separate automated underwriting from artificial intelligence. The core of the LoanPASS engine is made up of investor guidelines, rules, calculations and overlays - not an AI judgment generated fresh on every loan.


That distinction matters because a lender has to be able to explain why a decision was made, especially in non-QM. A rule can be traced to a requirement. A calculation can be reproduced. An overlay can be identified. The credit process remains understandable to the people responsible for it.


Anson does not want AI replacing that decision without transparency. The organization still needs to know what policy was applied and why the outcome followed from it.


Where Anson sees AI helping is around configuration and analysis. It can organize data, spot patterns, speed document analysis and help users navigate complex guideline material more efficiently.


Those are meaningful tasks because guideline interpretation itself can be labor intensive. AI can help people get to the relevant information faster without becoming the authority that decides whether the borrower is approved.


The dividing line is consistent with Anson's broader philosophy: use technology to reduce the manual work, preserve human visibility into the decision, and keep the rules themselves explicit enough to explain.


Flexibility across products

Anson also emphasizes that the engine is not limited to one product path. A lender can configure the system to evaluate a loan against one product or multiple product options, depending on how the institution wants to position the borrower.


That flexibility matters in non-QM because the same borrower may fit different programs depending on how the lender structures the credit box. A configurable engine can help surface those choices without requiring users to repeat the entire research process for every option.


The future Anson describes is therefore not just faster underwriting of one product. It is a more fluid way to evaluate how a loan fits across the lender's available offerings while maintaining consistent rules underneath each result.


Operations feels the change first

Anson expects operations teams to feel the immediate benefit because they are closest to the repeated touches that surround manual underwriting. Better turnaround time can change how files move, how quickly teams know what is missing and how often a loan has to be revisited simply because a requirement surfaced late.


Loan officers and brokers benefit from earlier direction. Processors benefit from clearer conditions. Underwriters benefit from less routine guideline work. The engine is useful because it gives several people the same starting point rather than optimizing only one desk. That broader workflow view is consistent with Anson's own career across processing, closing, underwriting, shipping, insuring and post-closing.


Expect operations teams to feel the immediate benefit because they are the closest to the repeated touches that surround manual underwriting.

- Tess Anson


Quality follows consistency

Anson's excitement about the AUS is tied to what she expects it to do for loan quality as well as speed. If the same guidelines are applied throughout the process, fewer requirements should depend on which person happens to touch the file.


Consistency does not guarantee a good credit policy; the lender still has to define the right rules. What it can do is make the execution of that policy more reliable and make deviations easier to identify.


That matters in non-QM because complexity amplifies inconsistency. The more specialized the product, the more valuable it becomes to separate the policy itself from the manual effort required to remember and apply it.


Anson sees automation as a way to protect quality by making the repeatable parts repeatable, then preserving human attention for the part that is not.


Faster without losing judgment

Anson has spent enough time underwriting loans to know that speed alone is not a meaningful goal. A fast process that misses the policy simply creates a different kind of problem. The value of automation appears when the file moves faster because the rules were applied correctly and consistently, not because important judgment was skipped.


Her version of automated underwriting is more disciplined. Let the engine apply the lender's defined rules. Let AI help organize and analyze the information around those rules. Let people review the file where context and judgment still matter. The objective is to reduce the number of times people have to repeat work that the system can handle reliably while preserving visibility into how the result was reached.


That division of labor can make non-QM lending feel less dependent on manual memory without pretending that every borrower or every file can be reduced to a machine decision. It also gives managers a clearer way to see where a process problem comes from: the policy, the configuration, the data or the human exception that genuinely needs attention.


The result Anson is aiming for is straightforward: faster loans, more consistent execution and an underwriter who still has time to underwrite.


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