Mortgage AI needs a standard of trust built on experience

Over the past two decades, the mortgage industry has repeatedly confronted the distance between originating a loan and understanding the risk that survives its sale. For AZP Insurance Specialists and its predecessor organizations, that challenge forms part of a mortgage risk and claims history spanning nearly four decades.

When the housing market turned in 2006, and the financial crisis followed, weaknesses in underwriting, documentation and risk assumptions became painfully consequential. The lesson extended beyond falling property values: Confidence in a process was only as strong as the evidence supporting it and the ability of responsible parties to meet their obligations when that confidence was tested.

The following decade brought renewed emphasis on loan quality, verification and accountability. Changes to Fannie Mae and Freddie Mac’s representation-and-warranty framework sought to clarify lenders’ obligations and available relief. For AZP and its predecessors, the practical questions remained familiar: Where did the process fail? Who was responsible? How could the resulting loss be addressed?

The refinancing surge of 2020 and 2021 brought another test: capacity. As rates subsequently rose and refinancing contracted, attention shifted toward cost and productivity. Today, AI offers another substantial advance. The experience accumulated across those cycles makes the questions surrounding its adoption especially important.

The next adoption decision

For lenders, an AI demonstration begins a much larger conversation: How will the mortgage technology perform within their processes, what evidence supports its results and what happens when an exception occurs? Decisions made during mortgage manufacturing can carry consequences long after closing.

Can an output be traced to source information? Who authorizes changes? How is borrower information protected? Which responsibilities remain with the lender? Vendors need credible answers that withstand scrutiny from credit, operations, compliance and technology teams.

A common assessment framework gives both sides a practical language for adoption. Lenders can ask consistent questions; providers can organize evidence around clear expectations. This becomes increasingly valuable as AI capabilities evolve faster than traditional procurement processes.

The timing matters because adoption now reaches into decisions previously supported by extensive manual review. Lenders under pressure to improve productivity need confidence that efficiency gains preserve visibility into how work is performed. Vendors seeking broader deployment need a way to demonstrate readiness beyond a successful pilot. A repeatable assessment can help both move from enthusiasm to an informed decision.

Why AZP’s experience matters

AZP and its predecessor organizations have placed mortgage representation-and-warranty insurance on more than 5 million mortgage loans representing more than $2 trillion in mortgage principal. That experience connects mortgage technology evaluation with underwriting, repurchase exposure and claims.

A claim brings the full process into focus: the information available when a decision was made, the controls that should have applied and the circumstances that produced a loss. That perspective informs the questions worth asking before adopting mortgage technology.

Turning experience into an adoption framework

The AZP Trust Platform’s AI Mortgage Technology Certification framework evaluates AI governance, security, controls, auditability, operational maturity and risk management. It examines the evidence supporting a provider’s readiness for mortgage production, including how exceptions are identified and addressed.

AZP’s ambition is to make that discipline a standard part of adoption: a shared basis for assessing performance, process and security before deployment expands. Certification supports a lender’s diligence and approval process; it does not guarantee outcomes or eliminate risk.

Gateless provides a concrete example

Gateless became the first provider to achieve Platinum, AZP’s highest certification level. Its final assessment scored 97.8, exceeding the 95-point Platinum threshold, based on evidence reviewed through August 28, 2026. The assessment covered Smart Underwrite and VeriClear across 11 categories.

Gateless helps lenders automate underwriting work, so loans move through production faster, with less manual effort and greater consistency, while underwriting teams focus on exceptions and decisions requiring judgment. Its approach combines agentic AI with human-defined rules engines and expert systems, applying mortgage expertise to document interpretation and underwriting workflows. The business objective is dependable, accurate execution that improves productivity while preserving lender control.

What supported the score was the operating discipline behind those capabilities. Gateless earned 99 in operational controls and AI governance, and 98 in information security and performance analytics. Those results reflected documented controls, testing and production evidence, with strengths assessed alongside areas for continued improvement.

The calculation audit framework used random sampling, source-document review and manual recalculation to test accuracy. It distinguished errors from observations, helping teams separate issues requiring correction from other findings. Sampled remediation records showed findings routed to product and engineering, linking quality review with corrective action.

That connection matters to lenders. Identifying an exception is one step; retaining evidence, assigning responsibility and tracking the response make the process usable in daily operations. The assessment recognized this connection between risk identification, operational review, exception handling and remediation.

Production reporting supplied another layer of evidence. Recurring customer reports tracked trends, adoption, automated clearance and operational outcomes. Additional audit, quality-control and performance evidence strengthened the final assessment. The score therefore reflected demonstrated practices, rather than relying solely on product descriptions or proposed controls.

The review also drew on a completed 2025 SOC 2 Type II report. Together with mortgage-specific operational testing, these materials connect technology diligence with the responsibilities of producing and selling loans.

The certification announcement reported that lenders used Smart Underwrite to manufacture and sell more than 230,000 loans without a single claim. That operating history adds context to the control assessment, while remaining distinct from a claim that every loan was defect-free or that future losses cannot occur.

For lenders, the result is a specific body of evidence to examine when evaluating Gateless. For mortgage technology providers, it illustrates what a meaningful certification can reward: mortgage expertise translated into documented processes, measurable performance and traceable corrective action.

Connecting confidence with financial accountability

That certification is complemented by access to AZP’s bespoke mortgage representation-and-warranty insurance program, rooted in decades of coverage placed by AZP and its predecessors and provided today through a Lloyd’s of London policy underwritten by Munich Re, subject to underwriting and policy terms.

For Gateless, the combination strengthens the case for adoption: Lenders can examine independently assessed performance and controls, while qualifying clients and transactions can also benefit from financial protection against covered losses. Certification establishes an evidence base for trust; separately underwritten insurance adds a financial response when a covered error results in loss, subject to policy terms, limits and exclusions.

Adoption must be rooted in trust, and trust must be earned through speed, dependability, accuracy and tangible business results. Gateless illustrates how demonstrated performance, independent certification and available insurance can work together to build lender confidence. For other technology providers, the same opportunity exists: substantiate the product, demonstrate the controls and connect that evidence with financial protection for covered exposures.