How TLS Cut Cost Per Loan 40% and Scaled to $2B+ in Monthly Volume with End-to-End Mortgage AI A top ten wholesale lender’s playbook for deploying intelligent mortgage automation—and the production results that prove it works. Every mortgage lender today is talking about AI. Conference stages are full of demos. Vendor emails promise transformation. But how many lenders can point to production results from end-to-end mortgage automation actually running at scale? The Loan Store (TLS) can. TLS, now a top ten wholesale lender, partnered with MOZAIQ to deploy intelligent mortgage automation across its entire loan lifecycle—not a patchwork of disconnected point solutions, but a single, unified AI platform integrated with its LOS. These are not projections. They are live production outcomes from a lender processing over $2 billion in monthly volume. The Problem TLS was founded to be the most efficient, lowest-cost wholesale lender while delivering superior broker service. In a commoditized market where the Mortgage Bankers Association (MBA) has reported negative origination margins industry-wide, the challenges were compounding: The market offers no shortage of AI buzzwords and narrow point solutions—a tool for document indexing here, an underwriting “support” automation there, with a chatbot thrown in—but none deliver connected, end-to-end mortgage AI automation across the loan lifecycle. Bolting together multiple vendors creates integration complexity, data silos, and gaps that manual effort still has to fill. TLS needed a partner who understood both the technology and the business—not a vendor still learning the industry. The Solution: One Platform, One Partner In a market crowded with generic AI vendors still learning the difference between a 1003 and a 1008, TLS chose MOZAIQ—mortgage professionals who became automation pioneers, with a leadership team carrying over 100 years of combined mortgage industry experience alongside deep AI and automation expertise. We didn’t learn mortgage from a whiteboard—we lived it. Then we built the AI to transform it. Over a four-year strategic partnership, TLS and MOZAIQ deployed—and continue to deploy new features—the Loan Assist platform, a unified, end-to-end intelligent automation SaaS platform, integrated with Encompass, and purpose-built to automate the entire loan manufacturing lifecycle: Document Indexing. AE Email Companion. Loan Setup. Appraisal Review. Credit, Asset, Income, Settlement, Insurance, Title, and Tax Analyzers. Underwriting Assist. Closing, Funding, and Post-Closing Reviews. And Loan Delivery. One platform, one integration, one partner—covering the entire loan lifecycle. No patchwork of disconnected tools. No integration gaps. No vendor finger-pointing. Every component shares a common data layer and orchestration engine, so automation compounds across the lifecycle rather than requiring challenging integrations. Critically, Loan Assist was designed with a human-in-the-loop architecture—processors and underwriters retain full control to review, accept, or override every AI recommendation, with complete audit trails. This transforms AI from a compliance risk into a compliance advantage. And, automation modules were deployed incrementally using a “try before you buy” model, proving ROI at every phase. The Results: Production Metrics, Not Projections MOZAIQ doesn’t just talk about AI in mortgage—we get it done. What This Means for Lenders Evaluating AI TLS’s journey reveals five lessons for any lender considering mortgage AI: The Loan Store didn’t just adopt AI—they committed to intelligent mortgage automation as a strategic foundation, and chose a partner that delivered it end-to-end with measurable production results. MOZAIQ didn’t just talk about AI in mortgage. We got it done. Ready to see what MOZAIQ can do for your operation? Schedule a conversation with our team.
The Guide to Adopting Agentic Mortgage AI
The Guide to Adopting Agentic Mortgage AI Today, we’re publishing “The Guide to Adopting Agentic Mortgage AI,” a white paper that distills everything we’ve learned over seven years of automating end-to-end mortgage fulfillment for some of the nation’s largest lenders into a practical framework for mortgage executives who are ready to move beyond the hype. The paper is grounded in MBA industry data, production metrics from a top-ten wholesale lender, and a clear-eyed view of where mortgage AI is headed — including the emergence of self-configuring autonomous mortgage AI agents that will fundamentally reshape how loans are manufactured. Below is the executive summary. To read the full white paper, download it here. The State of the Industry The mortgage industry is at a crossroads. After origination volumes collapsed nearly 60% from their 2021 peak and lenders endured six consecutive quarters of production losses — bottoming out at negative $2,812 per loan in Q4 2022 — the market has somewhat stabilized. By Q3 2025, 85% of institutions were reporting positive pre-tax net income, and net production income reached $1,201 per loan, according to the MBA’s Quarterly Performance Report (Q3 2025). But this fragile recovery masks a structural problem that has not been solved: the average cost to originate a mortgage remains over $11,000 per loan (Source: MBA; for Residential, Wholesale, and Correspondent lenders combined), with personnel accounting for more than 60% of that cost. The industry is barely profitable, but only as long as volumes hold. Why Agentic Mortgage AI Is No Longer Optional This white paper makes the case that intelligent mortgage automation — specifically, agentic mortgage AI purpose-built for mortgage fulfillment — is no longer optional. We believe that adoption of agentic mortgage AI will be the foundational operating capability that separates lenders who will thrive from those who will continue to struggle to survive irrespective of whether loan volumes return or not. The paper draws on MOZAIQ’s production experience automating end-to-end mortgage fulfillment for some of the nation’s largest lenders over the course of the last seven years. It examines why traditional approaches to mortgage automation have failed — or why they haven’t achieved the expected returns, what “agentic mortgage AI” means in the context of mortgage operations, and how lenders should think about adopting agentic mortgage AI in a way that delivers measurable ROI at every phase of adoption. Production Results, Not Projections The proof is in the production metrics: A top-ten wholesale lender incrementally deployed MOZAIQ’s Loan Assist platform — our end-to-end agentic mortgage AI product suite — over the course of the last four years and scaled from $10 million to $2 billion in monthly originations in under two years — while reducing cost per loan by 40%, doubling productivity, and improving loan turn times by more than 50%. These are not pilot results or projections. They are live production outcomes. Whether you are a CEO evaluating strategic technology investments, a COO seeking to transform mortgage operations, an underwriting executive looking for enhanced efficiencies, or a technology leader assessing agentic mortgage AI partners, this paper provides the operational framework and data you need to make the right decisions to enhance your competitive advantage through the adoption of agentic mortgage AI. Ready to see what MOZAIQ can do for your operation? Schedule a conversation with our team.
Agentic Mortgage AI is Here
The Age of Agentic Mortgage AI has Arrived For two decades, the mortgage industry has chased the promise of automation and fallen short. Wave 1 outsourced loan processing to BPO providers who lacked mortgage expertise, resulting in a cycle of outsourcing and in-sourcing that negated the savings. Wave 2 brought specialized mortgage outsourcers with deeper domain knowledge, but they still relied on manual processes that couldn’t scale with volatile volumes—they still had to hire and train resources when volumes increased, and fire them when volumes declined. Wave 3 introduced AI, RPA, and intelligent document processing, but these technologies were deployed as isolated point solutions that created integration gaps, broke when underlying systems updated, and struggled with the complexity of 300+ mortgage document types. Each wave moved the needle incrementally, but none solved the fundamental problem: end-to-end mortgage fulfillment remained people-dependent, fragmented, and difficult to scale. Wave 4: The Autonomous Mortgage Agent A fourth wave is now taking shape, one that will fundamentally redefine what is possible in mortgage automation. In Wave 4, intelligent mortgage agents move beyond executing pre-configured rules and workflows. These agents are capable of self-configuration, automatically incorporating the latest Fannie Mae and Freddie Mac guideline updates, adapting to changes in a lender’s product specifications, and adjusting their decisioning logic in real time without requiring manual reprogramming by technology resources. When an agency updates a guideline or a lender modifies its product portfolio or criteria, the agents ingest, interpret, and operationalize the change autonomously. The operating model looks fundamentally different. Swarms of specialized agents—each trained for a specific function such as income analysis, asset verification, appraisal review, or compliance validation—work a loan file collaboratively across the entire fulfillment lifecycle. The majority of conditions are identified, resolved, and cleared by the agents themselves. Exceptions that require human judgment are surfaced to experienced underwriters and processors, but these exceptions represent a fraction of the current workload. The human role shifts from performing routine work to governing the system and exercising judgment on genuinely complex scenarios. This progression begins with conventional conforming loans, where guidelines are most standardized and the volume of training data is largest. As the agents demonstrate accuracy and reliability on conventional products, they extend to more complex loan types, such as government loans, non-QM, jumbo, and specialty products, progressively expanding the frontier of what can be automated. The Foundation That Makes It Possible Today, specialized agents for foundational functions like document indexing, loan setup, and rules-based validation are already a reality. The data to train these agents exists, and the current generation of generic LLMs—Claude, ChatGPT, Gemini, and others—are capable enough to power them effectively. But the full Wave 4 vision—autonomous end-to-end mortgage processing—will ultimately require a purpose-built, mortgage-trained large language model. Today’s generic LLMs lack the domain specificity to handle the nuanced regulatory, compliance, and product knowledge required for mortgage decisioning. Ideally, an LLM pre-trained on the full corpus of mortgage data—agency guidelines, investor overlays, historical loan files, underwriting decisions, audit findings, and regulatory frameworks—and continuously fine-tuned as regulations and products evolve would be created. This is what will ultimately produce a true underwriting decisioning engine: an AI that can process a mortgage loan end-to-end, with no errors, with minimal human intervention, and with full auditability that satisfies regulatory scrutiny. The industry is not there yet. It will happen, but the economic models to train a mortgage-specific LLM have yet to be proven—is the investment in training a custom LLM worth the return? Because it’s also possible that the generic LLMs will advance rapidly enough to close the domain gap on their own. No one knows for certain. But the trajectory is clear: the lenders and technology partners who are building the foundation today, through end-to-end agentic mortgage AI, production-scale data generation, and human-in-the-loop feedback loops, are the ones who will get there first, regardless of which path the LLM landscape takes. Why MOZAIQ MOZAIQ is already there. This is not a roadmap for MOZAIQ: this is who we are now. MOZAIQ’s Agentic Mortgage AI platform is purpose-built to serve as the foundation for Wave 4. Every loan processed through the Loan Assist platform generates structured, validated, audit-grade data across every stage of the fulfillment lifecycle, from document indexing through investor loan delivery. Every human-in-the-loop decision refines the intelligence of the system. Every new customer, every new loan type, every new investor overlay expands the body of mortgage knowledge that the platform operates on. This is the data foundation that a mortgage-specific LLM would eventually require. And MOZAIQ is generating it today, at production scale, across the full loan lifecycle. No other platform in the market has this combination. Point solution vendors have data from one stage of the lifecycle, not all of them. Generic AI vendors have technology without mortgage domain depth. BPO providers have process knowledge but no integrated automation platform generating the data at scale. MOZAIQ is the only platform that combines deep mortgage domain expertise, end-to-end production automation, and the continuously growing data foundation that Wave 4 requires. The path to agentic mortgage AI runs directly through Wave 3, and MOZAIQ is further along that path than anyone else in the industry. Ready to see what MOZAIQ can do for your operation? Schedule a conversation with our team.
AI and the Future of Work
Session Summary: AI and the Future of Work From the MBA Annual Conference held in Las Vegas, NV, in October 2025 Speaker: Geoff Kramer, Financial Services Engineering Leader, Google (Head of Customer Engineering, Enterprise Financial Services North) Geoff Kramer’s session provided an in-depth look at how AI—particularly generative AI—is reshaping workflows across financial services, with a focus on the mortgage industry. Since joining Google in 2020, Kramer noted that 90% of his initial work centered on document processing for lenders and SaaS providers. What began as routine automation for FHA, Fannie Mae, and Freddie Mac documentation has evolved rapidly: in just one year, the mortgage industry has shifted from skepticism about chatbots to deploying cutting-edge GenAI solutions. Key Trends and Data Points Major Use Cases in Mortgage Automation 1. Underwriting Productivity Kramer traced the evolution from OCR to pre-trained parsers, noting that these approaches reached diminishing returns. Even with 99% accuracy per data point, the overall page accuracy fell to ~90%, making the “long tail” of document diversity a persistent challenge. Today, teams are leveraging LLMs and Gemini-based custom processors to improve accuracy and adaptability. To mitigate hallucinations, Google applies machine learning layers to verify extracted data against trusted sources—creating an auditable, transparent process. ROI Impact: Document extraction remains the single largest and fastest-returning GenAI use case, delivering 3–6x productivity gains. 2. Regulatory Compliance Generative AI is being deployed for automated policy comparisons—analyzing old vs. new regulations, summarizing amendments, and assessing loan-level impacts. Loan officers also use these tools to compare loan products and identify eligibility shifts in real time. 3. Loan Review and Approval Kramer highlighted how GenAI streamlines pre-approval by identifying missing documents, dramatically reducing underwriter cycle time and improving speed-to-close. At the end of the cycle, it helps teams adjust quickly to regulatory changes that occur mid-origination.He personally uses NotebookLM to generate audio overviews and summaries from uploaded materials—an example of GenAI’s assistive potential. 4. Lead Generation & Servicing Support Conversational AI is now in production across multiple financial institutions. Kramer outlined four key use cases: Kramer demonstrated a Gemini Live 2.5 Flash Preview (Native Audio, Sept 2025)—showing a real-time AI agent conducting a natural, audio conversation as a customer service representative. The system generates its own prompts and operates within pre-set guardrails. Key Takeaways Kramer closed by emphasizing that LLMs are inherently non-deterministic, and the key to enterprise adoption lies in setting the right system instructions, verification layers, and human oversight—turning GenAI from an experimental tool into a trusted, productivity-driving partner. Source: MBA Annual Convention Session – AI and the Future of Work
MBA2024 Session Summary Part I
AI, AVMs, and Model Management from a Regulatory Perspective This session started off on what I thought was the proverbial wrong foot: with disclaimers. Everyone on the panel was a lawyer (nothing wrong with that), but to introduce oneself with phrases such as “We’re not AI experts” and “The opinions expressed today are my personal opinions and do not represent those of my employer”, well . . . that wasn’t the way to grab the audience’s attention. But, they recovered with interesting and on-topic themes on how regulators are embracing AI, and provided sound advice for lenders that are adopting or looking to adopt AI-based solutions. What is the FHFA doing with AI? The Federal Housing Finance Agency (FHFA) issued an Advisory Bulletin in 2022 to guide the use of Artificial Intelligence (AI) and Machine Learning (ML) in lending, especially for regulated entities, though it can serve as a framework for the broader mortgage industry. This bulletin encourages a flexible, risk-based approach to AI implementation, emphasizing principles of transparency, accountability, and fairness. It introduces a common taxonomy and recognizes the anticipated growth of AI in lending, offering support for compliance and risk management. The FHFA also emphasized that “responsible AI” is crucial for the success of AI applications in this sector. I actually wrote about this in a blog post last year under the heading of Responsible AI as a Critical Success Factor. What are the practical challenges lenders face when implementing an AI-based solution? Lenders face various challenges in implementing AI solutions, including ensuring data quality and availability, as poor data leads to inaccurate results—the model is only as good as the data it was trained with. And, they must prioritize data privacy and security by establishing controls to keep the data used for training clean and secure. Addressing algorithmic bias and regularly monitoring the models and its outputs are critical in order to demonstrate compliance with fairness laws. Additionally, lenders must consider how AI integrates with existing systems, and manage the potential increase in costs and resource demands—not everyone has in-house AI expertise. Finally, model transparency and explainability are essential for understanding how AI reaches its conclusions, enabling better regulatory compliance. What AI issues are you currently working on? The FHFA is currently focusing on establishing its AI framework. The first step was to appoint a Chief AI Officer and aligning with the White House’s executive order to understand how to use AI for public benefit. FHFA is training its staff to: understand AI’s potential, what to look for in reviewing vendors’ AI practices, and how to establish a clear chain of accountability between vendors and clients—it’s crucial that the vendors that lenders engage are in full compliance with the regulations. Finally, the agency is committed to promoting responsible innovation, ensuring AI applications align with regulatory requirements (see prior point on resonsible AI). From a legal perspective, clients are grappling with deploying AI while remaining compliant with regulations published decades ago. Key challenges include determining necessary disclosures, identifying and mitigating biases, and ensuring model transparency and explainability. There is also concern with the implications of AI outputs—for example, combining data from multiple sources may meet the definition of a consumer report. What are the regulatory requirements around that? Additionally, there are questions around potential legal liabilities—if I get this wrong, what happens? How can AI can be used for good? AI offers promising opportunities to improve inclusivity in lending. The FHFA sees AI as a tool to enhance credit decisions by incorporating alternative data, such as cash flow from bank statements, in order to expand the reach to underserved communities (for example, using rent payment history to augment a borrower’s credti report). AI can also play a role in detecting biases, and assisting borrowers with limited English proficiency through translation. In compliance, underwriting, and fraud detection functions, AI can handle routine tasks, with human oversight reserved for exceptions—as MOZAIQ is doing today with our end-to-end automation platform, and could be used to identify potential ADA (American with Disabilities Act) non-compliance by analyzing images of the property. Finally, AI can improve valuation accuracy of automated valuation models, and support regulatory reporting and risk management, exemplified in areas like helping to report on Home Mortgage Disclosure Act (HMDA) data. The panelists: Full disclosure: This blog post was created by running my copious notes through ChatGPT and then editing the output. Why not use AI to write about AI? It’s the future. And I’ve embraced it.
MBA2024 Session Summary Part 2
The Promise and Possibility of AI Before I provide a synopsis of the panel discussion titled “The Promise and Possibility of AI”, I’d like to share some observations from this year’s MBA Annual 2024 conference. This year’s conference was well attended, but as with the past two years (I started attending the MBA in 2022 in Nashville), only approximately 35% of attendees were mortgage lenders or banks, the rest were all vendors (Full Disclosure: I was one of them). And it seemed that every other vendor booth had “AI” or “Automation” in their marketing materials, making it clear that the mortgage automation market is crowded with players, whether they deliver anything of substance or not. Regardless, it is increasingly difficult to stand out. However, from speaking with some of the larger service providers, lenders did come by and ask about automation and digital transformation, and when asked who else they met with, the answer was that these particular lenders gravitated to either the larger, better-known technology players in the market, or they met with smaller vendors that had a more complete product portfolio, and didn’t focus on providing just a single point solution e.g., a Pre-Qual solution, but nothing downstream in the process. The focus on either an end-to-end provider, or a larger scale provider was to alleviate the concern that lenders would end up having to manage multiple vendors, as opposed to only one when it came to AI and Automation solutions. Perfect for MOZAIQ, as we offer one of the few end-to-end, next-gen automation platforms specifically built for the backend mortgage fulfillment process that delivers a tangible ROI. But I digress. One of the many and more interesting AI-focused panels was hosted by Laura Escobar, 2025 MBA Chair and President of Lennar Mortgage. The panel was comprised of Olivia Peterson from AWS and David Tepoorten from NVIDIA (bio links are at the end of the blog post). How Not To Deploy AI I almost walked out of the session, though, when an “AI-bot” was used to introduce the panel. . . a case study in how not to leverage AI. A robotic, lifeless, monotone voice reading from a script. But the session quickly recovered and. . . wait, there’s a better way to do this. I’ll share the edited summary that ChatGPT created from my detailed notes and save myself some time. Here it is: How To Deploy AI The panel discussion highlighted the distinctions between traditional AI and generative AI (Gen-AI), emphasizing that AI is not a new concept, with companies like Amazon leveraging machine learning since 2010. As we know, traditional AI focuses on data-driven decision-making, while Gen-AI enhances learning by integrating past recommendations with new training data. The collaboration between AWS and NVIDIA since 2010 has been pivotal in advancing AI technologies, particularly in weather prediction through innovations like NVIDIA’s digital twin, Earth 2—the video demonstration was fascinating. This simulation model allows scientists to forecast weather patterns with improved accuracy, which has significant implications for industries like insurance that struggle with climate-related risks. Can you imagine the benefits if these models can accurately predict the weather beyond the current ten-day cycle? Sign me up! (I added the last two sentences). Likewise, AWS utilizes data from sources like NOAA to integrate climate risk into financial models, further enhancing decision-making for creditors and insurers. In the mortgage sector, the discussion focused on improving efficiencies through intelligent document processing (IDP), with companies like PennyMac using AWS to streamline document indexing and data ingestion, extraction, and validation. This use case wasn’t that revelatory, as MOZAIQ and a plethora of other companies already provide this “table stakes” functionality, the foundation required for downstream mortgage process automation e.g., Pre-Underwriting, Appraisal Review, Post Close Audit, CD Prep, and Loan Delivery to name a few. The importance of quality data in training AI models was underscored, alongside the necessity of establishing privacy protections and guardrails prior to model deployment. You can read my blog on Responsible AI here. The panel discussed the potential of digital agents in call centers to enhance employee training and improve customer service. These “buddy agents” could help new hires navigate customer queries more effectively, reducing handle times and minimizing repeat calls by providing on-the-job training and summarizing interactions. Additionally, digital agents can today summarize agent-customer conversations (saving time and money for the call center operator and enhancing customer service), and in the future could interpret customer sentiment more accurately than novice employees, augmenting overall customer engagement. The conversation highlighted the importance of maintaining a balance between AI assistance and human oversight, emphasizing that users must ultimately act on the recommendations provided by these technologies, and not the AI engine. Lastly, the panel addressed the trustworthiness of AI technologies from a regulatory perspective, noting that regulators are increasingly adopting these technologies themselves. Janet Yellen, the US Treasury Secretary (and a former Professor of mine at the HaaS School of Business at Cal Berkeley—Go Bears) emphasized the importance of understanding risks and implementing AI responsibly, warning that failure to adopt AI would put an institution at risk. And that’s how to deploy AI. The panelists: Full disclosure: This blog post was created by running my copious notes through ChatGPT and then editing the output. Why not use AI to write about AI? It’s the future. And I’ve embraced it.
MOZAIQ SOC 2 Type II Accreditation
MOZAIQ SOC 2 Type II Accreditation and its Importance for the Service Provider MOZAIQ is excited to announce that it has successfully completed a System and Organization Controls (SOC) 2 Type II audit, performed by Sensiba LLP (Sensiba). MOZAIQ’s SOC 2 Type II report did not have any noted exceptions and was therefore issued with a “clean” audit opinion from Sensiba. MOZAIQ’s SOC 2 Type II Accreditation report is an internal controls report that describes a service organization’s systems, whether the design of specified controls meets the relevant trust services categories, and whether the controls were operating effectively spanning an agreed upon review period. The standard is based on the Trust Services Criteria as defined by the American Institute of Certified Public Accountants (AICPA): security, availability, processing integrity, confidentiality, privacy, and it assures clients and stakeholders that their service provider has implemented effective controls over the client’s data. A SOC 2 Type II report ensures: The most important aspect of SOC2 Type II accreditation is that clients trust MOZAIQ with their data, know that MOZAIQ is in compliance with regulations, and understand that MOZAIQ is highly vigilant when it comes to the overall security and risk management of systems and processes. To find out more or request a copy of MOZAIQ’s Soc 2 Type II report, please contact us now.
AI Takeaways from the 2025 MBA Annual Convention
AI Takeaways from the 2025 Annual MBA Conference At this year’s annual MBA conference in Las Vegas there were several burning topics that attendees debated in the hallways, in the panel discussions, and at the bar: housing affordability differences across demographics; the benefits of the trigger lead legislation when it goes into effect; the rise of state-level compliance as the federal government takes a hands-off approach; if rates would go low enough to have an impact on loan volumes, and when; whether tariffs are impacting the housing market; and when regulators will allow the reform of the traditional tri-merge credit-report requirement. But the topic that garnered the most airtime was AI. It is no longer a buzzword. It’s real, it’s live, and it’s creating positive economic impacts for the early adopters. We are already seeing its near-term impact with tangible cost and cycle-time reductions for multiple sales and fulfillment functions, two of the highest cost categories in the loan origination lifecycle. And everyone was optimistic about it’s long-term impact, where the vision of creating a fully automated, continuously auditable mortgage ecosystem where efficiency, accuracy, and affordability built into every loan will become a reality. Here are the top 5 AI takeaways from this year’s MBA Annual Conference. 1. AI has crossed from experimentation to execution What was “chatbot talk” last year has become real deployments as lenders deploy AI in live production workflows including indexing & data extraction (document processing), pre-underwriting, quality control, and compliance. Some statistics courtesy of Google: → AI is now a line-item in budgets, and no longer a lab project. 2. AI has already transformed the mortgage fulfillment process Immediate wins are emerging in foundational areas like document classification (Indexing) and data extraction, and extending into the intelligent mortgage automation solutions for account executive (AE) and loan officer (LO) support, appraisal reviews, pre-underwriting support (loan pre-approval), underwriting assistance, and post-closing. Today, teams are leveraging LLMs and custom processors to improve accuracy and adaptability of the models. For example, to mitigate hallucinations, machine learning layers are applied to verify data extracted by the LLM models against trusted sources, creating an auditable, transparent process. These tasks are yielding 30–50% cost reductions and 2–3X throughput improvements when automated with Integrated-AI platforms. Human-in-the-loop validation remains a requirements for sensitive processes, like underwriting. Regardless, the productivity gains are now proven. Generative AI is being deployed for automated policy comparisons—analyzing old vs. new regulations, summarizing amendments, and assessing loan-level impacts. Loan officers also use these tools to compare loan products and identify eligibility shifts in real time. Lenders have also successfully deployed conversational “loan assistants” that let loan officers, account executives, operators, and underwriters talk to the loan file via an intelligent chatbot, enabling the asking of context-aware questions regarding the status of the loan, if there are any open conditions, or helping to rapidly escalate and resolve an issue. Loan Officers are even using conversational AI for lead generation, enhancing their productivity and filling their funnel more quickly. → The first real drop in cost per loan is coming from leveraging AI for data extraction and AE and LO support, with regulatory compliance emerging as a high-impact secondary area. 3. Long term, AI will redefine the loan manufacturing model Executives envision a fully connected “digital loan factory”, where data flows seamlessly from borrower through the lender and on to the investor. Within 3–5 years, agentic AI systems deployed by lenders will handle the bulk of file review, condition clearing, and exception management, with humans focusing on exceptions, as they build and maintain important borrower and LO relationships. For example, where deployed effectively, AI has cut initial-underwrite times from 48 hours to under 1 hour, automated 400-page document reviews in minutes, and unburdened underwriters from chasing missing documents. → The most efficient mortgage origination processes (the production line) will leverage AI end-to-end. 4. The human operator role is evolving, not disappearing All of the MBA panels repeatedly emphasized that AI will always require human oversight. But, the level of mundane tasks for underwriters, processors, and QC teams will significantly decrease. The responsibilities will evolve from performing boring tasks like data entry to more impactful exception handling and risk analysis, enabling tighter customer engagement. And remember, LLM models will always hallucinate. One can leverage technology to check for these hallucinations, like machine learning models as described above, or let humans handle the exceptions and validate or clear the edge cases that inevitably arise in mortgage. That’s why AI will make people better communicators and advisors by offloading administrative work. Trust and empathy remain essential, because mortgages are all about relationships, which is why human expertise will continue to define the borrower experience Let the machines handle precision, and humans handle trust. → AI will augment, not eliminate, human expertise 5. Speed-to-AI will separate the winners from the laggards Industry leaders like Rocket, ICE, and Newrez agree: success will hinge on how fast organizations deploy production-grade AI and scale it in a responsible, transparent, and auditable manner. And AI maturity will define the next generation of mortgage leaders. Firms that embrace change and leverage scale will thrive. Standing still is not an option. Continued consolidation will yield larger, more efficient lenders. The digital mortgage experience will be standard, not a differentiator, because the next generation of homebuyers is born digital; therefore lenders must deliver a streamlined, tech-native experience, or risk irrelevance. The customer will expect this. → The race is now about the velocity of AI execution with tangible ROI metrics. If you’re ready to achieve real business benefits with a winning automation strategy, follow the lead of Mortgage Lenders who choose MOZAIQ. Contact us today and discover how our Integrated-AI, End-to-End, Intelligent Mortgage Automation solutions can help you win. The sources for this blog post included these sessions:
Pure-Play BPO is Dead
Implications for Pure-Play BPO Players in the Age of AI The Capgemini–WNS merger highlights the existential challenge facing traditional pure-play BPO providers. Enterprises are accelerating the shift away from labor-intensive outsourcing toward AI-driven “Services-as-Software” models powered by Generative AI and Agentic AI. According to HFS Research in their April 2025 analysis, six out of ten enterprises expect to replace professional services with AI solutions within five years—a clear signal that the traditional “butts-in-seats” BPO model is becoming obsolete. The End of Labor-Intensive BPO Models For BPO specialists such as Genpact, EXL, and others, this market evolution creates a strategic dilemma. Unlike integrated firms such as Capgemini, Accenture, or the Big 4, most pure-play providers lack the deep technology platforms, AI expertise, and consulting depth required to deliver enterprise-wide transformation. As a result, they risk being confined to smaller transactional outsourcing contracts, while larger, high-value deals shift to providers that can deliver end-to-end AI-powered business transformation. Enterprises are no longer satisfied with incremental labor arbitrage—they want outcome-driven, technology-first solutions. How Generative AI and Agentic AI Are Reshaping BPO Generative AI and agentic AI are fundamentally transforming how enterprises—and in particular mortgage lenders—approach outsourcing. These technologies are automating complex workflows, enabling predictive decision-making, and improving compliance and risk management. By embedding AI directly into loan setup, underwriting, closing, post closing, loan delivery, and customer operations, mortgage lenders achieve faster cycle times, greater accuracy, and significant cost reductions—outcomes that no labor-heavy BPO model can match. Why Pure-Play BPOs Must Reinvent Themselves As renewal cycles approach, clients will demand that BPO providers: This means survival for pure-play BPOs depends on reinvention: building AI partnerships, investing in proprietary GenAI and agentic AI capabilities, and embedding consulting-level transformation skills. Without this pivot, traditional BPOs risk becoming “table scrap” providers, relegated to low-value contracts while high-margin enterprise deals migrate to AI-centric competitors. Partnering with MOZAIQ: The Fast Path to AI-First BPO Transformation This is why partnering with MOZAIQ makes strategic sense for BPO providers. MOZAIQ already has all the components necessary to deploy AI-powered services in partnership with BPOs: An end-to-end AI-powered mortgage automation platform already deployed with three of the top ten US mortgage lenders, experience in merging AI-centric solutions with BPO human-in-the-loop exception handling, pricing models aligned with business outcomes, and vertical—mortgage—domain expertise. By embedding Integrated-AI mortgage automation directly into their service offerings, BPOs can deliver immediate cost savings, faster cycle times, and measurable ROI—without the need to build AI-powered solutions from scratch. With MOZAIQ, pure-play BPOs can remain competitive, protect client relationships, and evolve into AI-first transformation partners that enterprises demand. The Bottom Line Enterprises are no longer asking if their BPO vendors can deliver AI—they are demanding it. For pure-play BPOs, the choice is stark: adapt and evolve into AI-powered transformation partners or risk steady decline in relevance and valuation. If you’re ready to achieve real business benefits with a winning automation strategy, Contact us today and discover how our Integrated-AI, End-to-End, Intelligent Mortgage Automation solutions can help you win. Table 1: Pure-Play BPOs Today vs. Enterprise Demands Tomorrow Dimension Pure-Play BPOs Today Enterprise Demands Tomorrow (AI-driven) Delivery Model Labor-intensive, FTE-based “butts-in-seats” contracts AI-driven “Services-as-Software” that reduce reliance on human labor Value Proposition Cost arbitrage via offshore resources Outcome-based efficiency gains, quality improvements, and risk reduction through GenAI and agentic AI Technology Depth Limited proprietary AI/tech capabilities; rely on partnerships Integrated AI/ML/GenAI platforms, proprietary assets, and deep consulting-led transformation skills Contracting Models Traditional multi-year, FTE-linked deals Flexible, modular, consumption-based pricing tied to business outcomes Innovation Pace Incremental process optimization, manual improvements Continuous automation and reinvention, leveraging agentic AI for self-learning and adaptive workflows Enterprise Perception Tactical vendor, focused on process execution Strategic partner, delivering end-to-end digital and AI-enabled transformation Source: HFS Research Note: This blog builds upon insights originally published by HFS Research and reinterprets them through the lens of financial services and mortgage automation.
Housingwire AI Summit Insights
Straight from the Conference If you are going to win in the mortgage industry, you don’t have a choice: you must adopt AI-powered solutions today. Last week I attended the HousingWire AI Summit held at the George W. Bush Presidential Center in Dallas, Texas. It was an informative day, with the key takeaway being that artificial intelligence (AI) is no longer a future concept, but a present necessity that must be embraced by all—lenders, brokers, and technology vendors—in order to compete and win. We all know why. The mortgage industry is in the midst of a digital transformation, with AI in mortgage lending quickly moving from concept to competitive necessity. Lenders and brokers face constant pressure from margin compression, regulatory complexity, and shifting borrower expectations, all while striving to close loans faster and at lower cost. Intelligent mortgage automation and AI-powered mortgage solutions are emerging as powerful levers to streamline operations, enhance compliance, and deliver more personalized borrower experiences. At the same time, technology vendors are accelerating innovation to equip the industry with tools that integrate seamlessly into existing workflows. What Lenders and Vendors Must Do In this post we summarize the top five reasons why lenders and brokers should adopt AI-powered services, and the top five priorities for technology providers building the next generation of AI-powered digital mortgage solutions. Why Mortgage Lenders Should Adopt AI-Powered Services What AI Offerings Technology Vendors Must Deliver On A Summary of What Was Learned Artificial intelligence is redefining the future of the mortgage ecosystem. For lenders and brokers, adopting AI-powered services means reducing loan cycle times, cutting operational costs, improving compliance accuracy, and strengthening the customer relationship by delivering a digital, borrower-first experience. For technology vendors, the mandate is to focus on AI solutions for mortgage automation that provide measurable ROI: targeted AI agents, seamless LOS/POS/CRM integration, built-in compliance guardrails, rapid deployment, and predictive analytics dashboards. Together, these advances in digital mortgage transformation will create a more efficient, resilient, and competitive marketplace—empowering lenders, brokers, and their customers to thrive in an AI-driven era. If you’re ready to achieve real business benefits with a winning automation strategy, follow the lead of Mortgage Lenders who choose MOZAIQ. Contact us today and discover how our Integrated-AI, End-to-End, Intelligent Mortgage Automation solutions can help you win.
