AI Mortgage Underwriting: Use Cases, Tools, and Challenges

Employment records, bank statements, tax returns – it’s a ton of documents! Sifting through it all manually can take up to a couple of months. AI mortgage underwriting can significantly speed up this process through automation. Fewer mistakes and less processing time are not the only benefits. A report by McKinsey & Company suggests that implementing AI automation in the mortgage industry could result in up to 20% cost savings. So, let’s find out how to reach such results.

Traditional Mortgage Underwriting and What’s Wrong with It?

Traditional underwriting is a lengthy, manual process that involves loan officers meticulously reviewing each application, poring over paperwork, verifying information, and making judgments on creditworthiness and risk.

Take the mortgage underwriting process in the US, for example.

1. The Application Flood & Manual Data Entry

It all begins with the borrower submitting a detailed application, FNMA 1003, along with a stack of supporting documents: bank statements, tax returns, and employment records. Let’s skip how terrible it is for the environment and consider how exhausting going through over 500 pages is for mortgage brokers and underwriters. Additionally, each piece of paperwork needs to be carefully reviewed and entered into the system.

And this presents the first challenge: manually sifting through all the documents can take weeks, even months, while manual data entry leaves ample room for human error.

2. Verifying the Verifiers

Once all the documents are entered, the real verification work begins. Underwriters have to contact employers, banks, tax agencies, and more to confirm that the information provided is accurate. But how do you verify the verifiers? Phone calls and letters can only do so much. Weeks more may pass while the credit, employment, and asset details are checked. Systems like The Social Security Administration’s (SSA) Consent-Based SSN Verification Service (CBSV) or credit reporting agencies like Equifax and Experian facilitate the process, but it still requires time and focus. That’s the root of another challenge: figuring out new sophisticated fraud schemes while the potential homebuyer awaits in limbo, wondering if their dreams of owning a home will be approved.

3. Crunching Numbers & Calculating Risks

With verifications (hopefully) complete, the underwriter can begin their financial review. They analyze debt-to-income ratios, credit reports, funding sources, available assets, and more to determine how reliable the borrower may be.

Once all financial factors are considered, including the property assessment, the real challenge begins—integrating them into an overall risk evaluation. The mortgage underwriter must consider how elements like job stability, income likelihood, payment history, and credit utilization interact and potentially offset each other.

Assembling a big picture out of vague little pieces requires systematic analytical skills, assiduity, and fairness of judgments. One miscalculation at this stage could end up costing the lender a lot of time down the road if defaults arise.

The biggest issue with traditional underwriting is not its time consumption but relying solely on the best guess of a human underwriter. Meanwhile, unbiased AI-powered automated mortgage underwriting software can augment human capabilities, conduct precise calculations, run documents through various databases, evaluate the property, predict risks, and provide recommendations for further consideration.

What is AI Mortgage Underwriting?

AI Mortgage Underwriting implies using artificial intelligence and machine learning technologies to facilitate and enhance the processes of evaluating mortgage loan applications, fact checking, document processing, data entry, fraud detection, and decision-making.

For underwriters, AI tools can highlight anomalies, inconsistencies, or discrepancies in an application and generate recommended risk ratings to assist in final decisions. For applicants, this means a potentially faster loan process as their application is automatically reviewed. Simple cases that clearly meet the lender’s criteria may even be instantly approved in some instances.

Of course, the final underwriting judgment – whether to approve, deny, or place conditions on a loan – still requires a human underwriter.

Mortgage AI Applications in Underwriting

Automated Document Processing

AI automates data extraction from loan applications, such as the FNMA 1003 form, through the Intelligent Document Processing technology. For example, AI can parse through the detailed personal and financial information provided by borrowers, extracting key details like income, asset values, and property information. This extracted information can then be filled into the respective fields in your Automated Underwriting System, eliminating the need for manual data entry.

Interface of Contract Summary AI by Ascendix

Moreover, these mortgage underwriting AI solutions can classify application documents based on their structure and format, facilitating easier future reference and more efficient storage.

We at Ascendix Tech have evolved our own AI data extraction module laser focused on real estate and mortgage needs. While not a ready-made solution, the framework can be seamlessly integrated into your current mortgage underwiring workflows and be customized based on your processes and operations. Speak to our AI experts.

Application Paperwork Review

Mortgage underwriting AI tools can automatically review loan applications for completeness and accuracy. For instance, AI can cross-check submitted data against standard requirements and identify any missing or inconsistent information. If some document is missing, the mortgage underwriting AI will alert the loan officer accordingly before submitting the application for further underwriting.

Income and Debt Verification

Mortgage AI can take on more than just basic calculations but streamline the entire verification process. Extracted data from pay stubs, tax returns, and credit reports serves as a basis to verify income, additional debts, and payment histories and calculate debt-to-income (DTI) ratios. Also, as a double check, AI mortgage underwriting systems usually compare provided data with tax returns and other databases.

Risk Assessment

AI-driven risk assessment has reduced mortgage default rates by 27%. How? AI can analyze a wide range of data, including credit history, loan-to-value ratio, debt-to-income ratio, and other factors, to assess loan risk and predict default likelihood.

Also, you can automatically extract meaning from textual data like applicant letters and emails as well as from unstructured data sources like photos and PDFs – you name it. No more reading every word line by line. The automated underwriting software understands the context and flags anything important.

Thanks to predictive analytics, the algorithms can spot trends in massive loan databases, linking specific variables to risk factors. Say they find people with four credit cards are more likely to default. Patterns like this help predict who’ll be a reliable morgagor.

Fraud Detection

Fraud detection is one of the most meticulous and time-consuming processes in mortgage underwriting. That’s why 85% of mortgage lenders use AI for fraud detection and prevention. This adoption has already brought promising outcomes: AI has helped reduce mortgage application fraud by half.

AI underwriting systems perform multifactor checks, including cross-referencing applicant information with public records, credit bureau data, and proprietary compiled data sources to flag any discrepancies in names, addresses, employment details, income statements, and other application facts.

Also, automated mortgage underwriting software analyzes suspicious activities like frequent address or job changes, credit inquiries in a short period, and questionable asset or income statements and efficiently detects these “red flags” that humans may miss from wading through piles of documents.

Applicants Support

AI chatbots and virtual assistants can receive mortgage applications and route them to the company system and free mortgage underwriter automatically, keep up applicants with the application status, and answer questions regarding the process, necessary additional documents, or terms and conditions of the mortgage offer.

Property Valuation

AI mortgage underwriting covers not only looking into potential mortgagors, but also the property they applied for to see whether the asking price is in line with the home’s determined value. Usually, at this stage, the property appraiser is involved, prolonging the mortgage underwriting process even more.

Fortunately, AI has evolved enough to power property valuation tools that analyze historical data, market trends, and property characteristics and return comprehensive results. Have you heard of Zestimate? Software like that would be a helpful addition to the underwriter’s toolkit.

Moreover, predictive analytics also consider probable changes so mortgage terms can be adjusted accordingly. If the house is so old that significant repair will be needed in a few years, the AI mortgage underwriting system will notify the specialist and recommend adjustments to the offer.

Condition Fulfillment

AI-automated condition management eliminates the headache of chasing incomplete files. Mortgage AI sends reminders to borrowers to submit required documents and tracks the responses in real-time. If extra data isn’t received within a reasonable window, notifications are escalated to loan officers for live follow-up.

Final Loan Decision

While human review remains critical, AI does much of the legwork for more informed final decisions. Automated mortgage underwriting software analyzes not just one applicant independently but also identifies trends across past loans to evaluate the borrower’s financial history and consider current economic conditions and historical data. For instance, AI could flag that a certain employment type has higher default rates during economic downturns.

Those are mortgage AI applications in underwriting, but mortgage lending encompasses other processes, too. A good part is that AI can help you facilitate most of them.

Will Underwriters be Replaced by AI?

No, underwriters will not be replaced by AI. The future of underwriting lies in a hybrid approach, often called “human-in-the-loop.” This model leverages AI for its speed and efficiency while retaining human underwriters for complex cases, quality control, exception handling, and customer interaction, where empathy and nuanced judgment are crucial.

For example, mortgage brokers and underwriters counsel mortgagors, explain requirements and find the best property for their needs—a task too complex for AI alone. Advanced virtual assistants can help with answering general questions or simple queries but not addressing complex what-if situations.

Also, automated mortgage underwriting still raises many concerns. If the algorithm is trained on unlabeled biased data (which are most of the historical data), the decision will be biased too. The AI or machine learning model can analyze a loan application, pull credit reports, verify income documentation, and even generate a suggested risk rating. However, approving or denying the loan, setting an interest rate, or placing conditions on approval requires a human underwriter. The AI acts more as a tool to help speed up the process and point out any issues or inconsistencies.

Besides veteran desktop apps like Fannie Mae Desktop Underwriter (DU) and Freddie Mac Loan Prospector (LP), there are a few promising off-the-shelf automated mortgage underwriting solutions that you might want to consider.

AI Mortgage Underwriters

There are many ready-made solutions on the market, but unfortunately, none of them can be called advanced, fully automated mortgage underwriting software capable of replacing human underwriters. Nevertheless, the following solutions are good enough to assist.

FundMore

FundMore has two products for facilitating mortgages: FundMore.ai for decision-making and FundMore.iq for document management. This automated mortgage underwriting software is focused on optimizing the pre-funding by automating manual processes.

Since mortgage underwriting involves handling huge amounts of sensitive data, FundMore.ai is now SOC2-compliant, ensuring top data protection.

Fundingo

Fundingo is an automated mortgage underwriting software designed specifically for non-bank lending businesses. It’s Salesforce-based, so you can find it on AppExchange. Fundingo offers automation of fundamental processes in AI mortgage underwriting:

• Automated Document Management to optimize document lifecycle stages like collection of applicant documents and generating offer letters and closing documents.
• Offer Wizard for appropriate pricing and terms of agreement.
• Application Routing to send the application to the most knowledgeable underwriter suitable for each particular case.
• Automated Data Entry to Salesforce CRM.

Fundingo – Automated Mortgage Underwriting Software

Also, Fundingo easily integrates with borrower verification data sources like LexisNexis, Experian, and DecisionLogic. Sounds impressive, but bear in mind the costs of the Salesforce license and further customizations.

Inscribe.ai

Inscribe.ai is a promising startup that aims to address one of the most tricky parts of AI mortgage underwriting: fraud detection and risk assessment. Inscribe.ai has multiple layers of checking document alterations: x-rays, masking, anomalies, whether the identical document has been submitted before, how recent the data is, and whether the name is featured in any blocklists. Based on this analysis, Inscribe.ai recommends further actions.

Inscribe.ai – Automated Mortgage Underwriting Software

TurnKey Lender Automated Mortgage Underwriting Software

TurnKey Lender has many lending solutions under its umbrella for various spheres, including consumer and commercial lending automation. Of course, TurnKey Lender has software to offer mortgage underwriters – AI-powered loan underwriting software.

Main functionality includes:

• Intelligent Risk Scoring
• Credit History Analysis
• Scoring Reports
• Document Management

All the results are presented with appealing diagrams on the interactive dashboards.

TurnKey Lenderi – Automated Mortgage Underwriting Software

Fintelite

Fintelite is not a fully-fledged automated mortgage underwriting software but a cloud-based tool focused on data extraction, document verification, and bank statement analysis. Mortgage brokers and underwriters can use those capabilities for fraud detection, better decision-making, and personalized offers based on spending habits and investment preferences. Although Fintelite is more of a fintech than a proptech solution, its functionality is great for automated mortgage underwriting.

Ascendix Can Help You with AI Mortgage Underwriting

As a seasoned partner with over 16 years of industry experience, we bring a unique blend of real estate expertise and technical know-how to the table. Prominent real estate firms like JLL and Colliers, along with over 300 other clients worldwide, trust us for their business automation.

Leveraging our deep real estate and lending technology expertise, the Ascendix team has developed an Intelligent Document processing framework for automating mortgage paperwork that can be a stand-alone ready-made solution, a powerful add-on to your existing system, or a solid basis for your custom solution, but cheaper and faster than beginning from scratch.

If you are interested in boosting your operational efficiency with advanced and secure AI solutions, we provide AI development services with a flexible fractional team engagement model.

If you already have a system suitable for your needs, let us enhance it with AI integrations after thorough testing and auditing of your system.

If you hesitate whether you need AI, our experts can provide you with elaborate AI consulting and answer all your questions.

Book a free consultation, and we will discuss how the latest technologies and our expertise can benefit your business.