Voyantt launched LoanCite with the purpose of facilitating the process of assessing the income of borrowers by non-conventional lenders as well as recording the necessary evidence, calculations, and decisions which contribute to successful underwriting.
Imagine a system that allows users to view a large number of bank statements and eliminate any cumbersome task of manually inserting each transaction into an Excel file.
That is exactly what LoanCite is set to achieve.
Non-standard lending usually refers to the borrowers whose income cannot be proven using traditional means such as tax returns or payslips, which include but are not limited to self-employed or freelance individuals and business owners whose income can be confirmed with some alternative documents such as bank statements, letters from accountants or any income document issued by leasing companies.
The Hurdle
Non-standard mortgage income analysis is seldom a simple data-entry exercise.
An underwriter may have to review 12 or 24 months of personal or business bank statements, determine qualifying deposits, eliminate transfers and transactions that are not income, apply an appropriate expense factor, reconcile supporting documentation and arrive at a defensible monthly income figure.
When this work is done manually, there are several problems:
- Relevant figures need to be copied from different documents into worksheets.
- Identify transfers, duplicate deposits, refunds, loans and other non-income items.
- Business and personal cash flow should be kept separate.
- Calculations will vary by the lender’s credit policy or loan programme.
- Exceptions, if any, should be identified and reviewed by the appropriate person.
- Quality-control teams need to piece together how the underwriter arrived at the final figure.
- Supporting evidence can be spread across spreadsheets, notes, emails and document folders.
The problem is more than a question of time.
A calculated income figure that has no clear link to its source can lead to operational, compliance, audit and repurchase risk. Even if the number is correct, reviewers need to understand what documents were used, what transactions were included or excluded, what calculation methodology was used and how exceptions affected the decision.
Traditional automation tools may extract numbers or may give recommendations, but extraction alone is not sufficient. Non-standard mortgage lenders need an underwriting process that gives them the ability to scrutinise every material figure and to substantiate every decision.
Our Solution
Voyantt’s AI underwriting platform, LoanCite, is designed to complement existing workflows, rather than replace them, for non-QM and bank statement mortgage lenders. It adds a layer of defensibility, linking together borrower documents, extracted data, income calculations, policy checks, exceptions and human decisions into a single reviewable record.
LoanCite’s Smart Document Review System is capable of working with a variety of income documentation such as bank statements, profit and loss statements and CPA letters. It pulls relevant information from loan files, reducing the need for underwriters to review multiple files manually.
Data extraction is based on provenance, so that each material figure can be traced back to its source, allowing underwriters to refer to the exact document and transaction from which it was derived, and to use confidence indicators for further investigation. This thoroughness answers an important underwriting question: where did each number come from?
LoanCite includes systematic income calculations, reviewability, as well as recurring deposits, transfers, non-income transactions, and unusual deposits. The income worksheet includes conclusions with the methods used, and any adjustments, so it is clear how the numbers are calculated.
Configurable rules help a lender match their credit policies and allow for verification against certain conditions, providing flexibility in underwriting. During the review, exceptions are classified so teams can identify whether something could be a roadblock or whether it needs to be looked at a little more closely.
The platform also supports a human-in-the-loop approach, allowing underwriters to validate AI findings with their own assessments, explanations and modifications where necessary. This combination of AI efficiency and human oversight ensures accountability in mortgage underwriting.
The Outcome
LoanCite automates the manual document review, revolutionising how underwriters handle and evaluate loan applications by streamlining repetitious and time-consuming work. LoanCite allows underwriters to automate bank statement examination, financial record verification and transaction data entry, which will be faster and more accurate than doing it manually.
- The main advantages are:
- Less time wasted on checking bank statements and recording transactions.
- Enhanced time for underwriters to analyze important discoveries and credit risks.
- Prompt credit decision-making.
The platform offers straight paths from final income values to source documents, leaving credit appraisers, reviewers, and auditors with an obvious proof of calculations and increasing transparency. LoanCite also gives possibilities to use:
- Customizable business rules;
- Structured income calculations;
- Uniform underwriting approaches across teams and locations.
Despite the emphasis on standardization by the platform, it does not disregard the role of professionals. It helps in identifying missing papers, suspicious transactions, and possible problem areas in advance, thus enabling faster examination of exemption cases.
In addition, LoanCite further enhances the quality control and audit readiness of the platform by keeping detailed records of activities undertaken, along with calculations and decisions made. This contributes to reduction of operational risk, possibility to identify mistakes before the loan is issued, and better proof during audit or review by investors. In general, the platform greatly improves the efficiency, accuracy, transparency, and governance of underwriting.
Conclusion
LoanCite demonstrates how it is possible to improve non-QM income analysis thanks to machine learning while avoiding the negative side of the underwriting process, where it is unclear how the results are obtained. It permits lenders to structure alternative income documents, collect numbers with references, make calculations that can be repeated, follow custom rules applied to credit, and confirm that the decision made by human experts has been recorded. It means that the process can be more efficient and predictable since at every step, there is a number connected to evidence, and the result can be traced back to the document.
Non-QM lenders need speed; however, they need to be able to justify the decisions made even more. LoanCite helps achieve both goals. Book a demo.