Troublesome bottlenecks in mortgage manufacturing are adding to industry pressures, but AI trained in the lifecycle and language of loans is adding flexible, affordable capacity.
In February 2020, as Covid-19 was beginning to grip the world, employees at the Jaguar Land Rover (JLR) manufacturing headquarters in Changshu, China were hurriedly packing suitcases. These workers for the British automaker weren’t preparing for a trip: they were stuffing the luggage with car parts.
Maintaining the flow of components is vital for any manufacturing process. Transporting automotive parts amid passenger luggage was an extraordinary measure, but Covid and the Lunar New Year had conspired to interrupt JLR’s finely balanced “just in time” supply chain. Bosses knew that they had mere weeks before factories in the United Kingdom would be forced to cease production.
The data supply chain bottleneck
Anything that delays or interrupts supply chains can involve huge costs and is a major distraction. Although loan manufacturing doesn’t involve physical components, lenders are responsible for a supply chain of documents. Data extracted from the documents is assembled to build profiles of borrowers, allowing loans to be manufactured with appropriate risk assessments, terms, and pricing.
Turning documents into data is a complex and costly process. The information in most documents is unstructured, so each page needs to be read and classified before the relevant data points can be identified and extracted. This work is both skilled and laborious – data classification and extraction tasks typically take an experienced lending professional several hours per loan to complete. The pressure to balance loan manufacturing costs against risk creates a constant tension between speed and accuracy.
The result is a data supply chain bottleneck, leading to higher costs and slower loan manufacturing processes. Time and money pressures also introduce hidden costs. Overburdened lenders run the risk of missing opportunities to maximize revenues. They lose the freedom to innovate and improve their competitiveness. Product leaders need to address both apparent and hidden costs to ensure a sustainably profitable business.
Cost challenges for product leaders
Manufacturing costs are straightforward. Like adding workers to a factory floor, expanding the size of an origination team allows loans to be manufactured more quickly but at a higher cost per loan. Automation technology has, until recently, struggled to match the capabilities and accuracy of trained people. However, task-focused artificial intelligence (AI) is now able to outperform people on docs-to-data conversion and data verification. Think of it like the reliable robotics that make modern car factories so efficient.
Opportunity costs are less obvious, but no less real. A sluggish data supply chain restricts loan manufacturing volumes – like a carmaker that can’t get enough cars to the dealership. Lenders are faced with three unpalatable options: spend more on people and accept a smaller margin, force more loans through the system by reducing quality (higher levels of inaccuracy and risk), or accept constraints on volume and miss out on revenue altogether. AI that is both faster and more accurate than people now exists, making automation in the data supply chain a more agreeable fourth option.
Innovation costs can be the hardest to calculate, but lenders that fail to develop their products or explore other market opportunities rob their business of possibilities. Think of an automaker diversifying with convertible and crossover models, or moving into an adjacent market for vans. Lenders innovate by devising competitive new loan products, moving into mortgage servicing, or investing in the customer experience. A clogged supply chain can thwart these efforts: longer times to close frustrate borrowers, and overburdened loan officers can struggle to adapt to new products and services. The outcomes are bad reviews that can damage a brand and lending businesses that lack resilience to changing market conditions.
Product leaders have been underwhelmed by data supply chain automation, with many resorting to hiring or outsourcing after experiments with technology. Today, tight labor markets, a worsening economic outlook, and fall in demand means many lenders can no longer afford to improve quality by simply adding more people.
The pressures are leading many lending businesses to look afresh at automation. They’re discovering that technology has moved on and experience about how best to apply it has evolved. These factors describe the TRUE Lending Intelligence platform.
Task specialization allows TRUE products – Trapeze for docs-to-data conversion, VerifAI for data verification – to slot neatly into data supply chain tasks and power fully automated performance – no humans in the loop. The underlying AI is trained to deliver a human-like understanding of the lifecycle and language of loans. It is uniquely able to review documents contextually, seeing the connections between the borrower and their data, and building a complete picture.
TRUE creates flexibility in the data supply chain, allowing it to scale up in busy times and down in quieter periods without committing to salaries or outsourcing contracts. It also gives lenders intelligence that reduces risks and surfaces opportunities. The result is a mortgage-making factory fully in control of its resources, ensuring consistent quality and readily adapting to demand.