During 2026, the share of UK mortgages originated through brokers is projected to reach 91% – reinforcing the highly intermediated nature of the country’s mortgage market. And partly due to this extra layer of complexity, the mortgage process here is painfully slow: a recent survey by Nottingham Building Society found that 41% of brokers believe it’s as slow or slower than two years ago, and 31% want to see it streamlined through new technologies.
Such figures underline the challenges around speed and efficiency facing every stage of the mortgage process in the UK. However, technological advances mean these challenges are now balanced by the potential for lenders to raise their game dramatically. The biggest opportunity? Adopting an agile core and AI engines to drive rising operational and financial performance at each step in the value chain.
A complex end-to-end process…
To set the context for this opportunity, it’s important to understand the dual role played by mortgage brokers. One aspect is conducting customer due diligence, saving lenders the cost and effort of doing it themselves. The other is advising customers on the available products, which the providers aren’t allowed do under current regulations. For lenders, the effect is double-edged: the intermediation by brokers relieves them of a heavy administrative burden — but also prevents them from controlling distribution or building direct customer relationships.
Against this background, the end-to-end mortgage process can be broadly divided into three “buckets”. The first is origination: essentially sales and distribution, including due diligence and risk assessment. Second, servicing, which kicks in after the loan is made and involves ongoing management of payments, statements, changes and so on. The third element is liquidity management by the lender to enable it to offer more loans.
…beset by embedded inefficiencies throughout
Ideally, the goal is to have high distribution and low operational costs in the origination bucket; low friction in servicing to minimize the cost-to-serve on each loan; and a large pool of liquidity to underpin rising lending. But the reality today is that all three areas are hampered by immature technologies, inadequate connectivity interfaces and high levels of manual intervention — all adding cost and risk.
The overall effect? Severe operational drag throughout the process. And in origination and servicing, the reliance on manual interventions is further increased by the fact that mortgage lending is an activity that’s high-risk, high-value and highly regulated. Together, these factors mean the cost of running, maintaining and changing systems across the industry is continuing to rise relentlessly, in turn further boosting risk. Put simply, it’s a vicious cycle that needs to be broken. The only question is how.
The way forward: an agile technology stack – overlaid with AI agents
The answer lies in cutting the cord and investing in better tech — which involves focusing on two areas. First, the core technology, with a view to modernizing the costly and ageing legacy systems at the heart of the business. The key here is to move to a more agile IT stack, bringing greater flexibility and the ability to adapt to market changes by dialing capacity up and down dynamically. Currently, some tier-two banks take about six months to launch a new product. With more modern platforms, they could do this in three weeks.
Having specified the right technology foundation, the next step is to plan out the adoption of automation and AI in operations across the business. While this might look daunting, it’s now imperative in order to remain competitive. The optimal approach is to start with one of the highest-value, lowest-risk use cases, use AI to create an agent to address it on the new technology stack, and prove that it works and delivers value. Then move on to the next use case, then another — and scale up to build what will become the agentic architecture driving future operations.
An example of a use case to target? Document case management. The underwriting process during origination requires a mass of detailed documentation and information shared by the customer. Currently it’s a time-consuming and labor-intensive activity. Using agentic AI, we’ve helped clients create an automated engine that extracts the necessary information, indexes it, cross-references it with relevant industry data sources, and classifies it to streamline and speed up underwriting. Our experience shows that the impacts can include a 50% to 60% decrease in underwriter effort and a 30% to 40% improvement in turnaround time.
Alongside transformation with AI, a further major area of opportunity for UK lenders is using nearshore/offshore operations to reduce cost-to-serve. This approach has already seen rapid take-up in territories such as the Netherlands, and the industry in the UK is ripe for a similar transformation. After relocating the selected operations, it’s possible to build a context layer on top that supports agentic interventions. Take mutations such as changes to phone numbers or employment status: done manually, these can take 25 to 30 minutes each — but AI can handle them in milliseconds.
A rising tide of value
The positive effects of the cost and productivity gains we’ve outlined extend well beyond the area of the business where they’re generated — and can combine to create a virtuous cycle linking every step of the mortgage process. How does this work? Say a lender reduces its cost-to-serve per loan. As a result, it can offer a higher interest rate on its savings accounts, attracting higher inflows and boosting liquidity – which in turn means it can lend out more money through mortgages and reap higher revenues. It’s a rising tide of value that companies can tap through the right technology. And early adopters will get a head-start. Are you ready to lead the way?
