Mortgage AI: The current reality? – Lusted

While the recent wave of generative AI tools and machine learning breakthroughs has sparked renewed excitement, the mortgage industry has quietly been using AI in various forms for years.

I was reminded recently of an event we hosted back in 2018 under the title ‘Mortgage AI: the current reality?’. I was joined by speakers from Microsoft, Eligible and Twenty7tec to explore the current (at the time) and future possible applications in the mortgage industry.

The immediate opportunities identified seem just as relevant today: 

1) Automated document processing, allowing lenders to use AI-powered tools to extract and validate data from payslips, bank statements and other documents to speed up the process and reduce manual work (and the risk of manual errors).

2) Utilising machine learning models to support underwriting and credit decisioning. The term “big data” was a hot topic back in 2018 and machine learning (a subset of AI) offered the promise of analysing vast amounts of financial and behavioural data to enhance credit decisioning, moving beyond traditional credit scores to build a more nuanced risk profile. 

Sponsored

How to get your first-time buyer clients mortgage ready

Sponsored by Halifax Intermediaries

3) Using AI algorithms to help flag suspicious transactions and identify patterns of potential fraud in real time.

4) Offering chatbot interfaces to borrowers and brokers to support handling queries, guide users through applications, and provide 24/7 support. 

5) Enhancing automated valuation models (AVMs) with AI to improve the accuracy of property pricing and speeding up underwriting decisions. 

We also debated a future of end-to-end mortgage automation, hyper-personalised mortgage products and predictive analytics being used to proactively intervene to prevent arrears. On the day, the speaker from Microsoft also gave a demonstration of the next wave of cloud-based tools it had launched with the aim of lowering the barrier to entry for those looking to leverage machine learning or offer a chatbot. 

Where does the mortgage sector stand with AI? 

Given seven years have passed, I think it’s useful to ask the question again: what is the current reality of adoption in the mortgage industry in 2025? 

While the speed, sophistication, and accessibility of AI technologies today have significantly increased – and the barriers to entry, therefore, lowered even further – I think it’s fair to say that the average mortgage case in the UK isn’t greatly impacted by the advent of AI yet.

That’s not to say that innovation hasn’t happened. There are some great examples of companies and services delivering on some of the immediate opportunities identified above, particularly when it comes to fraud detection, automated document processing, credit decisioning and chatbots. 

We have supported the development of a number of AI-based products in the mortgage industry, including the Vida Milo and Kensington chatbots, both of which were created to support brokers with common queries, including lending policy questions. 

But while there is an ever-growing fintech ecosystem of AI-powered tools and services available, many incumbents are still held back by their reliance on fragmented, legacy systems that limit integration, agility, and the real-time data access AI depends on.

For many lenders, these systems were never designed with AI in mind. As a result, even the most promising AI solutions often end up bolted on to outdated infrastructure, which limits their impact.

This doesn’t mean throwing out everything and starting again. It means taking deliberate steps to create a more flexible, modular tech environment – one that allows AI to be embedded seamlessly into key processes, not just layered on top. As many lenders have been busy in recent years undertaking the groundwork of modernising their core platforms, as well as improving data quality and interoperability, I think we’ll start to see greater impact in the real world in the coming years. 

I also think the cautious approach to date has been driven by growing regulatory expectations around fairness, transparency, and explainability, so it is equally important that AI is implemented responsibly. Lenders must invest not only in the technology but with the right safeguards that ensure these tools enhance outcomes without introducing new risks. 

The potential for AI to have a bigger real-world impact on the mortgage sector is clear, but to move from isolated use cases to industry-wide transformation, mortgage lenders must pair AI enthusiasm with the strategic, long-term commitment to rebuild the digital foundations.