The role of AI in mortgage advice is evolving rapidly. It’s transforming the way we do business in the UK mortgage market, improving brokers’ experiences interacting with lenders and, increasingly, making it easier for customers to self-serve. Borrowers encounter AI in much of their daily lives, but we should not overlook the fact they still need and demand valuable human touchpoints. Some 90% of mortgages go through brokers but about 65% of borrowers still want meaningful human interaction. AI is clearly part of a hybrid solution.
What do we mean by AI?
According to the Organisation for Economic Co-operation and Development, artificial intelligence – AI – is “a transformative technology capable of tasks that typically require human-like intelligence, such as understanding language, recognising patterns and making decisions”.
The latest artificial intelligence sector study 2023, published in October last year by the government, estimates that there are currently 3,713 active UK companies providing AI products and services. Most are small start-ups, though we are all becoming much more aware of AI assistants such as ChatGPT, Microsoft Copilot, Google Gemini and Perplexity now mainstream.
AI is a term often used to describe complex machine learning as opposed to truly cognisant software programmes that can grow organically. The AI we use today relies on quality data input at a large volume scale to allow algorithmic analysis that identifies patterns, trends and can extrapolate using experience-based logic. A common misconception in the industry is that the use of AI equates to the use of chatbots, which as we have outlined above is only a small part of it.
Where we started
The first firms in the mortgage market began claiming AI capabilities as early on as 2015. But while they were ahead on using decision-tree models with a slick, consumer-friendly interface, that level of technology wasn’t anywhere near where we are today.
Where we are now
From decision trees to fully interactive chatbots, AI-enabled assistants have got considerably better at providing useful support to brokers when interacting with lenders. We’ve seen that relationship develop over the last 12 to18 months, with some of our clients now using AI to underpin customers’ refinancing transactions directly. More importantly for intermediaries, is how that service is being used by lenders to support borrower choice, allowing for a direct transfer to an adviser when requesting a product transfer.
For borrowers who want a simple switch, this process is much faster and a lot less hassle. The option is there to self-select, but we are seeing many opt for advice, keeping their broker in the loop. Not only does this allow the borrower an easier experience, it also makes life much easier for their adviser – minimising repetition, administration while keeping comprehensive compliance records.
Where is AI going?
During last year’s interviews for our annual Mortgage Efficiency Survey research report, we saw that lenders were starting to look at borrower behavioural data using AI models, allowing them to interrogate large data sets previously too unwieldy for manual analysis.
This type of scrutiny is becoming more mainstream among mid-sized and smaller lenders, which have perhaps been held back by the cost of investing in technology and systems to support this type of work.
Under the consumer duty, spotting borrowers before signs of vulnerability present is central, with AI extremely well placed to support lenders to achieve this. Assessing borrower options and potential suitability for various product routes is also made possible at trend level, which will help to inform best practice for individuals in future.
It’s worth going back to the OECD definition of AI as “understanding language, recognising patterns and making decisions”. The last of these capabilities is where the future of AI really lies. Research house Gartner’s report Top Strategic Technology Trends for 2025: Agentic AI outlines this clearly.
What value can AI bring to the mortgage process?
According to Gartner, where AI is given agency, in other words the ability to select what actions to take for achieving particular outcomes, it will change the future of decision making. Having the power to analyse complex datasets, identify patterns and act quickly will start to move us away from labour-intensive data modelling, leading to better problem solving, reducing time to action and enabling new “concepts of scale”.
Agentic AI (with this agency capability) will dramatically upskill workers and teams, enabling them to manage complicated processes, projects and initiatives. This comes with huge upside opportunities for mortgage businesses. However, as Gartner points out, “the orchestration and governance of autonomously acting software entities require advanced tools and strict guardrails”.
The potential to improve customer experiences, allowing for swifter identification of borrower needs and giving systems the ability to make decisions based on those, is huge. But to maintain quality and guard the integrity of service, advice and ensure the right support from people is there when needed will require clear implementation and operational parameters.
For agentic AI to add real value, incorporating it into lenders’ processes must include legal and ethical guidelines on autonomy, liability, security and data privacy. There is no room for error – having the right partner with the experience and expertise needed to get this right will be key.
Hamza Behzad is business development director at finova
