If you were at the Building Societies Association Annual Conference in Edinburgh last month [April], you could not have missed the optimism: the sector is feeling bullish.
But underneath the positive noise about growth and mutual value, a harder truth was circulating in the exhibition hall.
We all know the building society sector needs to modernise – the problem is how we are going about it.
Bluntly put, the reality is that too many lenders are still tinkering at the edges, bolting new technology onto old systems to fix isolated problems.
At the moment, the industry is confusing experimentation with strategy. If we want to hit those ambitious growth targets, we need to accept that chatbots don’t cut it and start looking at technology that actually does the work.
UNDERSTANDING THE AI DIFFERENCE
The world is currently obsessed with generative AI. In the UK mortgage market, we are using it to write emails, summarise documents and answer basic policy questions. That is fine as far as it goes but it is not going to move the dial on lending volumes or operational efficiency.
The real game-changer is agentic AI. As we explore in a forthcoming independent research paper we commissioned, Intelligence in Motion, the difference is fundamental.
Generative AI can tell you what a document says; agentic AI can take that document, validate the data against your lending policy, flag any discrepancies and move the case to the next stage of the workflow. It does not just provide information, it executes multiple tasks.
For building societies, where capacity is often stretched and underwriters can spend half a day chasing missing payslips, that capability is transformative.
REMOVING THE FRICTION
We do not need to wait for a science-fiction future to start seeing the benefits. Task-oriented agentic AI is deployable right now.
Consider the front end of the process. Instead of an underwriter manually reviewing every uploaded document, an AI agent can ingest the file, extract the relevant data and check it against the application. It can spot potential fraud indicators in real time. It presents the underwriter with a clean, verified case file, allowing them to focus on what they do best: applying human judgement to complex lending scenarios.
This is where the mutual sector can really win. By using technology to strip out the administrative friction, we free up our people to deliver the relationship-driven service that sets building societies apart from the high street banks.
OVERCOMING THE INTERNAL HURDLES
So why aren’t we moving faster? The technology is ready, but the mindset often isn’t.
At board level, AI investment is still frequently viewed as a risky transformation cost rather than a tool for protecting revenue and driving down operational expenses. There is a fear of getting it wrong, particularly when it comes to regulatory compliance. But as our research shows, you do not have to jump straight into fully automated decision-making.
A smart way to start is by pointing agentic AI at your quality assurance process. Let it review historical cases to learn what good looks like. It is a low-risk environment that generates the data you need to build confidence before rolling the technology out to the front line.
You also do not have to build this capability yourself. Partnering with specialists who understand the UK mortgage market allows you to integrate agentic AI without taking on massive delivery risk.
The infrastructure is there. The technology works. The only thing holding us back is the willingness to adopt it.
If building societies want to turn the optimism of the Edinburgh conference into tangible growth, we need to stop tinkering and start putting intelligence into motion.
