The building society’s AI program puts mortgage member outcomes ahead of cost-cutting — and its governance model sets the example
Skipton Building Society is deploying AI across its 2,500-person workforce. Its group chief information officer has a clear message for mortgage lenders: AI exists to serve members, not replace the humans who advise them.
In July 2026, an internal test put a new AI tool to work on a query about building address handling. This is the kind of question that had taken several days of manual review earlier in the year. The tool returned an answer in four minutes. That tool is Q, a retrieval system built by two Skipton engineers on Azure Foundry using GitHub Copilot.
Q gives developers, architects and business analysts a way to query the rules underpinning the Society’s core banking system. Users can query rule definitions, trace system dependencies, or identify where a change might create knock-on effects. What it does not do is make the call. Colleagues interpret the findings, weigh the options and decide what to do next.
Stay ahead of the UK mortgage market. Subscribe to the Mortgage Introducer daily newsletter for broker news, rate updates, and the industry latest delivered to your inbox every morning.
Skipton’s purpose-led approach
Skipton is owned by its members — savings and mortgage customers — rather than shareholders. Wood says technology choices are always made “for the benefit of our members.” That means the Society is not simply asking how a tool might reduce costs. It is asking whether it can help colleagues solve a problem that matters to members.
“We can use [AI] to create more opportunity, and we can create things that help us serve our members better,” said Jenny Wood, Skipton’s group chief information officer.
Wood also said that Skipton was not seeking to move customers away from human contact. Members choose between 82 branches, a call centre, a chatbot or an app.
The FCA’s Mortgage Rule Review roadmap found broad consensus that AI should improve the mortgage process without replacing advisers. A close reading of the FCA’s AI roadmap for mortgages sets out exactly what that regulatory thinking means for the adviser’s role.
“Purpose-led organisations can always ground themselves in asking what’s the right problem to solve?” Wood said.
Building a culture, not just a capability
The Society’s AI deployment grew out of a cultural program called Enabling Curiosity. Staff are prompted to ask first whether a task needs doing at all, and only then whether technology should handle it.
Thirty colleagues were given Microsoft Copilot licences to explore the tool. The exercise was informal — colleagues shared observations and surfaced use cases that shaped the Society’s understanding of the tool. By July 2026, 80% of its roughly 2,500 employees were using Copilot Enterprise.
Charlie Moran, head of data capability at Skipton, says the deployment was never a headcount exercise.
“None of the work was approached from a cost-saving perspective. We didn’t say, ‘well, there’s 30 of you in this team, but maybe we could create the same output with 15’,” Moran said. “The accuracy is never going to be 100%. But my human accuracy isn’t 100% either.”
The case for guardrails over speed
The question of whether AI can genuinely replace mortgage brokers and advisers is a common issue across the industry. Skipton’s approach is a counterpoint: human judgement and AI capability need not compete.
Freedom to experiment comes with clear limits. A change cannot be rolled out beyond 10 to 15 people without sign-off from the relevant team.
“You can all have approved tools, but there are rules around how you use them. And we will monitor,” Wood said.
Q is part of a wider technology program that includes a Databricks environment on Azure, infrastructure virtualisation and a redesigned digital presence. Skipton’s track record of consistent mortgage lending growth reflects a lender that moves with intention.
That instinct is also part of its approach to AI. Clear guardrails and human oversight are not obstacles to the technology. Without them, Skipton argues, the tool is not ready.
