Summary
The UK has one of the most respected financial services sectors in the world. It is also evolving faster than the infrastructure that currently underpins it.
At Ohpen, working alongside our partners at Cognizant across both the Dutch and UK markets, we are seeing something shift. Not in the volume of AI conversation — that has been relentless — but in the specificity of it. What AI will actually do, how it will do it, and where it will land first in financial services is becoming clearer.
Mortgages are one of the most compelling places to watch.
What we mean by AI
It is worth being precise. AI is not a single capability — it is a category that spans machine learning, natural language processing, computer vision, deep learning, and expert systems, each powering different types of applications. What is most relevant to the mortgage market right now is a more autonomous variant: agentic AI, which does not simply analyse and advise but can act within predefined frameworks.
That distinction matters enormously in a regulated context. The old argument — that there is too much regulatory risk to give AI genuine agency in a mortgage transaction — is losing ground. Not because the regulation has changed, but because the models for operating within it have.
The problem agentic AI is solving
A mortgage transaction from end to end currently spans multiple siloes: product design, sales, credit decisioning, underwriting, valuation, fraud and AML checks, legal, operations, completion, and servicing. Data sources are fractured. Some of those siloes are entirely separate businesses in different industries. The underlying business processes have not fundamentally changed in decades.
Where open banking promised to bridge some of these gaps and largely fell short, agentic AI offers a more complete answer — because it can act on data continuously, not just at the point of application.
The shift is significant. Today, a customer applies for a mortgage. In an agentic model, they authorise their data once and an AI agent takes over: continuously collecting, validating, and updating relevant information from secure, live sources. Within predefined policy and risk frameworks, affordability is assessed, deviations are flagged, and follow-up actions are initiated — without the customer having to re-submit a document or chase a response.
Real-time property valuation through API connections with land registry and valuers becomes part of the same workflow. Borrowing capacity stops being estimated and starts being calculated.
What this looks like in practice
In the Netherlands, the combination of PSD2, iDIN (the Dutch bank-backed digital identity standard), and linked tax, employer, and asset data provides a live financial profile that makes this genuinely operational today. The UK is building toward a comparable foundation — GOV.UK One Login, open banking APIs and governance, commercial digital identity and KYC providers — and the direction of travel is clear.
Lead times that currently span weeks compress because the friction driving them is removed, not optimised around. The mortgage stops being a standalone application process and starts being integrated into the home-buying journey itself: financing options becoming available as customers search property platforms, borrowing capacity visible before a formal application begins.
Standard calculations and affordability checks are automated. This doesn’t eliminate advice — it relocates it. The question of where a broker’s time is genuinely valuable becomes sharper and more honest.
The broker’s role, redefined
Agentic AI doesn’t replace the independent mortgage adviser. It removes the administration that currently surrounds the role and has little to do with advice.
The added value of an adviser — interpretation, context, helping a client understand the long-term implications of a decision beyond interest rates and monthly payments — is not automatable. What is automatable is the surrounding coordination, document chasing, and process management. Brokers freed from that can focus on the moments where human judgement genuinely matters: complex life decisions, vulnerability, long-term risk.
Personalised advice also becomes scalable. Decisions around fixed-rate periods, flexibility, and repayment can be recalculated on an ongoing basis during the mortgage term, based on life stage, career development, and risk profile. That requires an API-first infrastructure where origination and servicing work as a connected system rather than separate processes.
After completion, the same capability that drives origination can monitor for life events, make proactive restructuring proposals, and link sustainability measures to property value and terms. Prevention replaces reactive collections — which aligns directly with the FCA’s Consumer Duty requirements around vulnerability.
Conclusion
The technology for this is not theoretical. It exists. The question is not whether the mortgage market will move in this direction — it is which participants will move with it early enough to define the terms.
Platforms, banks’ AI agents, advisers willing to redefine their role: the opportunity is open. The constraint is no longer capability. It is intent and pace.
Jerry Mulle is UK Managing Director at Ohpen. The original version of this article was first published in The Intermediary.