But few topics spark as much debate as the potential for artificial intelligence (AI) to ‘take over’ the role of the mortgage broker. It’s a question that’s not just theoretical; it resonates deeply with many professionals in our field.

In this article, I’ll draw on the latest insights to explore the feasibility of an AI-driven mortgage process, the realistic timelines involved, and why a hybrid approach might be the most effective way forward – all while ensuring that we protect jobs and maintain the human element that’s so vital to our industry. Please read the second instalment of this article, as I do believe there is light at the end of the tunnel.

Assessing the viability: Can AI really handle it all?

Let’s start with the basics: is a fully automated, AI-powered mortgage broker even possible? The short answer is yes – technically, it’s already 80-90% viable today for standard cases. We’ve got the building blocks in place, thanks to advancements in data access, APIs, and machine learning.

Consider data access, for instance. In the UK, open banking has matured significantly, allowing customers to grant secure access to their bank data, credit files, and income streams. This means an AI system could digest 12-24 months of transaction history in moments, crunching numbers on affordability and spitting out a decision in principle (DIP) faster than you can brew a cup of tea. Underwriting for employed borrowers with clean credit? That’s already highly automatable, with AI handling the heavy lifting on standard assessments.

Sponsored

Aldermore Insights with Jon Cooper: Edition 7 – Opening doors in a tougher first-time buyer market

Sponsored by Aldermore

However, it’s not all smooth sailing. Lender integrations vary – some, like Halifax and Nationwide, offer robust APIs, but others lag behind, requiring aggregator partnerships for full automation. Regulatory hurdles from the Financial Conduct Authority (FCA) add another layer; AI must demonstrate explainability, suitability, and fair treatment, which demands rigorous compliance features. AI-assisted property and legal completion processes are emerging but fragmented, with conveyancing APIs still in their infancy. And don’t get me started on crypto integration – it’s low viability today, as most lenders shy away without manual checks, though regulation is evolving.

Consumer trust is growing, especially among younger buyers accustomed to digital platforms like Revolut. Overall, I’d estimate that a digital mortgage platform could manage 70-80% of UK borrower types right now. By 2028-30, as lender APIs, AI transparency, and digital conveyancing advance, it could cover nearly all standard cases.

In essence, the tech is here – it’s the ecosystem and regulations that need to catch up, making full viability moderately challenging but achievable in 3-5 years.

Timelines and the human cost: Spreading the impact

Now to that alarming question about timelines and job losses. The fear is real: if AI can automate so much, what happens to the brokers? A full AI takeover could indeed disrupt roles, potentially making up to 90% of standard case handling redundant in the long term. But here’s the key – we don’t have to rush into it. A phased timeline isn’t just palatable; it’s practical and ethical.

For a minimum viable product (MVP) pilot, we’re looking at 6-9 months to get something operational. Full market readiness, including FCA approval, might take 24-36 months. To mitigate job impacts, I think the banking industry would advocate for a staged roll-out: start with automating admin-heavy tasks like data gathering and pre-qualification, then gradually integrate product selection and underwriting. This spreads the transition over 3-5 years, giving brokers time to upskill in areas AI can’t touch – like emotional support and complex case advocacy.

Think of it this way: rather than fearing redundancy, brokers can pivot to higher-value roles. In my coaching sessions, I often advise firms to view AI as a tool that frees them from drudgery, allowing more focus on client relationships. By spreading the adoption, we can retrain staff, perhaps through industry programmes, ensuring minimal disruption. It’s not about job losses, it’s about job evolution – and a thoughtful timeline makes that palatable for everyone.

The hybrid model: The best of both worlds

While a pure AI takeover is feasible, I believe the future lies in a hybrid model – where AI acts as the engine, and the broker as the pilot. This approach leverages technology for efficiency while preserving the irreplaceable human touch.

Imagine a client journey where AI leads on discovery: retrieving data via open banking, providing instant affordability insights, and generating product options. The broker then steps in for interpretation – explaining risks, adding nuanced advice like: “This lender is great for self-employed applicants,” and ensuring FCA compliance. For exceptions, such as inconsistent income or foreign deposits, the human broker crafts persuasive narratives that AI simply can’t replicate.

In terms of split: data verification? 100% automated. Emotional support and relationship management? 70-80% broker-led. This hybrid not only boosts efficiency but enhances client experience – transparent AI processes (“We’re analysing your finances now”) paired with personal check-ins at key moments, case submission and offer stages.

From a business perspective, this shifts the model from manual, transactional advice to outcome-driven coaching. Brokers compete on clarity, empathy, and speed, not just rates. In my work with firms, I’ve seen hybrids increase client retention through proactive AI-monitored remortgage alerts, blended with human outreach.

Tech-wise, it’s straightforward: a client portal for uploads, an AI layer for matching, and a broker dashboard for insights and support. Culturally, it’s a move from “I’ll get you a mortgage” to “I’ll optimise your finances for life.” And competitively? It positions you as credible with digital natives, compliant with FCA Consumer Duty, and differentiated from both outdated manual brokers and impersonal robo-apps.