{"text":[[{"start":7.65,"text":"When Daniel Hegarty first asked an AI platform what mortgage he should get and why, he was startled by the quality of the results. "}],[{"start":18.1,"text":"“These AI platforms are giving phenomenally precise and thoughtful mortgage advice,” says Hegarty, former chief executive of UK mortgage broker Habito. “It built sensitivity models for my mortgage around changes in rates and the pros and cons of fixing versus not fixing — all of the things that mortgage brokers advise on.”"}],[{"start":36.900000000000006,"text":"But when he started playing around with Claude, Anthropic’s AI platform, he was “shocked” by how prescriptive it was willing to be in what is a highly regulated advisory process. “It wasn’t just saying here are some representative numbers that you might be interested in. It was saying: ‘You should definitely get that one.’”"}],[{"start":57.2,"text":"The mortgage process — researching, advising on and executing a home loan — is ripe for disruption by AI. "}],[{"start":64.05,"text":"Live data on mortgage products, including their interest rates and loan-to-value ratio limits, is already widely available online, and the guidance offered by mortgage professionals is readily encompassed by the new technology, which can absorb the lessons of thousands of “good” and “bad” lending decisions in a moment. "}],[{"start":81.35,"text":"For a borrower willing to disclose their financial and personal circumstances, AI offers a quick answer to a big question. "}],[{"start":90.25,"text":"“AI platforms are clearly already giving an absolute tonne of mortgage advice as we speak,” says Hegarty, who is now co-chief executive of African mobile banking service Kuda. A 2025 survey of 5,000 people by Lloyds Bank found that more than one in five used AI for mortgage advice. "}],[{"start":107.65,"text":"For those enthused about the promise of the technology, it holds out the prospect of a speedier, frictionless, more reliable experience for borrowers and brokers in their relationships with lenders. And if it were to be coupled with modernisation of the conveyancing process, the upshot could be a steep reduction in time between offer and exchange of contracts — standing at a record 104 days in April, according to estate agency Connells — with fewer broken chains and more homebuyers willing to take the plunge on a transaction. "}],[{"start":138,"text":"But taking out a home loan is the biggest financial commitment most of us will make. Advising on mortgages is subject to strict controls designed to protect consumers from costly and potentially disastrous mistakes. "}],[{"start":150.75,"text":"The latest advances in machine learning have profound implications for lenders, brokers and others in the wider homebuying process. Can this relatively new technology be trusted to take on the task? And will consumers and regulators accept it? "}],[{"start":null,"text":"
"}],[{"start":null,"text":""}],[{"start":164.85,"text":"The UK has about 18,500 mortgage broking firms, nearly half of which are one-person operations. These intermediaries advise on and arrange more than eight out of 10 mortgages or remortgages. "}],[{"start":177.35,"text":"Some worry about the potential competitive threat to their industry from AI; one was recently heard at an industry conference suggesting that a big UK lender was secretly “building a Death Star to destroy us”. Others take solace in the belief that regulators, customers — and lenders — remain far from willing to place advised lending decisions in the hands of inscrutable algorithms. Some lenders say it would be counter to their interests to take back liability for mortgage advice from brokers on a large scale. "}],[{"start":207.35,"text":"But many are nonetheless considering how to capitalise on the opportunities presented by AI to improve their services, and in some cases experimenting with the tools in pilot schemes."}],[{"start":219.1,"text":"The first and most basic case for AI is to interpret the scans of documentation provided by loan applicants, such as payslips or bank statements, and converting that information into standardised formats for advisers or viewed by other authorised parties in a transaction. "}],[{"start":236.85,"text":"With the aim of saving time and reducing human error, it is the low-hanging fruit of AI integration, especially when linked with open banking technology that can be used to verify documentation as well as build a picture of a borrower’s spending habits and credit score. "}],[{"start":253.04999999999998,"text":"What then follows? David Morris, head of homes at Santander UK, sketches scenarios in which AI over time takes on more of the advised process. It might start to give prompts to brokers or borrowers to say: “Have you considered X, Y or Z?” helping them understand their options. "}],[{"start":270.9,"text":"One step further along, it could do more of the heavy lifting, analysing various options and offering a recommendation to be validated or checked by a trained human adviser before this is submitted to the client. Finally, in a “blue-sky” scenario, there may be a future in which AI replaces humans entirely, taking charge of the entire process."}],[{"start":292.34999999999997,"text":"Morris says the first stage is “relatively uncontroversial” and the second is arguably not a great leap from that. “The bit I worry about is an unconstrained jump to that end point without thought about the risks.” "}],[{"start":306.34999999999997,"text":"And the risks, according to Morris and several other industry figures interviewed for this article, are numerous. "}],[{"start":312.95,"text":"First, AI remains notoriously vulnerable to error. As people increasingly apply the technology to everything from improving their golf swing to tackling relationship problems, choosing a mortgage would seem an obvious candidate for interrogation. But the recommendations it offers can be based on out-of-date or simply wrong information. "}],[{"start":334.2,"text":"Andrew Montlake, managing director of broker Coreco, says that a client who was recently seeking an offset mortgage replied to his recommendation by relaying alternatives suggested by ChatGPT. Montlake knew immediately that the lenders would not consider the loan, but humoured the client by asking them. “Not one of those lenders would do the offset,” he says. "}],[{"start":355.75,"text":"This flaw partly reflects the tendency of lenders not to make all of the criteria crystal clear they may apply when deciding whether to lend — factors that brokers only learn through trial and error."}],[{"start":366.25,"text":"Adrian Anderson, managing director of broker Anderson Harris, says: “Lenders have so many quirks and nuances. What counts is the knowledge we all have as experienced mortgage brokers.”"}],[{"start":378.85,"text":"He, too, has had his share of clients make futile requests after consulting AI. One asked for a deal with a fixed rate of interest for the life of the loan and no early repayment charges — something that is not possible in the UK market. He believes AI’s inherently facilitative approach — its disinclination to shut down an unattainable premise — is part of the problem. “They’re set up to keep you engaged.” "}],[{"start":401.55,"text":"Another caution commonly voiced by brokers is that AI is ill-suited to eliciting more personal information from a prospective borrower. This includes delicate matters that might have a bearing on an affordability assessment, such as any medium-term plans to start a family or retrain for a new career, or their appetite for risk. "}],[{"start":421.7,"text":"“AI will have a really difficult time being empathetic,” says Stephanie Charman, chief executive of the Association of Mortgage Intermediaries. “It’s going to be really difficult to give you reassurance. It will struggle to identify vulnerability and when you might need further support in different areas.”"}],[{"start":null,"text":""}],[{"start":438.9,"text":"For Morris at Santander, a major concern is the ability of lenders to be able to justify a machine-led decision. Regulations stipulate that banks must have the ability to show how they came to the conclusion they judged optimal for the consumer. "}],[{"start":453.54999999999995,"text":"But AI decisions are notoriously opaque. “How do you evidence decision making, when you’ve got a black box, especially one that’s been developed outside UK financial services? How do you make sure you can evidence and prove advice and consent?” he says. "}],[{"start":468.34999999999997,"text":"As yet, none of these concerns have stopped big lenders from exploring the opportunities of the new technology. How far advanced are they in bringing AI into their mortgage businesses, either in their direct relationships with customers or through brokers? "}],[{"start":483.54999999999995,"text":"Lloyds, one of the big six UK banks, is running pilot programmes to test the benefits and limits of AI in mortgages. In the second half of this year it will launch the pilot of a chat tool for brokers to pose multi-stage, nested questions about their clients’ applications, with a view to speeding up the mortgage process and keeping participants informed about the application’s progress. "}],[{"start":507.15,"text":"It is also looking at developing an AI tool that can give customers general guidance about mortgages, such as current rates and the application process. Andrew Asaam, mortgage director at Lloyds, stresses that this service will steer clear of advice-type queries. “But it is where I think we’ll end up.” "}],[{"start":526.25,"text":"Because mortgage advice is so strictly regulated, he insists the bank is treading carefully. But he adds: “We’ll have sufficient understanding of a customer’s context, background and preferences to be able to have a conversation that says . . . shall I get a tracker, a two-year or a five-year fix? That is what I think will happen.”"}],[{"start":547.4,"text":"He also sketches the idea of an AI aide lurking in the background of a conversation between a Lloyds’ mortgage adviser and a consenting client, chipping in when it detects something the adviser might have missed, or a particular risk preference. “Clearly, these colleagues have already got lots of experience and are qualified mortgage advisers. But it starts to help them shape that conversation. It could run as guardrails,” he says."}],[{"start":null,"text":"
"}],[{"start":572.15,"text":"A bigger prize still is on offer, Asaam says, of which mortgages are just one part: the digitisation of homebuying, which is currently dogged by antiquated rules and paper trails. "}],[{"start":583.1999999999999,"text":"Industry-wide efforts to modernise the conveyancing process over the past two decades have ended in failure. The time between offer and completion continues to lengthen, even as other financial processes quicken. But some believe the advent of AI, though only tangentially related to the flaws of the homebuying process, may act as a catalyst for reform. "}],[{"start":605.1999999999999,"text":"Coadjute, a UK tech company, aims to put property transactions on a blockchain, or digital ledger, to bring together all those involved in a purchase, including lenders, estate agents, conveyancing lawyers and mortgage brokers. It has big property and finance names among its backers: Lloyds, Nationwide, NatWest and Rightmove contributed as part of a £10mn investment round in 2024. Lloyds also last month launched an initiative with Connells and conveyancing group LMS to create a “fully digital” homebuying process. "}],[{"start":638.6999999999999,"text":"Other recent developments include NatWest’s partnership with Australian digital platform Pexa, which aims to fast-track the mortgage process, starting with remortgages. Last month, it carried out its first transaction, completing a remortgage within two days. "}],[{"start":653.8,"text":"Nationwide, meanwhile, in February opened the path to digital signatures on mortgage deeds in a partnership with HM Land Registry and others. And last month the Financial Conduct Authority announced a second wave of firms, including Barclays, Lloyds and Experian, to live-test AI applications. "}],[{"start":672.9,"text":"“If you’d asked me two or three years ago about the feasibility of these things, I’d have said maybe,” says Asaam. But he now believes the technology is advancing so rapidly, with a slew of serious initiatives under way, that significant change is possible. “You can see there’s momentum building.”"}],[{"start":690.65,"text":"The transition to a tech-enabled future is unlikely to be pain-free, for property companies as well as mortgage brokers. Rightmove’s share price has halved over the past nine months, and last week it delivered its latest rejoinder to worries over the impact of AI on its business by announcing a suite of new AI tools to be integrated into its services, as well as expectations of higher revenue."}],[{"start":714.55,"text":"As for mortgage brokers, there is a consensus that disruption is inevitable. But some believe the increasing complexity of mortgage financing — for example, the growth of portfolio working, irregular employment, or the challenges facing first-time buyers — points to a reshaping rather than annihilation. In this world, AI might handle “plain vanilla” transactions, while human brokers step in for more complex cases."}],[{"start":740.1999999999999,"text":"“The danger is that people think AI can solve everything,” says Morris of Santander. “What we have to focus on is a thoughtful, responsible application of the technology. How do we adopt it in a way that blends the best of human nuance and interpersonal qualities with the known benefits that come from a technology that can help us speed things up and reduce costs?” We may not have much longer to wait before we get an answer."}],[{"start":773.8,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1778917000_7417.mp3"}