By Liz Syms — Co founder of mortgagelab.ai, CEO of Connect for Intermediaries, Chair of the Society of Mortgage Professionals.
I read the recent Aveni article about ‘Hazel’ which launched last month. Hazel is the AI advice platform launch that sent UK wealth management into a quiet panic. One line was really poignant. Joseph Twigg argued that the panic itself was the story. Not the product but the reaction.
He’s right. And every mortgage adviser reading this should consider that for a minute, because the same wave is coming for us.
This week I read another piece. Different lens, but same horizon. The headline framing was simple: AGI by 2030.
Demis Hassabis, CEO of Google DeepMind, now puts the odds of artificial general intelligence arriving by 2030 at roughly 50/50, and he describes the resulting economic transformation as “ten times the scale, ten times the speed” of the Industrial Revolution. Dario Amodei, CEO of Anthropic, has written about a “compressed 21st century” meaning decades of progress packed into years. Sam Altman calls his version a “gentle singularity” and pencils in 2026 as the year AI systems begin producing genuinely novel insights. Peter Diamandis is blunter: AI will redesign every job in the next three years, and white-collar work is first.
If even half of that is right, the question for a UK mortgage adviser is not whether AI is coming for the role. It already is. The question is what the role looks like on the other side.
Here’s the good news. We start with an advantage almost no other professional services category has.
The mirror Hazel held up
The panic over Hazel, Twiggs says, exposed how poorly our industry understands what it is actually selling. For years we have justified our fee with expertise and access, you pay a human to know things you don’t, and to be there when you need them. AI doesn’t just compete with that model. It makes the underlying logic harder to defend. If a machine can run a suitability assessment faster and more consistently than a person, “you’re paying for our expertise” stops being a value proposition.
Now translate that into mortgages. Affordability calculations. Criteria search across 90 plus lenders. Document assessment. Application packaging. Compliance summaries. Initial fact-find. Sourcing etc
Every one of those tasks is, right now already in 2026, is becoming something an AI agent can do. The FCA has openly said it wants brokers to use AI to deliver better, faster advice as part of the upcoming mortgage market reforms. Mortgage Solutions is already writing about “the agentic adviser.” The Intermediary has run a piece titled “from mortgage broker to relationship manager.”
This isn’t speculation. It’s policy direction, vendor roadmap and consumer behaviour all moving in the same direction at the same time.
What an adviser is actually for
So if the analytical work is being absorbed by AI, what is an adviser for? I would suggest just four things, and only the first one is partly technical.
Judgement. The ability to look at a complicated, messy, half-told human story and decide what matters. Self-employed income that doesn’t fit a lender’s template. A divorce mid-application. A first-time buyer whose parents want to gift the deposit but not all of it. AI can search criteria; it can’t decide what the criteria should be in the context of someone’s whole life.
Trust. The thing the regulator keeps calling “good outcomes” under Consumer Duty is really a trust problem dressed up as a compliance one. Trust is built over years, not seconds. AI can be transparent, but it cannot yet be loyal.
Coaching. Most clients don’t actually want a transaction. They want someone to tell them whether to move now or wait six months, whether to overpay, whether to fix for two years or five. That is closer to coaching than mortgage broking, and most advisers are already doing it, they just don’t price it that way yet.
Context. Contextual reasoning such as knowing that this client says she’s relaxed about rate risk but her last three life decisions suggest otherwise. Knowing that this couple needs more time to think, not more product options. Knowing when to push and when to hold back. AI is improving fast on cognition. It is not improving on this.
The unfair advantage we already have
A typical UK mortgage adviser sits on top of two assets that AI cannot manufacture: a multi-year relationship with a client, and a rich, long record of that client’s financial life. We’ve walked them through a first purchase, a remortgage, a second child, a porting, a buy-to-let, and a protection review. We know what their parents do for a living. We know their attitude to risk because we watched it shift after the mini-budget.
A start-up cannot replicate that. An AI advice platform cannot replicate that. A direct-to-lender consumer app cannot replicate that.
The risk isn’t that AI replaces us. The risk is that we use the next eighteen months badly and miss the opportunity. By competing with AI on speed and accuracy we will lose the race, instead of using AI to amplify the things only we can do.
What to do this quarter, not next year
Use AI to compress the admin. Document collection, fact-find pre-population, compliance file building, initial criteria screening, follow-up chasers. Aim to give seceral hours a week to every adviser in your firm and spend that time on client conversations, not more cases.
Audit your client data. The richer your CRM, the more valuable you become and the harder you are to displace. If client records are scattered across email threads, spreadsheets and a CRM, fix that this quarter. A good data layer is what makes you AI-augmentable rather than AI-replaceable.
Rewrite your client journey around relationships, not transactions. Annual reviews are not a compliance task. They are the most defensible thing you can do. Schedule them. Run them as coaching conversations. Map them against life events, not product anniversaries.
Practise being a coach. The advisers who thrive in 2030 will sound less like product specialists and more like trusted advisers. Improve soft skills and relationship selling skills. Take Consumer Duty vulnerability training seriously. Learn to ask better questions than the AI can.
Get loud about governance. The industry’s compliance infrastructure was not built for AI at scale The networks, lenders and brokers who get ahead of this, who can explain to a client and a regulator exactly how AI was used in a recommendation, will own the trust premium. Don’t wait for the FCA to write the rules. Write your own first.
The honest bit
Some firms may not make this transition. Some advisers will retire earlier than they planned. The economics of solo adviser firms, which is already down, will keep tightening. That is the realistic part.
But the optimistic part is real too. The mortgage adviser who walks out of 2030 with the right tools, the right data, the right client relationships and the right governance is not a diminished version of today’s adviser. They are a more valuable one. Closer to a financial coach, more aligned with Consumer Duty, more profitable per client, and far harder to displace.
The AGI-by-2030 framing tells us what the timeline is. The bit nobody else can do for us is decide what we want to be when our analytical work is no longer the product.
