Most people in financial services have heard the pitch by now. AI will automate advice, cut costs and make planning accessible to everyone. Some of that is true. Algorithms are genuinely good at certain parts of the job, and they’re getting better quickly. But anyone who’s sat across from a client going through a divorce, or tried to unpick a family argument about inheritance, knows that a large chunk of financial planning has nothing to do with numbers.
The FCA’s Mills Review, published in July 2026, flagged this tension, warning that AI will be a systemic driver of change while stressing that existing consumer protection rules still apply. So where is AI actually making a difference, and where does it keep falling short?
What AI Already Does Well
Portfolio Optimisation
The parts of financial planning that suit automation tend to be data-heavy, rules-based and repetitive. Portfolio optimisation is the obvious one. Algorithms can rebalance thousands of portfolios in seconds, maintaining target asset allocations far more efficiently than a human could. Tax-loss harvesting works the same way. An AI system can scan a portfolio daily, spot positions sitting at a loss and execute swaps to offset gains automatically.
Document Search
Document parsing has moved fast too. Pension statements, tax returns and trust deeds used to take hours to read and summarise. Machine learning tools can now extract key data points and flag inconsistencies in minutes.
Risk Profiling
Client risk profiling has improved along similar lines. Traditional questionnaires ask clients to rate their comfort with volatility on a scale of one to ten. Newer AI tools go further, analysing spending behaviour and past investment decisions to build a picture of how someone actually behaves with money.
Where the Algorithms Hit a Wall
Here’s where it gets more complicated. A client might say they want to retire at 60 with a comfortable income. But after a few conversations, it turns out what they really want is to stop working for their current employer and spend more time with grandchildren. Those are different problems with different solutions, and no algorithm is going to tease that out.
Family dynamics are another blind spot. Inheritance planning sounds technical, but in practice it’s deeply personal. Who gets what, and when, often depends on relationships, second marriages and unspoken expectations. A planner who’s known a family for years will pick up on things that don’t appear in any dataset.
Then there’s behavioural coaching. During market downturns, the biggest risk to a client’s long-term returns is usually the client themselves. Talking someone out of panic selling requires empathy, trust and sometimes a firm conversation. AI can send a reassuring email, but it can’t replicate the relationship that makes a client actually listen.
The Hybrid Model That’s Taking Shape
What’s emerging in practice is a middle ground. Technology handles the heavy lifting on data, modelling and compliance, while a human planner stays at the centre of the relationship.
In most hybrid setups, cashflow modelling and scenario analysis map out a client’s future under different assumptions, but a dedicated planner runs the conversation, interprets the results and adapts the plan as life changes. Rathbones financial planning services work on that basis, pairing technology with an adviser who knows the client. It’s the model most regulated firms are settling on, and it’s not hard to see why.
Cashflow models are only as good as the inputs. A planner who understands the client’s real situation will build better assumptions than any default dataset.
What the FCA Expects
The regulatory picture matters here. The FCA has been clear that it won’t introduce AI-specific rules. Instead, it expects firms to show that existing frameworks, particularly the Consumer Duty and the Senior Managers and Certification Regime, already cover their use of AI. The House of Commons Treasury Committee reinforced this in January 2026, recommending practical guidance on how consumer protection rules apply to AI-driven advice.
For planners, the Consumer Duty creates a specific challenge. Automated tools need to demonstrate good outcomes for clients, not just technically suitable recommendations. If an algorithm suggests a product that’s compliant on paper but wrong for the client’s actual situation, the firm is still responsible. That’s another reason the hybrid model looks like the direction most regulated firms will take.
A Better Split, Not a Full Takeover
AI is making financial planning faster in areas where speed and data are what matter. But the clients who benefit most from professional planning tend to be the ones with the most complexity in their lives, and that complexity is human. The firms that get this balance right will use AI to free up their planners’ time for the work that actually requires a person in the room.
You should know: There’s no certainty with investments. Values and income can decrease as well as increase, and you could end up with less than you started with. How an investment has performed in the past doesn’t tell you how it will perform in the future.
