The debate around AI in financial advice has moved well beyond whether firms should adopt the technology and towards the harder questions of data quality, governance and who bears responsibility when something goes wrong. Chris Davies, founder and director of Model Office, sets out why those questions matter more than the tools themselves.
The data problem beneath the AI opportunity
Most advice firms already hold substantial client data. The difficulty is that it sits in fragments: spread across CRMs, cashflow tools, suitability reports, document stores, compliance systems and spreadsheets. Before AI can add analytical value, that information needs to be connected, consistent and traceable. Davies describes this as a regulatory data layer, or RegData layer, capable of turning a firm’s scattered technology estate into usable management and regulatory intelligence.
The stakes of getting this wrong are real. Incomplete or inconsistent data affects Consumer Duty monitoring, vulnerable customer identification, suitability oversight, complaints analysis and regulatory reporting. AI magnifies both sides of that equation. Good data enables better analysis. Poor data, as Davies puts it, allows inaccurate assumptions and decisions to be produced faster and at greater scale. Speed without accuracy is not an improvement.
The FCA Mills Review found that 95% of large advice firms were already using, or actively considering using, AI, which underlines how quickly the profession is moving. The same review identifies the potential transition towards increasingly autonomous AI and recommends the development of a trusted public-interest AI ‘advice like’ financial capability service.
AI in financial advice and the trust gap among consumers
Consumer attitudes present their own complexity. FCA research cited by Davies found that one in five consumers would be likely to use AI capable of acting autonomously within pre-set goals. Yet the Financial Conduct Authority‘s own research also found that 44% of less experienced younger investors surveyed incorrectly believed AI-generated financial information was regulated. That misunderstanding creates a direct risk for firms: clients may arrive at planning meetings carrying DIY financial plans generated by large language models, assuming those outputs carry some form of regulatory oversight.
Davies frames this as both an opportunity and a responsibility. AI can analyse financial and behavioural data, identify patterns, interrogate client records and support suitability processes. But financial planning, he argues, is not a data-processing exercise. People make decisions about retirement, inheritance, illness and financial security within the full context of their lives. Data provides evidence. AI provides analytical capability. The qualified adviser provides judgement, context and empathy, and helps a client understand the decision they are actually making.
What governance must follow adoption
For boards and senior managers, Davies argues the strategic question should shift. Rather than asking where AI can be applied, firms should be asking whether they have the data quality, governance and human oversight required to trust it. The FCA’s 2026 wealth management survey reinforces this: firms remain responsible for outcomes, including where technology and third parties are involved. That accountability cannot be delegated to a platform or a model.
In practical terms, building an AI-ready advice firm means understanding data ownership, quality, lineage and accessibility; establishing clear AI governance and accountability frameworks; and creating regulatory data environments capable of continuously evidencing conduct, risk and customer outcomes. The likely result is not one monolithic system but an interconnected ecosystem in which data moves securely between CRM, planning, platform, compliance, risk and reporting tools through APIs and, increasingly, AI agents.
Firms that build on that foundation will not be replacing financial planners with AI. They will be giving planners better intelligence and automating more of the procedural and monitoring work around them. With the FCA’s Consumer Duty supervision becoming progressively more evidence-based through 2026, the quality of a firm’s underlying data architecture may prove as consequential as the AI tools sitting on top of it.
