The FCA wants to supervise at AI speed. Here is what that takes underneath.

In mid-October the FCA set out its direction plainly. It is not preparing new AI rules for the firms it regulates. Instead it intends to use AI itself: reading the documents, emails and interview records firms send it, and automating supervision and enforcement wherever it can.

That is a shift in who has to move fastest. For years the question was whether firms could adopt AI safely. Now the question is whether supervisors can keep pace with firms that already have.

What the FCA actually said

The direction follows the Mills Review, which recommended an AI-enabled agentic supervisory model: tools that watch outcomes across firms in near real time, triage what firms submit and test evidence against expectations. The FCA has been clear that its people stay in charge of decisions. The AI does the reading at a scale no team of analysts could.

The numbers explain why. Tens of thousands of firms are coming into scope, and the firms themselves are moving faster every quarter. Supervision built on reading every document by hand cannot stretch that far.

Why it matters beyond the UK

The UK is not an outlier. Three in four supervisors worldwide now use AI, up from one in four a year earlier. And the bar regulators set for firms is rising: in October, Singapore's MAS issued AI risk management guidelines expecting firms to keep an inventory of their AI, assess how material each use is and keep human oversight in place.

The foundations underneath an agentic regulator

Agentic supervision is only as good as what sits underneath it. Three things matter most. Data that arrives structured and validated, so the AI is not cleaning up after the fact. Rules that are deterministic and owned by the supervisor, so the calls that must be repeatable are made the same way every time. And an evidence trail, so any finding can be traced back to the data point and the rule that raised it.

Get those right and AI becomes a force multiplier for supervisors. Get them wrong and it becomes another system nobody can explain.

What it means for smaller supervisors

Not every authority has the FCA's budget or headcount. The good news is that the foundations do not need a multi-year programme. Most of the value sits in one or two high-volume workflows where analysts spend their days restructuring data. Start there, prove it in weeks, and build out.

If you want a quick view of where your team stands, our three-minute readiness check scores the same foundations. Or see how the Line4 platform puts AI inside the supervisory workflow.