RegTech and SupTech are converging. Real-time supervision is where they meet.

For a decade RegTech and SupTech have been treated as two markets. RegTech helps firms comply. SupTech helps supervisors oversee them. We think the line between them is about to blur. Firms' reporting systems and supervisors' platforms are heading for the same place: validated data flowing straight through, and the measures that matter watched as they change. By 2030 we expect most supervisors will want to work this way.

Banks now change every day

The pressure comes from the firms, and banks are starting to publish the numbers. NatWest says more than 12,000 of its coders now use AI, writing around 35% of the bank's code, and that it puts new features in customers' hands four times faster than in 2021. Monzo ships hundreds of changes to production every day. Citi says AI has freed 100,000 hours of developer capacity a week. One smaller UK bank reports its monthly production deployments nearly quadrupling in 18 months, with fewer failed changes, not more.

With AI now used by 90% of software professionals, this is becoming the norm, not the exception.

A bank that ships hundreds of changes a month is a different thing to supervise from one that released a few times a year. A quarterly return describes it the way a photograph describes a moving car. And supervisors are adopting AI too: three in four now use it, up from one in four a year earlier, according to Central Banking.

Two markets, one direction

The Financial Stability Board saw this coming in 2020. It found that SupTech could “generate real-time indicators of risk to support forward looking, judgement based, supervision”, and it pointed to the potentially catalytic role of data standards. Both halves of that sentence matter. Real-time indicators need data from firms. Data standards are what let a firm's system and a supervisor's system speak to each other without a person re-keying numbers in between.

That is where the two markets meet. A firm's RegTech produces validated data. A supervisor's SupTech consumes it. The join between them is the reporting pipe, and today, for most authorities, that pipe is still a spreadsheet attached to an email.

What real-time supervision looks like

Real-time supervision does not mean watching every transaction. It means seeing the handful of measures that matter for each firm as they change, rather than a quarter later. Some authorities are already part of the way there. As early as 2017 the National Bank of Rwanda was pulling some types of data daily from the firms it supervises, so its supervisors could monitor the market as it moved. In Austria, banks and the central bank built a shared, granular data model together, and almost every report is derived from it.

The Bank of England's Future Banking Data paper asks the questions underneath: the right data, at the right frequency, through processes that are still fit for purpose. The ECB's Integrated Reporting Framework will replace many separate collections with one granular one. The direction is the same everywhere: fewer, richer, more frequent flows.

Defending the whole market, not one firm at a time

The bigger prize is what supervisors can see together. Risk rarely sits inside one firm. In the BIS Innovation Hub's Project Aurora, analysing payment data across institutions with privacy-enhancing technologies detected potentially up to three times more money laundering in complex schemes, and cut false positives by up to 80%, compared with each institution working alone. Sharing critical signals in anonymised form, between firms and supervisors and between authorities, is how a supervisor protects the whole market and the consumers in it.

Where to start: the portal is step one

None of this needs a big-bang programme. It starts with a submission portal that validates data the moment it lands, on a structured data model, with rules the supervisor owns. Once that exists, firms and their RegTech providers can connect to it for straight-through reporting, wherever the supervisor wants it. Email and Excel can stay during the transition, as long as every return ends up as structured, validated data.

Why validation speed is the unlock

Granular, frequent data only works if checking it is fast. When validating a large return takes hours, firms hear about errors days later, and a supervisor cannot test a new rule against a year of filings before it goes live. When it takes seconds, both become routine. In our own benchmark, a large FINREP return validated in 16 seconds, against 4.2 hours for the open-source reference processor, with 99.95% agreement across 417,531 checks. That speed is what turns real-time supervision from an ambition into an operating model.

The foundations take a few years to build, which is why the time to start is now. See how regulatory data collection works on the Line4 platform, or score your readiness in three minutes.