02.11.2026
Most banks are paying a fragmentation tax, and a lot of leaders do not realise it. The money goes into separate, air-gapped modelling architectures: one for risk, another for fraud, another for capital. Each built in isolation. Each with its own team, its own assumptions, its own version of the truth.
After dozens of conversations with risk and compliance leaders on my podcast, Women Who Make It Happen, I keep noticing a shift. The industry is moving from tick-box compliance to something more data-led, and the banks that get there first will have a structural advantage over the ones still patching fifty systems by hand.
The vision of the “AI super model”
One of the sharpest versions of where this goes comes from Nidhi Agarwal, Chief Model Risk Officer at Virgin Money. She argues the current setup, where separate teams and models handle capital, underwriting and financial crime, has had its day. The future is a single pipeline that acts as one source of truth.
As she said in an episode with Sue Saunders, “eventually there may be a need or a world where there’ll be an AI super model which will be the super brain for all the core modelling.”
Pull the core modelling into one architecture and you strip out the manual handoffs and friction that slow legacy banks down.
Risk measured, not guessed
Conduct risk and compliance have spent years treated as a qualitative exercise, second to the hard numbers of credit and capital. Vicky Stubbs, Managing Director in Compliance at Barclays, says that is changing fast, pushed by expectations like the Consumer Duty.
The goal is to make conduct risk as data-driven as credit risk. You cannot understand customer outcomes from stories and isolated examples. You have to look across the whole customer journey and catch the variations from the norm before the regulator does.
“The plural of anecdote is not data.”
– Vicky Stubbs
One pipeline, fewer manual interventions
Unifying the architecture does something practical: it removes the manual handoffs between systems and the repeated annual model reviews that fragmented setups force on you. Nidhi Agarwal points to exactly this kind of repetition as the tax of staying manual.
But a single pipeline raises the stakes on oversight, which is where Fleur Hicks, a twice-exited founder, makes the case for data plus people. The numbers validate decisions, but a unified model concentrates a lot of risk in one place, so it needs human judgement reading it, not just trusting it.
The speed versus safety paradox
Banking is caught between two cultures: the rigour and discipline of traditional institutions, as Vicky Stubbs puts it, and the dog-years pace of fintech, as Claire Barratt describes it. Fintechs move fast because they are wired to get things done, but that speed bites as they scale.
Hazal Muhtar, Senior Director of Data at Wise, has lived it. She helped scale the analytics function at Wise from 50 to 300 people, and her approach treats every challenge as solvable if the data architecture underneath is strong enough.
A unified model is what makes that solvable at scale. When technology outruns policy, you update one super brain to reflect a new regulation, rather than retrofitting fifty fragmented systems.
Someone has to be able to challenge the model
A unified model is only as good as the team willing to argue with it. Ellen Watson Hicks frames psychological safety as the layer that makes that possible: people confident enough to question an output rather than defer to it.
This matters more, not less, once a single super brain sits behind your core decisions. When the model produces something that does not look right, you need someone who will put their hand up. A pipeline no one challenges is not a source of truth. It is a single point of failure.
From fragmented functions to connected intelligence
The shift from fragmented silos to unified architecture is coming. But the technology is only half of it. Whether your bank can innovate behind that architecture depends on the calibre of the people in your governance, risk and compliance functions.
That is what we do. At Morson Edge, we supply the specialist GRC talent that steers these models and turns rigour into a competitive advantage. You can see how we support the sector on our Morson Edge Financial Services page.
So here is the question for your next strategy session. Is your data architecture a bridge to the future, or a barrier keeping your best people stuck in the grunt work of the past?
