Last year, I drew the org chart of the risk team I would hire if I were building from zero. I drew four boxes. Fifteen years earlier, when I first joined a risk team inside an Indonesian bank, the chart I worked under had eleven boxes — analysts feeding into officers feeding into heads feeding into a Chief Risk Officer, with parallel ladders for credit, operational, market, and compliance risk. The shape of the function was a pyramid. The shape I was now drawing was a flat trapezoid.
The pyramid is collapsing — not from above, where most of the attention goes, but from the middle. And the people working their way up through it are the ones who will feel the shift first.
The middle is going first
The analyst tier, the layer where most of us learned this craft, is the one being automated first. Risk dashboards that used to take a team a week to compile now generate themselves overnight. Reconciliation work that used to require three people now requires a workflow. The young analyst who used to spend three years learning the function by doing the compilation is no longer being hired into the seat that taught them.
Two roles that did not used to exist are quietly emerging in the space the analyst tier used to occupy. The first is the model overseer — someone who can read what a model is doing, validate its outputs against domain knowledge, and produce the evidence a regulator will accept. The second is the cross-functional translator — someone who can sit between engineering, risk, and product, and arbitrate between the languages each side speaks.
The Chief Risk Officer role is changing too. Three years ago the CRO’s job was largely to defend the function inside the organisation — to argue for risk constraints against revenue pressure, and to package the function’s work for the board. Today the CRO is increasingly an integrator. They must understand how AI models touching credit, fraud, and compliance interact with each other, and how the firm’s risk posture shifts as each of those models is retrained or replaced.
Also Read: The future is full of humans working with humans, AI systems and other technologies
What I should have changed sooner
I did not flatten the teams I was hiring for soon enough. I kept hiring analyst-tier roles into 2023 because the previous pyramid was familiar, because I worried about losing the apprenticeship pipeline the analyst tier represented, and because the alternative team shape was still genuinely uncertain. Twelve months of those hires turned out to be redundant within two years — not because the people were not capable, but because the work they were hired to do had already been automated underneath them. I should have spent that budget on the model overseer roles that turned out to matter.
The CRO pipeline problem
The flattening creates a new problem nobody in the industry is solving yet. The traditional pyramid was, among other things, a training apparatus. Analysts grew into officers, officers grew into heads, heads grew into chiefs — and each layer taught the person above it some of what the next layer needed to do.
Without that ladder, the function depends entirely on lateral hires. The supply of people who have already learned to be a CRO without going through the pyramid is small, and most of them are sitting in their current jobs at large institutions that built them. The next generation of CROs in ASEAN will either be poached or invented. There is no obvious third path yet.
What good team design looks like now
The teams I am helping build today look nothing like the pyramid I came up through. The shape is closer to a small, senior cell than a hierarchy. Two or three model overseers. One or two cross-functional translators. A small, very senior decision layer at the top. Apprenticeship happens through rotation rather than ladder — people move between the model side, the policy side, and the engineering side, picking up the domain by exposure rather than by promotion.
Also Read: Hiring an AI-fluent junior is easy, building one with judgment is the problem
The teams are smaller. The seniority per head is higher. The compensation envelope per role is bigger. And the work done by each seat is more cross-domain than any single seat used to require.
The shape you draw next matters more than the tools you buy
If you are building or reshaping a risk function in 2026, the question is not how to digitise the pyramid you have. The question is whether the pyramid was ever the right shape for the work you now need to do. The teams that will sit inside ASEAN’s regulated institutions five years from now look nothing like the teams that staffed them when most of us learned this craft.
That gap will not close by adding tooling on top of the old structure. It will close by drawing a different shape, and being willing to hire against it before the rest of the industry catches up.
—
Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.
The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.
Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.
The post The flattening: How AI is collapsing the middle of the risk function appeared first on e27.