CBA Redesigns Work for AI
CBA is shifting from job replacement to task redesign, a signal for how large enterprises will adopt AI.
Most companies think AI replaces jobs.
The real shift is that work itself is being dismantled into tasks.
CBA is preparing for AI by mapping tasks and skills, not targeting roles for reduction.
The CEO publicly acknowledged AI disruption will be uneven and uncertain in pace.
Competitive advantage will hinge on workforce architecture, not model access.
Commonwealth Bank is signalling something important about how large enterprises are approaching artificial intelligence.
Not automation targets.
Not headcount reduction.
Work redesign.
In a national interview this week, CEO Matt Comyn outlined how the bank is preparing for AI-driven change. The emphasis was not on replacing roles, but on decomposing them.
CBA is mapping:
the tasks inside each role
the skills attached to those tasks
adjacent roles employees could transition into
structured internal mobility pathways
Comyn described the goal as creating “transparency and opportunity” so employees can see how their work evolves as AI changes workflows.
This sounds procedural.
It isn’t.
It signals that a major regulated bank is treating AI adoption as an organisational design challenge.
The unit of change is no longer the job.
It is the task stack.
Why this matters now
Across industries, executive teams are under pressure to show measurable returns from AI investments.
The most common starting point is headcount:
Which roles can we automate?
How many staff can we reduce?
How quickly will savings appear?
But large regulated organisations operate under a different constraint structure.
In the same interview, Comyn emphasised that humans remain accountable for decisions inside the company. He pointed to functions requiring judgment, contextual interpretation, and social awareness as “inherently very human.”
That distinction is not philosophical.
It’s structural.
AI can scale execution.
Responsibility remains human.
And in regulated sectors, that responsibility cannot be delegated to software.
The uneven nature of disruption
Comyn was also explicit about uncertainty.
“The impact will be uneven,” he said, noting that the pace of change remains difficult to forecast.
That single admission cuts through much of the AI hype cycle.
Workforce disruption will not unfold evenly across departments.
It will fracture along task boundaries.
Some tasks will compress rapidly under automation.
Others will persist because they embed judgment, liability, or trust.
Organisations that assume linear, role-level replacement will misread the shift.
Those that analyse work at the task layer will see it earlier.
From job titles to task architecture
This shift from roles to tasks isn’t theoretical.
It reflects a broader pattern emerging across large enterprises.
When work is decomposed into task graphs, leadership can classify each component by:
automation suitability
judgment requirement
liability exposure
training transferability
From there, roles can be rebuilt.
Some become narrower and more technical.
Others become more relational and oversight-focused.
Many become hybrid.
This approach changes the conversation from:
“How many people can we replace?”
to:
“How is work actually constructed?”
I argued last year that generative AI would reshape jobs by reshaping tasks first. What we are now seeing inside large institutions is that logic being operationalised.
The signal inside the $90 million program
Alongside the interview, CBA announced a $90 million workforce preparation initiative tied to AI transition.
The significance lies less in the number than in the allocation.
The investment is directed toward:
skill development
career pathway visibility
internal mobility systems
workforce transition support
In other words, adaptability.
Comyn framed the broader challenge succinctly: organisations, and countries, “have to get really good at adopting this technology.”
Adoption, in this context, does not mean deploying software.
It means rewiring how work flows through the enterprise.
For leaders actively running AI deployments, the rest of this briefing outlines:
the three workforce models emerging in enterprise AI
the early warning signs your rollout is targeting the wrong work layer
the operational metrics boards should demand before approving AI-linked reductions
The three enterprise AI workforce models
Across large organisations deploying generative AI, three structural approaches are emerging.
