From experimentation to simplification: The real opportunity in agentic AI
Much of the discussion around agentic AI is focused on autonomy. How much authority should AI agents have? How can they be monitored? What governance frameworks are required before they can make decisions or take actions independently?
These are important questions. But I believe they are not the most important question.
The more important question is whether organisations are using AI to solve problems or simply to navigate around them.
Executive summary
- AI should be used to remove complexity, not hide it. Lasting value comes from addressing the root causes of operational friction rather than masking them.
- The real opportunity is simplification. AI can help identify inefficient processes, ambiguous policies and control gaps, enabling organisations to streamline and automate with greater consistency.
- AI dependency can become a new form of operational risk. Organisations must avoid replacing key-person dependency with overreliance on AI systems.
- Governance and accountability remain non-negotiable. Human oversight, clear ownership and effective controls are essential for scaling AI safely and sustainably.
- The most successful organisations will treat AI as an accelerator. Those that use AI to strengthen resilience, transparency and customer outcomes will achieve greater long-term value than those focused solely on automation.
AI in financial services: Why hiding operational complexity is not the same as solving it
Across financial services, many organisations continue to operate with the accumulated complexity of decades of growth and change. Legacy technology, fragmented processes, overlapping controls, ambiguous policies and critical knowledge concentrated in a small number of individuals all create friction for their customers and the institutions themselves. Understandably, there is a temptation to view AI as the answer.
In many cases, AI can be remarkably effective at navigating that complexity. It can locate information, interpret policies, connect disparate systems and guide users through convoluted processes.
However, there is a danger.
If used incorrectly, AI can create the illusion that complexity has been solved when it has merely been hidden:
- Technological friction can still remain
- Processes can continue to be unwieldy
- Policies can be an ongoing hidden impediment
The organisation simply becomes less aware of it because the AI has absorbed the burden.
The CRO's opportunity: Using AI to drive process automation and policy simplification
For CROs, this distinction is critical.
The real opportunity is not to use AI to compensate for organisational weaknesses. It is to use AI to identify and eliminate them in the best interests of customers and operational efficiency.
A poorly written policy is a good example. AI can help interpret ambiguous requirements and provide guidance to colleagues. But the better outcome is often to rewrite the policy, remove ambiguity and codify the rule so that it can be applied consistently and transparently.
The same principle applies to processes and controls. Where rules are clear and outcomes are deterministic, organisations should aim to codify and automate them. AI can accelerate this work by helping identify gaps, expose inconsistencies and support redesign. However, the objective should be a simpler and more deterministic operating environment, with customers at the core, not a permanent dependence on AI to interpret complexity.
The hidden risk of AI dependency: Why governance and human accountability still matter
This leads to a second challenge that deserves greater attention: organisational sovereignty.
Historically, organisations worried about key-person dependency. Certain processes, technologies or decisions could only be understood by a small number of individuals. When those individuals left, capability and resilience were put at risk. AI appears to offer a solution, but there is a risk that organisations simply replace key-person dependency with key-AI dependency.
Over time, people may become less familiar with how decisions are reached, how processes operate and where accountability resides. Human judgement can erode. The ability to challenge outcomes weakens. Organisations become increasingly dependent on capabilities they may not fully understand or control.
The dependency changes form but does not disappear.
That is why the fundamentals of risk management remain unchanged.
This is not a new discipline. It is the next evolution of automation.
Authority must still be clearly defined. Accountability must still belong to people. Controls must still be effective. Monitoring must still be continuous. Customers must stay at the heart of all thinking. Governance remains the mechanism that allows innovation to scale safely and sustainably.
In fact, the organisations that will achieve the greatest value from AI will not be those that deploy it the fastest. They will be the organisations that learn to trust it, govern it and apply it to the right problems.
The winners will recognise that AI is not a cure-all. It is an accelerator.
Used well, it can help expose complexity, simplify operating models, codify knowledge and strengthen decision-making. Used poorly, it can simply automate around underlying weaknesses and make them harder to see.
As agentic AI continues to evolve, the objective should not be to maximise autonomy for its own sake. The objective should be to build organisations that are simpler, more transparent, more resilient, more customer centric and easier to govern.
The organisations that succeed will not be those with the most AI agents.
They will be those that use AI to remove complexity rather than manage it.
Explore AI’s adoption in risk management at the largest gathering of global risk leaders: RiskMinds.
