Article

Agentic AI: why CIOs and CTOs need to rethink the operating model now

Heather Barnes 28 Aug 2025

cloud technology concept Global network of metaverse systems exchange of information global business Megatrends on the Internet for Telecommunication

The rise of agentic AI marks a turning point in the role of technology leaders. For years, CIOs and CTOs have delivered incremental transformation, digitising services, migrating to cloud and embedding data into decisions. Agentic AI does not offer an increment. It introduces a different paradigm altogether.

These systems do not just automate tasks. They pursue goals, make decisions, and orchestrate workflows across tools and teams with limited supervision. They are not tools that wait for instruction. They are agents that act, reason and adapt. For technology leaders, this is not just another evolution in AI capability. It is an operational, cultural, and architectural inflection point.

What agentic AI actually means

Unlike traditional AI, which relies on human prompts or predefined decision trees, agentic AI can interpret a business goal, plan how to reach it, act, monitor its impact, and iterate, sometimes across multiple systems and stakeholders. A procurement agent, for instance, might scan the supplier landscape, negotiate contracts, align to ESG targets, and loop in finance only when strategic thresholds are breached.

This shifts AI from a passive function into an autonomous actor. The CIO or CTO becomes responsible not only for its performance but also for its alignment, boundaries, and escalation logic. You are no longer just managing code. You are managing intent.

Impact on the technology function:

Architecting for autonomy: legacy systems are built around request-response logic. Agentic AI needs infrastructure that supports decision autonomy, context switching and continuous feedback loops. This will force a rethink of integration layers, orchestration tools, and how systems signal intent and constraint. CIOs will need to invest in architecture that allows AI agents to traverse functions, legal, finance, ops without fragmenting data or breaching compliance. This is less about data lakes and more about system composability and AI-native permissions.

A new governance mandate: autonomous systems can go rogue – quietly and at scale. A procurement agent making unauthorised commitments, or a customer service agent escalating issues incorrectly, can introduce reputational and financial risks before anyone is aware. CTOs will need to design in safeguards. This includes behavioural audits, escalation triggers, traceable decision logs and real-time simulation environments. It also means defining the “why” behind decisions, not just the “what.”

Redefining technology talent: with agentic AI absorbing much of the process logic and execution work, the technology function will shift toward a mix of orchestration, oversight, and design. Roles like AI product lead, prompt engineer, systems ethicist and simulation strategist will emerge as core to the operating model. Technical fluency alone will no longer be enough. CIOs will need teams who can design for ambiguity, align systems to strategic intent, and interrogate machine-led decisions in real time.

Macro of many silicon Wafer with semiconductors in box inside clean room

Strategic implications for the business:

Agentic AI will stretch far beyond the technology team. It will reshape decision rights, workflows, and accountability across the enterprise:

Distributed decision-making: AI agents can triage decisions, identify bottlenecks, and propose changes without waiting for a quarterly review cycle. This requires a mindset shift in the executive team. If the AI can spot a problem and resolve it, who owns the outcome? Who signs it off?

Collapse of traditional hierarchies: when a digital agent can escalate to a C-suite dashboard, propose a change, and A/B test the solution all within a morning, the middle layers of oversight begin to shift. CIOs and CTOs must help the organization determine what should be automated, what should remain human, and what should be a hybrid approach.

Speed vs. control: agentic AI systems can introduce strategic agility but also operational volatility. Firms that scale too quickly without the right guardrails may find themselves exposed. Technology leaders will need to strike a balance between empowerment and containment.

Inside computer, hardware motherboard components and circuits

The leadership challenge

For CIOs and CTOs, agentic AI is not a technical choice. It is a leadership test. You are no longer just overseeing systems that support business strategy. You are defining how intelligence flows through the organisation and how machines and humans make decisions together.

This will demand more than architecture diagrams. It will require a new playbook, one that combines engineering rigour with ethical foresight and cultural awareness.

Firms that treat agentic AI as a strategic capability, not just a productivity lever will redefine how value is created. Those that wait for a best practice template may find themselves playing by someone else’s rules.

Now is the moment to lead.

Heather Barnes

Heather is a Partner in Leathwaite’s global Technology, Data & AI Practice, advising boards and executive teams on the leadership needed to drive digital transformation and enterprise innovation. She appoints CIOs, CTOs, Chief Data, AI Officers, CISOs, and other senior…

See full profile