From IT to HR: Why Agentic AI Is Becoming a Workforce Issue

Artificial intelligence has spent the last few years being discussed as a technology deployment problem. Organisations have focused on choosing tools, integrating models, and experimenting with pilots. Yet a new phase of AI adoption is already beginning to emerge. That phase centres around agentic systems.

Agentic AI refers to systems capable of completing complex tasks with limited supervision. Instead of simply responding to prompts, these systems can plan, reason, and act in order to achieve defined goals.

In practice this means AI is beginning to move beyond the role of assistant. Increasingly it behaves more like a digital worker that can execute multi step workflows and make contextual decisions along the way.

This shift is important because it changes how organisations think about AI. When software starts performing tasks that resemble human work, the challenge is no longer purely technical. It becomes organisational.

The management of agentic systems will not sit solely with IT. It will increasingly involve HR, operations, and leadership.

From automation to agency

To understand why agentic AI represents a shift, it helps to look at how automation has historically evolved.

Traditional automation tools such as robotic process automation follow predefined instructions. They execute tasks exactly as programmed. If the workflow changes, the automation must be redesigned.

Agentic AI systems operate differently. They can interpret goals, plan sequences of actions, and adapt behaviour based on context and feedback.

In practical terms this means an AI system can manage parts of a workflow rather than simply completing one step within it. Research suggests that this transition from passive tools to autonomous systems is one of the defining shifts in the next generation of artificial intelligence.

This capability allows agentic systems to automate forms of knowledge work that were previously considered difficult to automate. The implications extend well beyond traditional IT systems.

Why this is not just an IT issue

The introduction of agentic systems changes how work is structured inside organisations.

When employees use tools, responsibility and execution remain clearly human. When employees supervise systems that perform tasks on their behalf, the structure of work changes. People move from performing tasks to overseeing outputs.

This change affects job design, training requirements, performance management, and organisational culture. These are areas that traditionally sit within HR and operational leadership rather than IT.

Employees will need to learn how to supervise AI outputs, interpret results, and intervene when systems produce unexpected outcomes. Roles will evolve as certain processes become automated and new supervisory responsibilities emerge.

HR therefore becomes central to managing workforce transitions and maintaining psychological safety during this shift.

The management questions agentic AI creates

As agentic systems begin operating inside organisations, leadership teams will face new management questions.

One concerns accountability. When an AI agent takes an action that affects a business process or a customer outcome, who is responsible for that decision? In highly autonomous systems this question becomes increasingly complex.

Another concerns supervision. If AI agents are capable of executing significant portions of knowledge work, how many agents can one employee realistically oversee? Will workers manage one agent, several agents, or entire fleets of them?

These questions begin to resemble workforce management challenges rather than traditional software administration. They require organisational policies, governance frameworks, and leadership oversight.

Research into agentic AI governance highlights similar concerns. Autonomous systems introduce new risks around accountability, oversight, and decision transparency that organisations must address as they scale deployment.

In other words, the challenge is not simply building agents. It is managing them.

The risks of treating agents like tools

If organisations treat agentic systems as just another category of software, several risks can emerge.

Autonomous systems without clear accountability frameworks can make decisions that no one fully owns. Without governance structures, uncontrolled autonomy may lead to operational errors or inconsistent outputs.

As multiple agents begin interacting across workflows, diagnosing failures can also become more difficult. Errors may emerge from interactions between systems rather than from a single point of failure.

Industry research suggests that scaling agentic AI requires new governance structures specifically designed for autonomous systems rather than traditional application oversight.

The challenge is not avoiding agentic systems. The challenge is integrating them safely into operational structures.

The opportunity: the rise of the digital workforce

While agentic systems introduce new risks, they also open up significant opportunities for organisations.

Instead of replacing workers outright, agentic systems allow employees to scale their capacity dramatically. Individuals may supervise multiple agents performing research, analysis, drafting or operational coordination.

This creates a new model where human workers act as orchestrators of digital labour rather than executing every task themselves.

Large organisations are already beginning to explore this model. Research from Deloitte describes a future where companies manage a "silicon based workforce" alongside human teams, requiring new management frameworks and organisational thinking.

When managed correctly, agentic systems can improve operational consistency and allow organisations to handle greater workloads without linear increases in headcount.

A historical parallel

History provides a useful lens for understanding this shift.

During the early stages of the industrial revolution, the steam engine itself was impressive. Yet the deeper transformation came from what it enabled. Entire manufacturing systems, labour structures and organisational practices were redesigned around it.

Agentic AI may follow a similar trajectory.

Today’s systems are still in their infancy. However the organisational changes they enable could prove far more significant than the technology itself.

The steam engine did not simply improve work. It changed how work was organised. Agentic AI has the potential to do the same for knowledge work.

The leadership challenge

As agentic systems become more integrated into organisations, leadership must think carefully about ownership and governance.

A clear point of accountability is essential. Increasingly this responsibility may fall to a Chief AI Officer or a similar leadership role responsible for coordinating AI strategy, governance, and operational integration.

Without central ownership, agentic systems risk being deployed inconsistently across departments. This creates governance gaps and operational confusion.

This is why organisational design is becoming as important as technical capability in the age of AI.

For a deeper discussion on this topic, see our article on where AI ownership should sit in an organisation.

The next two to three years

Agentic systems are still emerging, but development is accelerating quickly.

Many organisations are currently experimenting with agents in limited roles. Over the next two to three years these systems will likely move deeper into operational workflows as capabilities mature.

Companies that treat agentic AI as infrastructure and design organisational frameworks around it will be able to scale these systems safely.

Those that treat agents as simple tools may struggle with governance, accountability, and operational complexity.

The Silmaril view

At Silmaril we believe agentic AI represents the beginning of a new operational layer inside organisations.

The challenge is not simply deploying these systems. The challenge is integrating them safely into workflows, governance frameworks, and workforce structures.

We help organisations design the operating models required to support agentic systems so that experimentation, governance and workforce transformation move together.

Agentic AI will not just change the tools organisations use. It will reshape how work itself is structured.