AI-Agents worden Werknemers: Wat de Opkomst van Agentic AI Betekent voor HR en Governance

The boundary between software and staff is dissolving faster than employment law, HR policy, or boardroom governance can track. When an AI agent receives a task assignment, escalates decisions, interfaces with external systems, and files a completion report — it is not behaving like a tool. It is behaving like a member of your workforce. The question is no longer whether your organisation will employ AI agents. The question is whether you are equipped to manage them, govern them, and answer for them when they get something wrong.

This is not a hypothetical inflection point. It is the operational reality of 2026 — and most leadership teams are structurally unprepared for what it demands.

Salesforce's Agentforce now serves more than 10,000 enterprise clients across financial services and logistics, with agents functioning not as productivity accessories but as autonomous units embedded in core business processes. ServiceNow has gone further still, deploying agents that resolve IT incidents from initial diagnosis through to closure — without a human hand touching the workflow. The speed advantage is real. So is the displacement pressure. When an AI agent consistently outperforms a first-line employee on resolution time, the organisational logic of that employee's role requires honest examination.

What drove this acceleration is not a single technological breakthrough but a convergence: large language models capable of multi-step reasoning, mature API ecosystems that allow agents to act across enterprise software stacks, and enterprise buyers under relentless pressure to reduce operational headcount costs. The agentic turn was commercially inevitable once the underlying capability crossed the threshold of reliability. The vendors moved fast. The governance frameworks did not.

The psychological dimension of this shift is underappreciated at the board level. McKinsey's early 2026 research establishes a meaningful distinction: employees who use AI as a passive tool report manageable disruption; employees who work alongside AI agents — entities that receive tasks, complete work, and report outcomes — report significantly elevated role uncertainty. The cognitive shift from "AI helps me" to "AI does my job" is not a communications problem solvable with a town hall. It is a structural threat to adoption velocity. Organisations that deploy agentic AI without redesigning role architecture and investing in psychological safety frameworks will generate the very resistance that undermines their AI ROI.

Business Implications

If you are a CHRO, your function is now a regulatory frontline. The EU AI Act, fully enforceable for high-risk applications from August 2026, explicitly classifies AI systems used in recruitment, selection, performance evaluation, and termination decisions as high-risk. That classification triggers mandatory conformity assessments, human oversight requirements on every significant decision, detailed audit logs, and transparency obligations toward affected employees and candidates. An agent autonomously screening CVs or analysing attendance patterns is not a grey area — it is squarely within scope. Dutch organisations deploying such systems without rigorous legal analysis face penalties reaching 3% of global annual turnover or €15 million, whichever is higher. The Dutch Data Protection Authority has named automated HR decision-making a 2026 enforcement priority. This is not future risk. It is present exposure.

If you are a CFO, the governance infrastructure required to deploy agents responsibly is a capital allocation decision that belongs in this year's budget, not next year's. The organisations getting this right — a major Dutch insurer and a mid-sized bank among them — are building internal agent handbooks: per-agent documentation specifying permitted task scope, decision authority boundaries, and human approval triggers for edge cases. This is not bureaucratic overhead. It is the compliance architecture that makes agentic AI defensible to regulators, auditors, and boards. The cost of building it now is a fraction of the cost of retrofitting it after an enforcement action.

If you are a CTO who currently owns this dossier without direct HR and Legal co-ownership, the organisational model is wrong. Agentic AI deployment decisions carry employment law implications, data protection obligations, and workforce relations consequences that sit entirely outside a technology function's mandate. The CTO can build the agent. The CTO cannot govern it alone.

The winners over the next 24 months will not be the fastest deployers. They will be the organisations that can demonstrate — to regulators, to employees, and to clients — that every agent in their workforce operates within a defined, auditable, and human-accountable framework. That capability, once built, compounds. It becomes a procurement differentiator in regulated industries, a talent retention signal for employees who want to understand the boundaries of their AI colleagues, and a regulatory moat that late movers cannot easily replicate.

ZeroForce Perspective

The Zero Human Company thesis has always contained a paradox: the closer organisations move toward full automation, the more consequential human governance decisions become. Agentic AI makes this paradox operational. Every agent you deploy is a delegation of organisational authority to a non-human actor — and delegation without accountability is not efficiency, it is liability. The boardrooms that understand this are not treating governance as the brake on transformation. They are treating it as the engine. In a regulatory environment where the EU AI Act is the floor and not the ceiling, the organisations that build accountability infrastructure into their agentic architecture from day one will hold a structural advantage that no late-mover can buy their way into. The race is not who deploys first. The race is who can account for their agents when the regulator asks.