From implicit to governable agency.
Public procurement in the age of agentic AI.
Public procurement is already a distributed, multi-agent system. Across Europe's public institutions, thousands of decisions emerge daily from interactions between demand owners, buyers, legal advisors, finance controllers, and suppliers — interpretations, judgment calls, and institutional habits that no single person can fully reconstruct after the fact.
The dominant framing of agentic AI treats autonomous decision-making as something external – to be carefully admitted, or kept out. It implicitly assumes a baseline of human decision-making that is traceable, consistent, and governable. In most procurement organisations today, that baseline simply does not exist.
Agentic AI does not introduce this distributed agency – it makes it visible. Designed with governance as a first-order concern, it offers the first serious mechanism to make procurement decisions explicit, bounded, and auditable. The paper elaborates on
-
Why public procurement is already a multi-agent system – and what copilots leave untouched
-
The misconception shaping current AI policy debate in government procurement
-
What agentic procurement looks like in practice: bounded agents, encoded policy, preserved context
-
Five governance design principles – and concrete recommendations for governments
Download the paper to learn how agentic AI can move public procurement from implicit to governable agency — and what that means for your institution.