Executive whitepaper by Marc Hofheinz, who has worked in SAP logistics projects for almost 30 years – the third and final part of our series on AI in SAP logistics landscapes.
The technology for autonomous agents is coming. The decisive question is not whether agents can decide autonomously, but what a company must have regulated before it lets them: Who defines what an agent is allowed to do? Who can stop it? Who answers for its decision the next morning?
This whitepaper does not describe a technology and does not replace legal advice. It presents a governance structure that enables companies to grant autonomy in SAP EWM and SAP MFS responsibly – the EU AI Act is the outer framework here, not the main subject.
Download the whitepaper for free
Executive perspective for CIOs and those responsible for SAP logistics
Four core theses: Autonomy is a permission, not a feature
Governance per decision class instead of per agent
Chain of responsibility with four staffed roles: business owner, technical owner, approval authority, operational oversight
The EU AI Act as the outer framework – with citations and a clear separation between the wording of the law and our assessment
Two case examples from SAP EWM and SAP MFS
Four recommendations for CIOs
In a Deloitte survey, roughly one in five companies (21 percent) said they have a mature governance model for autonomous agents – yet 74 percent expect to use agentic AI at least to a moderate extent within two years.
No law closes the gap between these two figures. Only your own organization can. The whitepaper shows how.
Marc Hofheinz
Managing Partner | Qinlox Consulting GmbH
Autonomy is granted
Whether an agent acts without individual approval is not decided by the software. It is decided by the company.
Responsibility needs a name
Responsibility that lies “with the system” or “with the project” lies with no one.
The deadline is build time
The rules for high-risk systems apply from December 2027. The structure is needed for the first autonomous class well before then.
The whitepaper shows how companies anchor responsibility for AI agents in SAP landscapes so that autonomy can be granted in the first place: per decision class, with named individuals and a formal level change.