Share with your CISO
The DXP category is dissolving, with 13 of 28 tracked CMS vendors changing their positioning since 2023, and “agentic” branding rising from one vendor to ten. But the governance gap hiding inside that rebrand is what demands attention. A documentation review of 10 vendors finds only three (Optimizely, Acquia, Kontent.ai) give AI agents their own account and permissions. The other six run agents on the requesting user’s credentials, which collapses the audit trail and breaks the separation between who requests work and who approves it.
What this means for your business
If your content platform is mid-renewal or heading into evaluation, the vendor’s new name tells you almost nothing about their actual control model. The question that decides exposure is simpler and testable before any demo: does the agent have its own account, or does it borrow the credentials of whoever clicked “run”? Six of ten vendors reviewed here do the latter, which means the audit log attributes AI-generated content to a human, and that human is often the same person who approves it.
That second fact is the structural problem. Finance calls the safeguard the four-eyes principle, the requirement that the person who makes a change and the person who approves it are different people. When an agent operates on the requester’s permissions, all three roles (requester, author, approver) can collapse into one name in the log. Three weeks later, when legal finds a claim that was never cleared, the record can’t distinguish between what the person wrote, what the agent wrote, and what the person approved without reading. That isn’t a hypothetical failure mode; it’s the default configuration for the majority of vendors in this review.
The leading indicator to watch is whether vendors start publishing agent-specific audit log samples in their public documentation, not just asserting that audit logging exists. Acquia’s current documentation explicitly states that its Work mode provides no customer-visible record of agent activity. Sitecore and Adobe confirm audit logging without specifying that agent actions are labeled as agents. If a vendor can’t show you a sample log entry that names the agent as the actor, the governance claim is marketing, not architecture. I’d revise this assessment if any of the six proxy-model vendors ships and documents a separate agent identity layer within their current contract cycles, but right now the documentation says otherwise.
Concept deep-dive: The four-eyes principle
The four-eyes principle is a controls standard, common in finance and regulated industries, requiring that any significant action be performed by one person and independently reviewed by a second. The analogy is a bank requiring two signatures on large transfers. In a content workflow, it means the person who requests a change shouldn’t be the one who signs off on it. When an AI agent borrows the requester’s identity, the system can’t enforce that separation even if a human approval step technically exists.
Based on reporting from The DXP Label Is Fading. Agent Documentation Shows Who Governs, originally published 2026-11-13 01:56:00.

