This article is part of the How Do YOU CM2? blog series in collaboration with the Institute for Process Excellence (IpX). Although I receive compensation for writing this series, I stand behind its content. I will continue to create and publish high-quality articles that I can fully endorse. Enjoy this new series, and please share your thoughts!
CM2 states: “A dataset cannot be used until it has been released and cannot be released until it has been properly validated.”
Simple enough. But here is the problem nobody talks about. Validation is supposed to ensure that a dataset is clear, concise, and valid for all its users. In practice, the creator validates against their intentions. The designated user validates against their needs. And the dozen other downstream users who also depend on that dataset? They find the problems after release.
This is not a people problem. It is a scale problem. No single designated user can realistically represent every downstream use case.
The use case: AI supports both sides of CM2’s co-ownership model in distinct ways. On the creator side, AI supports technical review by checking the dataset against higher-level requirements, flagging inconsistencies with related baselines, and verifying that the content meets the requirements. On the user side, AI runs persona-based reviews, simulating different downstream users of the documentation to identify where the dataset is unclear, ambiguous, or incomplete for specific use cases.
What CM2 provides: CM2’s validation and release record standards define the co-ownership model that makes this work. Each dataset is co-owned by a creator and one or more users. The designated user must coordinate with other users to make the dataset “free of difficulty for all users.”
This co-ownership structure defines the AI’s operating boundaries. The creator’s AI review stays within the technical domain. The user’s AI review stays within the usability domain. CM2 defines who owns what, so AI knows what to check and from which perspective.
The human role: The creator reviews AI’s technical findings and decides what to address. The designated user reviews the persona-based findings and determines which downstream concerns are valid. Both humans retain ownership of their validation responsibility. AI does not validate. AI helps humans validate more thoroughly.
The critical distinction: the creator still signs as creator. The designated user still signs as user. AI expanded their field of vision, but it did not replace their judgment or their accountability.
The CM2 role: Without CM2’s co-ownership model, AI-assisted review becomes unfocused. Which perspective matters? Who resolves conflicts between the creator’s intent and a downstream user’s needs? CM2 answers these questions structurally. The framework ensures that AI-generated feedback routes to the right owner through a governed process, not just a comment thread.
How many downstream users does your validation process actually represent? Or is “free of difficulty for all users” still just an aspiration?