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!
Engineering changes consume 30 to 50% of engineering capacity in most organizations. Up to 30% of engineering effort is lost to rework when downstream consequences surface too late.
The uncomfortable truth? Most of this waste is preventable. The problem is not a lack of data. It is a lack of structured data and access to it.
This is where AI becomes genuinely useful, but only within the right framework.
The use case: Interactive, AI-assisted impact analysis during engineering changes. Not AI dumping a list of impacted items for a human to approve. That is just automation with a rubber stamp. Real AI-assisted impact analysis is a conversation. The engineer and the AI work through the product structure together, level by level, decision by decision.
A 2024 systematic literature review found 39 publications applying AI methods in engineering change management, with Bayesian Networks emerging as the dominant approach for change propagation analysis.
The potential is real. But potential without structure is just noise.
What CM2 provides: The CM2 closed-loop change process defines 30 operating standards governing how changes are assessed, reviewed, and implemented. This gives AI a consistent, repeatable data structure to traverse alongside the engineer.
Without CM2’s baseline discipline, AI has no reliable “before” state to compare against. Without CM2’s naming and numbering standards, AI cannot reliably identify what changed. The framework is not optional. It is the foundation that enables interactive AI analysis.
The human role: The Engineer and the AI walk through the impact together. AI surfaces a dependency and explains why it is impacted. The engineer confirms it, rejects it, or asks AI to dig deeper. Crucially, the “why” is captured at every step. Knowing that a thermal housing is impacted is not enough. Knowing that it is impacted because the changed interface increases heat dissipation beyond the housing’s rated capacity enables the CRB to make an informed decision. At each step, the human builds an understanding of the change’s reach.
The CRB owns the decision. But the interactive process ensures that the people presenting to the CRB actually understand what they are presenting.
The CM2 role: CM2 structures this interaction. It defines the baselines, relationships, and change objects (IR, CR, CN) that provide both AI and people with a shared language. Without that shared structure, an interactive session becomes two parties looking at different versions of the truth.
If a human is not part of the analysis, they cannot be accountable for the result. That is not oversight. That is theater.
How interactive is your impact analysis today? Or is someone just signing off on a list they did not help create?