Change management: indispensable, omnipresent, and on the engineering department’s popularity scale right behind budget cuts and Monday-morning meetings. But seriously, why is that? It’s in the nature of us engineers to change and improve things.
The change itself isn’t the problem either.
The impact analysis problem
A classic example: impact analysis, the necessary companion of every change. Which parts are affected, which documents, which requirements? What is the current state in the first place, in PLM, in ALM, in ERP?
In practice, this question often ends with having to ask the one colleague who has always known. The one who happens to be on vacation. And before hacking your way through that jungle, you might prefer to just leave the change as it is.
The vision: request a change, get the answer in seconds
If I could wish for anything, it could already work like this today:
Me: “Computer, part X has to be replaced by Y because it’s no longer available. What does that mean?”
Chatbot: “Requirement A47 will then no longer be fully met, because of Y’s reduced stability. And the assembly workstation has to be adapted, since it needs a different tool.”
Follow-up activities could then be triggered in a similar way.
What it really takes
This still sounds like a vision, but technologically we are much closer than some think. Provided the fundamentals are right. No AI model can save an impact analysis if the data underneath is wrong. Three things are decisive:
- A Digital Thread that makes dependencies between requirements, design, and software visible
- Clean versioning, so it’s clear which state is the valid one
- An information architecture that defines what lives where and what is connected
By the way, these three points are timeless. They don’t just contribute to the feasibility of AI chatbots; they are the foundation of any working impact analysis, with or without AI.
Change management doesn’t have an acceptance problem because engineers shy away from changes, but because the impact analysis behind it often runs on hallway talk instead of data. Digital Thread, clean versioning, and a clear information architecture are the foundation for answering the question “what does this change mean?” in seconds instead of days, with or without a chatbot in front.