We’re a mid-sized industrial equipment manufacturer and we’ve been piloting generative AI for component design over the past eight months. The technology works—our AI system can generate optimized component alternatives that meet structural, thermal, and cost requirements in a fraction of the time. The problem we hit is governance: who actually owns these designs, who approves them, and how do we document the human involvement needed to protect IP?
Our traditional change management process was built for human engineers submitting discrete design proposals. Now we’ve got an AI generating dozens of alternatives per request, each requiring evaluation against manufacturing constraints, regulatory requirements, and cost targets. We ended up building a multi-gate review process where the requesting engineer specifies design objectives and constraints upfront, reviews AI-generated candidates, selects and refines one, then routes it through standard approval with full documentation of human decisions at each step. We also updated our IP policies to clarify that the organization owns designs created using company data and infrastructure, regardless of AI involvement.
What’s worked: embedding governance into the workflow from day one rather than bolting it on later. What’s been harder: getting engineers comfortable with the idea that they’re responsible for validating and signing off on AI recommendations, not just passively accepting them. Curious if others have tackled similar approval workflow changes and how you’ve handled the ownership question.