A department head can support an AI initiative and still object to how it changes the work. I don’t treat those as contradictory positions. When a leader remains accountable for delivery but has little say in the new process, an executive mandate can create an operating problem before implementation begins.
When I rebuilt operating processes as COO of a regional M&A firm, clarifying roles and handoffs between departments was part of the work. I bring that operating perspective to AI efforts: department heads need more than an explanation of the decision. They need a meaningful part in shaping how it will work.
My argument is straightforward. Partnership doesn’t mean giving every department a veto. It means involving the people responsible for the work while there are still consequential decisions to make.
Bring them in before the plan hardens
I want department heads involved before the implementation plan becomes something they’re expected to defend to their teams. Asking for input after the scope, sequence, and responsibilities are fixed leaves little room for genuine participation.
At that stage, I’d make a clear distinction between the business objective and the proposed method. Leadership may have settled the outcome that matters. The workflow, review requirements, division of responsibility, and order of implementation may still need to be worked out.
That distinction gives the conversation boundaries. I can hold the objective steady while taking an operating objection seriously. I also make clear which decisions remain open and who has final authority. An invitation to contribute shouldn't imply authority that someone doesn't actually have.
Ask for operating knowledge, not endorsement
I don’t need department heads simply to tell me whether they like AI. I need their understanding of the work: where judgment matters, what information arrives incomplete, which exceptions require attention, and what their teams need from other departments.
I’d ask them to identify what a proposed change requires their people to do differently and what would make that difficult. That is more useful than asking for general support. It turns a broad reaction into specific requirements I can assess.
I also separate different kinds of concern. A disagreement about the business objective isn't the same as a workload constraint, an unclear approval rule, or a concern about losing decision authority. I need to understand the objection before deciding whether to change the plan, resolve a dependency, or explain a boundary.
Keep the handoffs in the room
My responsibility is to look beyond any department’s preferred arrangement. A change that makes one team’s task easier may leave another team checking more work, correcting missing information, or carrying an unresolved decision.
I’d bring the affected department heads together to examine those handoffs. The questions are practical: what must be complete before work moves, who accepts it, who handles exceptions, and who can stop the process when something isn't right.
This is where partnership needs executive direction. I’m not looking for a compromise that preserves every existing preference. I’m looking for a workable arrangement across the business, with responsibility that each leader understands. If priorities conflict, I make the tradeoff explicit rather than leaving teams to negotiate it during daily work.
Give participation consequences
I judge participation by whether it can affect the plan. If a department head identifies a missing prerequisite, that should receive a decision — not merely a note in a meeting summary.
Before implementation, I want the plan to state who owns each part, what has to be ready first, how the work will be evaluated, and what is outside the scope. Department heads should be able to see where their knowledge shaped those choices and understand the reasoning where it didn't.
I also include adoption in the workload discussion. Learning a revised process, reviewing work produced with AI assistance, and resolving exceptions require attention. I won't treat that attention as free simply because it doesn't appear in the software price.
Hold the direction and listen to the operation
Once work begins, I stay responsible for the sequence and for evaluating whether the change addresses the intended problem. Department heads remain partners in that evaluation, not just recipients of progress requests.
I want them to surface difficulties without having to withdraw their support for the overall effort. That requires a clear distinction between commitment to the objective and agreement with every implementation choice.
My role is to hold the business direction, make the necessary calls, and give operating knowledge a real place in those decisions. That is partnership with accountability — not a softer version of a mandate.