There is a VP of Operations sitting in a steering committee meeting sometime in late 2025. She has the budget. She has the vendor. She has the kickoff slide deck with the four-quarter rollout timeline. What she does not have is any agreed-upon answer to the question of who owns the decisions the AI tool is supposed to make faster. Or what happens when the tool's output conflicts with a department head's read of the situation. Or how her team will be measured now that execution has been partially handed to software.
Six months into the rollout, the initiative is behind schedule. The vendor is pointing at change management. Her boss is asking for adoption rate reports that nobody can agree how to define. A new training program is being proposed. And the original question, the one about decision rights and accountability, has never been answered.
A 2026 survey found 79 percent of organizations are facing significant challenges in AI adoption despite high investment. More specifically: 54 percent of C-suite executives say their AI initiative is actively tearing their company apart. Those numbers get read as a technology problem, a skills gap, or an onboarding failure. They are a reflection of something that was already broken before the rollout began.
What the Accelerator Actually Reveals
The pattern that keeps surfacing in organizations struggling with AI at the executive level is not about the technology at all. It is about what happens when you push capability into a system with pre-existing design failures. Blurry decision rights become visible at scale. Managers who were promoted for output delivery, now overseeing AI agents that handle execution, are still being measured on outputs rather than on the quality of their judgment and direction. Informal agreements that were quietly holding the org chart together stop working when the pace of decision-making changes.
Forbes published a piece this week that described it plainly: a manager promoted for results is now overseeing agents that handle execution, but still measured on outputs rather than on the quality of judgment, direction, and ownership they bring. That is a design failure. Training does not fix a design failure. A communication plan does not fix a design failure.
The question for most organizations is not whether to invest in AI. That decision is already made. The question is whether the organizational infrastructure underneath those investments was built for the speed at which decisions are now happening.
Why Leaders Keep Reaching for Change Management
The change management frame is durable because it locates the problem at the employee level. If adoption is slow, add training. If people are resistant, build a communication plan. If the steering committee is not aligned, add another layer of governance on top of the existing governance. The org chart stays intact. The measurement systems stay intact. The decision rights that were already murky stay murky.
What 54 percent of C-suite leaders are actually describing, when they say AI is tearing their company apart, is role ambiguity at scale. The informal agreements that made the old system function were precise enough for a world where every output required a human hand. They are not precise enough for a world where AI handles execution and the value of the human layer is supposed to be judgment. Nobody made that explicit before the rollout. Now the gap is expensive and visible.
The organizations that keep adding to the change management stack tend to produce a particular kind of report at the 12-month mark: adoption metrics that look acceptable, business impact that is difficult to demonstrate, and a growing tension between the teams using the tools and the leaders measuring them by standards that predate those tools by several years.
The Question That Gets Skipped
The organizations that are not struggling have generally done something before anyone signed a contract: they asked who owns the judgment calls the AI is supposed to accelerate. That sounds like a simple question. In practice it tends to surface everything the org chart was not saying clearly. Who is actually accountable when the AI output is wrong? How will managers be measured in a world where their team's execution output is partially automated? What decisions require human sign-off and which ones does the tool make autonomously?
These are governance questions, not configuration questions. They require executive-level clarity about accountability and decision rights. And because they expose existing ambiguity in the org design, most leadership teams find reasons to defer them until after the rollout is underway. By then, deferring is no longer an option. The tool is live, the friction is visible, and the conversation that should have happened in a conference room is now happening in an incident review.
The companies getting AI right are not necessarily the ones with the best tooling or the most sophisticated change management apparatus. They are the ones whose senior leaders were willing to use the rollout as a forcing function for decisions they had been avoiding. That kind of clarity is harder than buying software. It is also the only thing that actually closes the gap.
What organizational design question is your AI adoption surfacing that you haven't yet answered directly?
