Originally published at Forbes on Sep 03, 2026
Every CEO I know is asking some version of the same question right now: how fast can we get AI into the business? It’s an understandable question. But it’s also the wrong first question. The better question is: how much of the business is actually ready for AI?
Because AI is not just another technology to roll out. It is a stress test on the organization itself. It shows leaders where decisions are unclear, where workflows are bloated, where meetings are performative, where managers are overloaded, and where activity has been mistaken for value.
Everywhere I go, I hear the same good news: people are using AI, and they are getting faster. They are drafting, analyzing, and producing faster. That does matter. But speed alone is not transformation. I think the important questions are: What actually changed about the work? What new value did we create for the customer?
Because in too many organizations, AI is being layered on top of work that was already broken. It’s helping people create better versions of things that should not exist. It’s turning messy meetings into clean summaries without fixing the lack of alignment underneath. It’s helping teams move faster through workflows no one has had the courage to challenge. The work has not been transformed. It has been accelerated.
My head of research at Ferrazzi Greenlight, Wendy Smith, recently spoke with organizational theorist Dr. Richard Claydon, whose work is some of the sharpest I have seen on what AI is really doing to organizations. Dr. Claydon has been arguing that AI is moving faster than the organization can absorb it.
One of Dr. Claydon’s most important points is that AI is not removing management work as neatly as some people suggest. In many cases, it’s pushing more work into the part of the organization that is already least visible: the integration layer. Dr. Claydon talks about this as the “serve” layer: the layer that connects strategy to execution, translates priorities into usable work, manages handoffs and exceptions, interprets ambiguity, and keeps the system from breaking. It is the layer many executives underestimate because it doesn’t show up cleanly on an org chart.
And right now, that’s exactly where the load is landing. Executives say, “Use AI.” Teams start producing faster. But someone still has to check the quality. Someone still has to understand the context. Someone still has to decide whether the output is right, useful, risky, or misleading.
That someone is usually a manager. This is the management work many AI elimination arguments miss. They see the report. They do not see the judgment behind the report. They see the deck. They do not see the sensemaking behind the deck. They see the workflow. They do not see the human integration required to make the workflow usable across functions, geographies, customers, and cultures.
Dr. Claydon made a point in the interview that every executive should think about: if everyone can produce more reports faster, but no one has the capacity to judge quality, then the organization has not become smarter. It has become busier. That is the trap: as production gets cheaper, judgment gets more valuable. AI can create more analysis, recommendations, dashboards, and content than an organization has the capacity to absorb.
That is what Dr. Claydon refers to as the AI confusion tax. The phrase has been used in other contexts, but he uses it to describe the hidden work people take on when AI is added to already unclear work: the extra checking, the rework, the employee trying to figure out whether the AI-generated answer is useful, risky, or just confidently wrong.
AI can reduce that tax when it is applied thoughtfully. But layered onto unclear work, it can make the underlying problem worse. That’s why the real AI question is not, “Where can we apply the tool?” It is, “What work deserves to exist in the first place?”
That forces leaders to ask: Why does this process exist? Who uses the output? What decision does it support? Where are we asking people to spend hours preparing things no one acts on? Where have we mistaken the artifact for the outcome?
A report or dashboard is not really the outcome. The outcome may be a better decision. A team that knows what to do next. A leader who sees the risk clearly enough to move. Dr. Claydon’s insight is that many organizations are adopting AI before they have clearly defined the work AI is supposed to improve.
For years, companies have treated process as the answer. Improve the process, make the machine more efficient, and performance should improve. That may still matter in some domains. There are places where work is clear, repeatable, and measurable. In those places, AI can help automate, speed up, and standardize. But much of management is not that kind of work.
Management lives in the handoffs, exceptions, and competing priorities. It must address the tension between what headquarters wants and what the local market requires, and in the moments when the process says one thing but the situation demands another.
Dr. Claydon’s point is that too many organizations are using tools built for speed in environments that require sensemaking. And that is a leadership problem. The best managers have always done more than move information. They help people make sense of complexity. They see contradictions before the system breaks. They know when something sounds right but is not right. They know when a process should be followed and when the process has become the problem.
AI doesn’t eliminate that work. In fact, Dr. Claydon raised one of the most powerful possibilities in the interview: AI could become an extraordinary sensemaking tool if leaders actually knew how to use it that way. It can help teams explore multiple perspectives faster, test assumptions, translate across functions, surface contradictions, and accelerate the human process of making meaning.
This is not AI replacing managers. This is augmented intelligence strengthening managers. If AI helps a team produce a report in five minutes but no one understands the issue better, that is not progress. But if AI helps a team see the issue from five different perspectives, understand the tradeoffs, align around the decision, and move with greater trust, that is progress.
But that only happens if managers are developed for this new work. Many companies are telling people to use AI without developing the human capacity around it. They are offering prompt training when the real need is judgment training. They are measuring usage when the real need is better decisions.
“Human in the loop” has become one of the most overused phrases in AI. But Dr. Claydon asks the question underneath it: who is developing the human? Because it is not enough to put a human in the loop if that human is overloaded, undertrained, unclear on accountability, or unable to challenge the machine. It is not enough to say a manager should review the output if the manager does not know what standard they are reviewing against. And accountability means little if no one has defined what it actually looks like in AI-enabled work.
AI creates a dangerous temptation. It makes it easy to believe friction is the enemy. But friction is not always the enemy. Dr. Claydon highlighted, “If you remove all friction from an organization, you may also remove its capacity to adapt”.
Good management creates the right kind of friction. It slows the team down when the answer is too easy. It challenges the assumption everyone is rushing past. It forces the conversation that the process tried to avoid. Bad friction should be removed. But productive friction should be protected. That’s why the future of management is not fewer managers doing less. It’s better managers doing more valuable work.
Every company should be asking these three questions again and again:
That third question is the one most organizations avoid. It’s also the one that separates real transformation from another productivity program. Because if AI only helps us do more of the same, we will miss the moment. The answer is not to slow AI down. We cannot, and we should not. The answer is to speed up the organization’s ability to learn, decide, and redesign.
This means giving people permission to let go of work that no longer deserves to exist. AI is landing inside companies full of old habits, sacred meetings, legacy reports, and outdated assumptions about what good work looks like. The risk is that we automate and accelerate those things before we question them.
Dr. Claydon is right to warn that AI is moving faster than the organization. The challenge for leaders is to make sure organizational redesign can keep pace. Adoption is already happening. The real test is whether we can redesign management, judgment, and sensemaking fast enough to deserve the technology we are putting inside the business.