Alex spent two weeks inventorying the AI activity already underway inside his company. He expected to come back with a recommendation leadership could act on. He came back with a spreadsheet: fourteen pilots, fourteen owners, fourteen budgets, and fourteen good reasons to keep going.
He called Marcus, a chief executive at another company who had run a transformation of his own, and said the sentence that should worry any leadership team in the same position.
“None of them are obviously wrong.”
That is what made the list dangerous. If the spreadsheet had been full of weak ideas, the decision would have been easy. These were not weak ideas. The chatbot could reduce response times. The finance model could catch anomalies earlier. The marketing experiments could speed campaign development. The search assistants could help employees find answers buried across systems. Every one could win a meeting on its own. Together, they still did not explain where the company was going.
Marcus had the spreadsheet open on his end. Rank these by return, sponsorship, or readiness, he said, and you will get an ordered list. An ordered list is not a strategy.
The Request at the Bottom of the Sheet
At the bottom sat one more line: additional funding to establish an AI center of excellence to coordinate the growing number of initiatives.
Marcus read it back slowly, then asked the only question that mattered.
“Coordinate them toward what?”
The company did not need more energy. It had energy. It did not need more ideas, and it did not need more conviction, because every sponsor already had plenty of both. What it lacked was direction strong enough to turn conviction into coordinated action.
This is the failure that is hardest to see from inside, because it does not look like failure. It looks like progress. A dashboard fills with pilots. A spreadsheet captures owners and budgets. Roadmaps appear. Status updates improve. Leaders see motion and assume the organization is maturing. What is actually happening is that the company is becoming more active without becoming more aligned. It is building plans faster than it is making choices.
A plan explains what a team intends to do. A strategy explains why that work belongs in the future the enterprise has chosen. A vision explains what that future is and why it is worth having. Without the last two, every plan becomes its own argument for attention, and the more good ideas accumulate without a shared direction, the harder it becomes to say no.
Without the ability to say no, strategy has not begun.
The Vision Statement that has Never Stopped Anyone
Before the call ended, Alex mentioned a vision statement someone had circulated the week before, and read it aloud. Become an AI-first organization that leverages advanced artificial intelligence to drive innovation and efficiency across the enterprise.
Marcus laughed. He had heard that sentence in two other boardrooms and in nearly every peer group he belonged to. It had never stopped anyone from doing anything.
That is the whole problem in one line. A statement broad enough that no sponsor could disagree with it is not a vision. It is a mood.
A usable vision describes the business rather than the technology. It identifies who will experience the difference and how. It states what the company intends to do better than its competitors. And it remains true even if every model, vendor, and platform underneath it changes.
Marcus had seen two companies get it right. A specialty insurer whose board he sat on committed that within three years, every customer question about a claim would be answered accurately the first time, at any hour, by whoever handled it, and that underwriters would spend their days exercising judgment rather than processing paperwork. An industrial distributor in his peer group was blunter: we will quote in minutes, not days, on every product, at every branch, without adding headcount to do it.
Neither sentence mentions AI. Both make AI unavoidable, because neither is achievable without it. And both immediately ended real projects that had already been funded.
“That’s the test,” Marcus told him. “Read it to your sponsors. If it doesn’t stop anything, you haven’t chosen yet.”
Four Questions, and The one that Takes Longest
Choosing requires the executives who own customer outcomes, own margin, own the workforce, and control funding to be in the same room. Not the technology function, which cannot commit the business to an outcome. Not a communications team, because language that offends no one changes no one’s behavior. Everyone who will later be asked to give something up should be present when the choice is made.
Alex put four questions on the screen. Which customer or business outcomes matter most? Which sources of advantage should AI strengthen? Which operating constraints must change? And which attractive ideas are we willing to leave behind?
The first was harder than it sounded, because the first answers named everything the company cared about, which is another way of naming nothing. The second exposed a disagreement that had gone unresolved for years: two executives believed the company’s advantage was its depth of product expertise, two others believed it was the breadth of its service network. That mattered now, because AI applied to product expertise would build a different company than AI applied to service reach. The third turned ambition into a funded constraint. The fourth took longest, because leadership had to name work it would not do.
Agreement did not arrive through consensus. It arrived through drafting. Alex asked each executive to write the vision as a single sentence about the business, without using the word AI. The sentences went up unattributed. Most were interchangeable, which was itself instructive. Two were specific enough to be wrong, and the room spent the rest of the session on those two.
Two executives still did not agree, and Alex did not treat that as failure. Unanimity was not the goal, and waiting for it would have handed the decision back to the portfolio. What the vision needed was a decision the dissenters could operate under and defend to their own teams. He asked each of them the same question: not whether they preferred the sentence, but whether they could fund it, staff it, and explain it. One could. One asked for wording that made the outcome measurable, and got it.
What a Chosen Direction Costs
Alex had asked every executive in his network that week what their AI vision actually said. Two gave him a sentence he could still repeat a week later. The rest gave him something he could not tell apart from the page on his own desk.
The difference was not eloquence. The two he remembered could be disproven. One named a service outcome a customer would notice and accepted that a bad quarter could disprove it. The other named the work the company would stop staffing by hand and absorbed the internal argument that followed rather than soften the sentence. Neither was comfortable to say out loud. Both had cost their author something to defend.
By the end of the session Alex’s leadership team had a sentence that named no technology, was not elegant, and was specific enough that at least three of the fourteen pilots clearly did not belong to it.
That was the point. A vision that costs nothing to hold will not survive its first funding argument, because it will never stop anyone from doing anything.
Adapted from Feasible, Not Achieved: Why AI Transformation Stalls and the NINE Decisions That Finish It, by Weston A. Jones, MBA and Jacob C. Roxon, PhD (JRNine Press).
The Jones-Roxon Nine™ is the Operator’s System for AI Transformation.