Leadership & Intent: Where AI Absorption Begins
Part 1 of the AI Absorption Framework series
I had dinner with a friend recently, and she described how her company rolled out Claude at the start of the year: an all-staff email, a login link, a few training docs. Essentially, a “go forth and prosper” mandate.
When I asked how she and her team were using it, she said: “Everyone's using it now, but it's mostly used as an advanced google search tool and writing assistant.”
A powerful, expensive tool, deployed with no direction, and absorbed into exactly the work people were already doing.
That’s not a technology problem. It’s a failure of intent. And it’s where AI value quietly disappears.
The promise of AI is seductive: it will boost productivity, cut costs, change everything. But the risk no one’s watching is unmanaged absorption — AI scaling faster than the organization can govern its costs, redesign the work, or turn any of it into value. Usage climbs. Spending climbs. The business looks exactly the same.
That gap doesn’t start with technology, budgets, or training. It starts at the top, with intent: the first dimension of the AI Absorption Framework, and the foundation the other three stand on. If leaders can’t answer why — clearly, honestly, and with ownership — nothing downstream can save the investment.
Getting intent right comes down to four things: a real reason, honest expectations, leaders who can actually lead it, and clear ownership of what AI does. Each is a question worth answering before you scale.
Purpose & Business Alignment: What are you actually trying to solve?
Go back to that all-staff email. What was missing wasn’t the tool, the login, or the training. It was a reason.
No one told those people what problem AI was supposed to solve, so they solved nothing in particular, and it became a writing assistant. When the goal is “use AI,” people will use it on everything and nothing, with no way to tell the difference.
This is the most common failure of intent, and the hardest to see in yourself, because it doesn’t feel like a mistake. It feels like progress. Adoption is up, people are busy, something is clearly happening. In many organizations, this is where AI adoption gets mistaken for AI transformation.
But “adopt AI” isn’t a strategy. It’s an input. A strategy names an outcome: a cost to take out, a bottleneck to clear, an experience to improve. AI is how you might get there, not the destination.
So the honest question is why you’re really doing it. Could you name the business problem in a single sentence? Or are you moving because everyone else is, and standing still feels like falling behind?
That fear is real. But fear is a terrible design brief. Organizations that start from “we can’t be left behind” buy capability and hope for value. Organizations that start with “here are the three processes costing us the most” tend to get both.
Value Realism & Assumptions: What are you expecting, and what is that based on?
AI investments rest on assumptions about the value they'll return. Those assumptions are rarely written down, so they go untested and quietly inflated.
Leaders tend to expect the payoff to be cost savings, and to see it quickly. But faster work and lower costs are two different things. Productivity shows up first: the same work gets done in less time. Savings come much later, if they come at all, and only if the organization redesigns the work to turn that freed-up time into something real.
So the question isn’t only what you expect AI to deliver. It’s when, and in what form. Are you expecting returns in two quarters or two years? And when time is freed up, does someone own turning it into value, or does it just get absorbed back into the day? This doesn’t mean AI won’t pay off. It means the payoff is slower and more conditional than the hype suggests, and the leaders who get real returns plan for that from the start.
Leadership Capability & Change Leadership: Can your leaders actually lead this?
Leading change has often allowed a certain distance. Leaders set the strategy and the goals; teams figured out how to translate that into action. That works when the path is known, when someone, somewhere, already understands how to get it done.
A 2026 Stanford study of more than 6,000 senior executives, co-authored by economist Nicholas Bloom, found that nearly 70% of CEOs, CFOs, and senior leaders use AI less than an hour a week. In a normal change, that distance from the details would be fine. With AI, it’s a problem. No one has figured out the path yet, and it can’t be handed down.
This is where leadership capability actually gets tested. Announcing an AI initiative is easy; funding one is easy. Turning it into real change — redesigning how work gets done, carrying people through the disruption — is a different skill, and far fewer leaders have it. AI is new to everyone, so where it creates value has to be worked out together: leaders and the people doing the work, figuring it out side by side rather than from a distance.
That doesn’t mean being the most fluent person in the room. But it does take enough understanding to set honest expectations, make real tradeoffs, and lead the change through.
Governance & Accountability: Who owns what AI does?
Most people would file this under “operating model.” I put it at the leadership level on purpose. The mechanics of monitoring cost may land with a CIO. But whether the organization takes accountability seriously at all, how much risk it tolerates, what happens when something goes wrong, is set at the top long before it’s written into policy. You can’t delegate the tone of accountability.
AI raises the stakes in two directions at once.
The first is cost. AI spend scales with usage, quietly, and can run away fast. Just four months into the year, Uber had reportedly burned through its entire annual AI coding budget (per TechCrunch). If a company with Uber’s resources can lose control of AI spend in the one place it’s most tightly measured, the question for everyone else isn’t “could that happen to us.” It’s “who would even notice, and whose job is it to stop it?”
The second is liability. As AI starts talking directly to customers, it makes calls that used to belong to people, and someone still owns the outcome. When Air Canada’s chatbot invented a refund policy that didn’t exist, the airline argued the chatbot was “a separate legal entity responsible for its own actions.” The tribunal called that remarkable and ruled against them: a chatbot is part of your website, and you’re accountable for what it says. You don’t get to blame the bot.
So the real test is simple: when an AI-driven decision goes wrong tomorrow, could everyone in your leadership team name who owns it? If the answer is a pause, the accountability isn’t there yet. That’s a decision to make before the tools scale, not after an incident.
Before you scale AI, sit with four questions
AI doesn’t absorb into an organization on its own. Leaders have to decide what it’s for, what they expect, whether they can lead it, and who owns what it does:
What specific business problem are we solving with AI, and could we name it in one sentence?
If you had to defend your AI business case with evidence instead of optimism, could you?
Do your leaders understand how much this will ask of them, and are they willing to give it the time and hands-on involvement it needs, rather than handing it down?
When AI makes a bad call tomorrow, do you already know who answers for it, or would you be figuring it out in the moment?
Answer those well, and you’ve built the foundation everything else stands on. Leave them unanswered, and no amount of adoption will save you from unmanaged absorption.
Next in this series: Work & Workforce — what happens to the work itself, and the people doing it, once AI enters the picture. If you’d like to initiate a conversation with me on this topic, you can reach me at lisa.taylor@taylorsteinconsulting.com.
Comments