AI Is Changing the Decisions You Make. And How You Make Them.
Consider a hypothetical midmarket industrial distributor that has just approved its three-year growth plan. The largest commitment is a sales expansion into two adjacent regions: twelve new field sales representatives, a second branch warehouse, and the inventory to support it. The logic was sound. The company had won and kept accounts for decades through rep relationships, the margin model assumed current order sizes and pricing would carry into the new territories, and historical data showed a new rep reaching full productivity in about eighteen months.
Now suppose that within a few quarters, the conditions underneath that commitment begin to move. Purchasing teams at several key accounts start using AI tools to compare suppliers on price, availability, and lead time before a rep ever calls. Some routine reorders shift to automated purchasing, which weakens the relationship advantage on the repeat business that carried the margin. A larger competitor introduces AI-assisted quoting that answers requests within minutes, and buyers in the new regions begin to expect that speed. In this scenario, none of these shifts proves the plan wrong, yet each one makes its assumptions about rep ramp time, relationship-driven wins, and repeat-order margins less certain.
Meanwhile, the CFO produces three AI-assisted scenarios in an afternoon: proceed with the full expansion, stage it by entering one region and holding the second, or redirect part of the investment toward inside sales and digital ordering. Each model is clean and internally consistent. None of them settles the question.
The original investment may still be sound. What has changed is how much confidence its assumptions deserve. Sales leadership defends the full expansion because the territory plan carries its name, while finance quietly argues for holding the second region. The CEO has to decide whether to continue, stage, or redirect, and the harder discipline is keeping new information separate from sufficient reason to reverse a decision the company has already approved. The choice also shapes what remains possible later. Once the reps are hired, the lease is signed, and the inventory is placed, moving that capital toward inside sales or digital ordering becomes slower, more expensive, and in some cases impractical. Approving the current commitment also means deciding which growth paths the company is willing to give up.
That tension comes from two developments running at the same time. AI is changing some of the business conditions executive decisions depend on, and it is changing the tools executives use to evaluate those decisions. Together, they make it easier to approve a plan with confidence while the ground under it is moving.
The Assumptions Behind Your Decisions Are Moving
When you approve a plan, you are really approving a set of beliefs about how the world works: what customers want and how they buy, what competitors can afford to do, what talent and capacity will cost, and how demand will hold.
AI is changing some of those beliefs, unevenly and at different speeds depending on the industry. In a distribution business, it may show up first in how customers reorder. In a manufacturer, it may show up in how buyers specify and source parts. The unevenness is what makes it hard to track, because a plan approved on last year's assumptions can be sound in one area and already wrong in another.
A few operating examples show how this plays out:
Go-to-market. If AI changes how your buyers research, compare, and self-serve, the sales capacity model you funded last year rests on a buying process that is changing shape underneath it.
Operations and inventory. If AI changes how customers forecast and reorder, the warehouse capacity and inventory levels sized on current order volumes need a fresh look.
Pricing and channels. If customers can get acceptable terms through a different channel, such as digital ordering or automated purchasing, the pricing and margin assumptions behind your revenue projections become less dependable.
Investment priorities. Capital committed on three-year payback assumptions deserves scrutiny when the underlying economics can shift inside that window.
Your plan may well still be right. What matters is knowing which assumptions are most consequential to your commitments, and which of those are becoming less reliable. It is common for leadership teams to approve plans without writing those assumptions down, which makes it very hard to notice when one of them breaks.
Confidence Can Rise Faster Than Evidence
The second development offers real analytical help, with conditions attached. Some scenarios that previously required days or weeks of preparation can now be developed much faster, a proposal can be pressure-tested against alternatives nobody would have had time to model, and the strongest argument against your own plan is available for the asking. These are genuine advantages, and used well, they can make for better-informed choices.
The condition worth watching goes beyond outright errors. Polished analysis can raise confidence faster than it raises the quality of the evidence underneath it.
In the meeting itself, a set of scenarios arrives clean, consistent, and well formatted shortly after the question was asked, and it reads like diligence completed. Three different scenarios can look like a thorough examination of the alternatives, even when all three depend on the same untested estimate of customer demand. Speed also changes where scrutiny has to come from. Faster production makes deliberate source checks, assumption reviews, and challenge more important, because when analysis can be produced that quickly, those steps happen only if leadership builds them in. Otherwise the first serious questions arrive at the decision meeting itself, and the leadership team can leave more certain than when it arrived while the evidence beneath the decision is roughly what it was before.
The questions worth asking of AI-assisted work are the ones a good CFO has always asked of any analysis: where the data came from, how current it is, which claims actually carry the decision, and whether the alternatives presented are credible or simply attractive. The meetings that concern me most are the ones where nobody in the room can say which single estimate the whole recommendation depends on.
Where Analysis Ends and Judgment Begins
This is where the two developments meet. AI can materially improve difficult analytical work: modeling variables that interact, testing sensitivity across scenarios, and surfacing alternatives a team under time pressure would miss. That work is hard, and doing it better matters. What stays with leadership is settling what the organization is actually deciding, which alternatives it is willing to reject, and which tradeoffs it can live with.
Those questions carry the weight of the decision. Over three decades in executive leadership, I have seen consequential decisions falter because assumptions were never surfaced, tradeoffs were never owned, or the commitment was never clear enough to survive pressure.
AI can make the analysis behind a decision substantially better. Accepting the tradeoffs and carrying the accountability for the choice remain executive work.
That has a practical consequence. As analysis becomes easier to produce, the volume of recommendations reaching your leadership team can rise, and with it the number of plausible paths to choose between. Leadership teams can then find themselves reopening the same decision each time another credible alternative appears. More analysis is useful, but a new scenario alone is not sufficient reason to reverse an existing commitment. The scarce work shifts toward framing: which question actually matters, which assumptions the commitment depends on, and what evidence would justify changing course.
Accountability Did Not Move
A consequential decision requires an owner, whether the supporting analysis came from your finance team, from consultants, or from a model. No tool takes on the responsibility for committing the organization's resources, and if a recommendation turns out to rest on a broken assumption, the consequences land on the business and on the leadership team that committed to it.
This is where decision discipline earns its keep: clarify what is actually being decided, surface the tradeoffs honestly, align ownership of the choice with the authority to make it, translate the decision's downstream implications into execution, and then either hold the commitment or deliberately reassess it when material conditions change. That discipline is what I call Decision Governance, and AI makes it more valuable because conditions can move faster and the inputs can arrive with more confidence than they have earned.
What This Looks Like in Practice
The habits that matter here are easy to skip when a plan already looks well supported, and the first is to start with the assumptions before the commitment is made. In the distribution example, the plan depended on rep ramp time, the share of business won through relationships, and the margin on repeat orders. Get beliefs like these on paper while the decision is still open, then ask plainly how AI could change each one over the life of the commitment. Knowing which assumptions are holding and which are moving tends to change how the decision is structured. Sometimes that means staging an investment rather than committing it all at once. Sometimes the plan holds, and the right response is closer attention to a few numbers with no new decision required.
When AI-assisted analysis supports a recommendation, treat it the way you would treat work from a capable new analyst: useful, fast, and not yet trusted on the points that matter. Check the sources behind the two or three claims the decision actually depends on, and ask what the analysis did not have access to.
Before anything significant is approved, name one person accountable for the decision, with authority that matches the responsibility. Then assign responsibility for monitoring the critical assumptions, including who will bring material changes back to the decision owner. Those roles can sit with the same person, but that should be decided deliberately. When ownership stays vague, nobody has the standing to close a question once it has been answered, and each new piece of analysis becomes another reason to revisit it.
Finally, settle the reassessment conditions at the moment of commitment, while the room is still calm. Agreeing in advance on what evidence would justify reopening the decision keeps commitments stable against ordinary noise. In the distribution example, if leadership chooses the staged expansion, that means reviewing first-region rep productivity, repeat-order margins, and changes in customer purchasing behavior against agreed thresholds before releasing the second-region investment, and naming who will bring that evidence forward before additional hiring, lease, and inventory commitments are made. The standard for reopening is information that materially changes an assumption the decision depended on, and that standard is far easier to hold if it was set before the pressure started.
The Decision Is Still Yours
AI can change the conditions your commitments rest on, expand the information available to you, and compress the time between question and answer. What it leaves unchanged is the responsibility for choosing.
The leadership teams that handle shifts like this well tend to treat new analytical capability seriously and treat their own judgment as the asset it actually is: knowing which assumptions matter, which are weakening, and what the organization is truly committing to when it decides.
The conditions will keep moving. The accountability stays with you.
David Cote is the founder of TrueNorth Strategic Advisory, an independent executive advisory firm working with CEOs, founders, and leadership teams when important issues require clear judgment and consequential decisions. After three decades in technology and executive leadership, he now works with leaders to recognize what deserves attention, determine what actually requires a decision, and bring independent judgment to the choices involved. When a consequential decision is required, Decision Governance provides the discipline for making it clear, owned, executable, and durable.