Applying AI to Strategy Without Letting It Make the Decision

The organisations staying ahead of disruption are not the ones using AI the most. They are the ones using it at the right point in the decision, and no further

Every leadership team now faces pressure to apply AI to strategic planning, and the pressure is not unreasonable. AI can process market signals, competitive data, and internal performance metrics at a scale no strategy team could match manually, surfacing patterns and risks that would otherwise stay buried in spreadsheets nobody has time to fully examine. Organisations that ignore this capability entirely are choosing to make decisions with less information than their competitors, and that choice carries real cost.

But there is a meaningful difference between AI informing strategy and AI making it, and the distinction gets lost precisely when the pressure to move fast is highest.

What AI Is Actually Good At Here

Pattern detection across large datasets is where these tools genuinely excel: identifying a shift in customer behaviour before it shows up in quarterly revenue, flagging a competitive move buried in public filings, surfacing a supply chain risk that a manual review would catch only after it had already caused disruption. This is real strategic value, and it is available to organisations of a size that could never previously afford this level of analysis.

Scenario modelling is similarly strong: running many more variations of “what if” than a human team could construct manually, giving leadership a wider view of how a decision might play out under different conditions. Used well, this expands the options a leadership team actually considers before committing, which is one of the most common failure points in strategic planning: not a bad decision, but a narrow set of options that were never seriously widened.

What Strategy Still Requires From a Person

Where this goes wrong is not in using these capabilities, but in treating their output as the strategy itself rather than an input to it. A pattern detected in data does not know what an organisation values, what risks it is willing to accept for the sake of its mission, or what commitments to its people and community constrain what it should do even when the data suggests otherwise.

Strategy, at its core, is a judgement about which future to pursue and what a business is willing to sacrifice to get there. That judgement involves values, relationships, and responsibilities that do not appear in a dataset: a commitment to employees during a downturn that a pure efficiency calculation would not recommend, a decision to serve a community a purely profitable strategy would abandon, a refusal to pursue a legal but exploitative opportunity a model would flag as attractive.

This is where the Catholic principle of subsidiarity offers real guidance: decisions should be made at the level closest to those who understand the full human context, with support from tools and information gathered more broadly, rather than deferred upward to whatever system can process the most data fastest. An AI system can inform the leader who makes this judgement. It cannot make the judgement, because it has no access to what the organisation is actually for.

The organisations navigating disruption well are not the ones deploying AI most aggressively into their strategic process. They are the ones who have been precise about where the tool’s contribution ends and human judgement, accountable and values-driven, begins.