The AI Matrix Was Never a Snapshot

Where this began

A few weeks ago, I argued that AI doesn't fail to scale because of models. It fails because decisions aren't designed to work together. The constraint is structural, not technical.

To make the point, I introduced a matrix that maps the role of AI in decision-making - automate, support, make - against the scope of decisions involved, from single-function to cross-functional. That matrix is where this follow-up begins.

What I underweighted

I treated the matrix as a snapshot.

It is, in fact, a trajectory.

A direction enterprises are moving in - whether or not they have explicitly chosen it.

Two variables that move together

As AI moves from automating tasks to making decisions, and from single-function to cross-functional, two dynamics emerge in tandem.

Value Potential Rises. More of the enterprise's economic logic - pricing, demand, supply, workforce, capital allocation - becomes addressable by coordinated optimization.

Structural Pressure Rises. Trade-offs that previously surfaced sequentially - between people, in meetings, on quarterly cycles - now surface continuously, between systems, at machine speed.

They are the same curve.

Value and structural pressure rise on the same curve. There is no version of high-value enterprise AI that does not also impose meaningful structural demands - the conditions that let AI compound value across the enterprise are the same conditions that strain a structure designed for independent, function-led decision-making.

The trajectory most enterprises are on

Investment is asymmetric.

Autonomy → Heavy Investment. More agents, more automation, more decisions embedded directly into models, faster cycle times.

Connection → Significantly Less. Structures, decision rights, and trade-off logic that allow those decisions to cohere across functions - most often deferred until problems become visible.

Each additional use case, however well designed in isolation, deepens the gap.

The danger zone

Most enterprises today, without realizing it, are accelerating toward the danger zone - investing heavily in autonomy without investing in connection.

The autonomy curve is steep. The connection curve is flat.

The space between them is where value is quietly being lost - not through weak execution of individual use cases, but through the absence of a structure capable of holding them together.

What the matrix is actually asking

Used as a trajectory, the matrix raises a different set of questions than the ones most leadership teams are asking today.

01. Where will our AI portfolio sit in 18 to 36 months - given the roadmap already in motion?

02. Is the operating model evolving at the same rate as the autonomy curve - or is the gap between them widening with each additional deployment?

03. Is the value being created compounding at the enterprise level - or is it being absorbed by coordination cost elsewhere in the system?

The strategic choice

The choice is not whether to coordinate. That choice has effectively been made the moment AI begins to influence interdependent decisions across the enterprise.

The actual choice is whether coordination is designed in advance - while the architecture is still flexible - or imposed in retrospect, once the asymmetry has compounded.

The difference between enterprises that compound AI value and those that don't will not be located in the quality of the models. It will be located in whether the operating model evolved alongside the autonomy curve - or lagged it.