Strategic Foresight & Scenario Planning

A sound decision today can become the wrong architecture tomorrow. An institution may commit significant capital to owned AI infrastructure only to find that changes in regulation, economics, or technology alter the need for it. Another may build its control model around a strategic provider only to find that provider acquired, restricted, disrupted, or no longer acceptable to the institution.

Neither decision need have been wrong when it was made. The vulnerability lies in designing a strategy that works only while its original assumptions remain true.

The AI Control Assessment™ establishes the direction. Scenario Planning tests its resilience.

Scenario planning is not forecasting. It is not prediction.

The Oxford Scenario Planning Approach (OSPA) provides a disciplined method for constructing a small number of plausible future operating environments and using them to test decisions before those decisions become difficult or expensive to reverse.

Applied to institutional AI, it asks four questions:

1. What future are we implicitly planning for?
2. What breaks if that future does not arrive?
3. Where must control, resilience, or optionality be built in now?
4. What should we build, rent, compose, partner on, preserve, or exit as conditions change?

The objective is not to choose the correct future.

It is to make better decisions across several plausible ones.

We begin with the decision that matters.

Where should the institution own? Where can it depend on others? What capital should be committed? Which capabilities must remain portable? Which decisions would be difficult to reverse?

Scenario planning begins with the institutional choice — not with speculation about the future.

We identify the external forces capable of changing the answer: regulation, model capability, open weights, compute economics, geopolitical restrictions, vendor concentration, infrastructure availability, market structure, and other forces specific to the institution.

The focus is on uncertainties that are both consequential and outside management's control.

We combine those uncertainties into a small number of coherent future operating environments.

These are not forecasts. Each scenario is deliberately plausible and sufficiently different to expose assumptions that would remain hidden in a single-base-case strategy.

The institution's chosen direction is tested against every scenario.

• What remains sound?
• What becomes vulnerable?
• What fails?
• Where is the institution locked in?
• Where does it retain options?

The result is a clearer distinction between decisions that are resilient and decisions that depend on one version of the future being right.

Scenario planning becomes useful only when it changes what the institution does today. We identify:

• No-regret moves — actions that make sense across the scenarios.
• Strategic options — capabilities or choices preserved until uncertainty resolves.
• Decision triggers — observable signals that conditions are changing.
• Decision rights — who acts, and what changes, when a trigger is reached.

The result is not a set of scenarios. It is a strategy designed to move when the world does.

Institutional AI applies Scenario Planning as the strategic extension of the AI Control Assessment™. The Assessment establishes the institution's demonstrable control position — deep across the AI stack and wide across the delegation chain.

Scenario Planning tests the strategic choices that follow from that position against multiple plausible futures.

Together, they answer two different questions:

1. Where do we stand now?
2. Will what we do next still make sense if the world changes?

 A Scenario Set

Three to four plausible future operating environments, built around the forces most capable of changing the institution's AI strategy.

The institution's strategic direction tested across every scenario — showing what remains resilient, what becomes vulnerable, and what may require redesign.

Choices across Build, Rent, Compose, partner, preserve, and exit — structured around the conditions under which each becomes appropriate.

Actions the institution can take now because they strengthen its position across multiple futures rather than depending on one forecast.

Observable indicators tied to defined management actions, allowing the institution to respond as conditions change rather than after they have changed.

A coherent account of the uncertainties considered, the alternatives tested, the decisions taken, and the conditions under which those decisions should be revisited.

The institution is no longer dependent on one assumed AI future being right.

Large, difficult-to-reverse investments are tested before commitment, while optionality is preserved where uncertainty remains high.

Provider concentration, infrastructure reliance, portability constraints, and other strategic dependencies are surfaced before they become points of failure.

Management knows which signals matter, what they mean, and what actions have already been agreed when conditions change.

The institution does not merely establish control over today's AI environment. It designs the ability to retain control as that environment changes.

Control should not depend on predicting what comes next. It should be designed to survive it. Bring strategic foresight to your AI control program.

Schedule a confidential conversation.