Practical guide
AI fluency for corporate strategy: test portfolio dependencies
Compare fictional portfolio options by strategic premise, shared dependency and failure condition without inventing returns.
Corporate strategists can use AI to expose relationships across initiatives while checking every assumption against available evidence. This fictional portfolio exercise produces an option map, not an investment recommendation, valuation or forecast.
Frame the portfolio choice
A list of initiatives is not yet a strategic option.
A fictional company is considering entering one adjacent segment, strengthening its current segment or sequencing both. State the choice, horizon and decision owner. The packet supplies no budget approval or forecast.
Ask AI to group initiatives by the premise they depend on, then verify the grouping. Keep proposals separate from current commitments and do not describe an option as funded.
Write the strategic premises
Each option should make its theory of advantage inspectable.
The adjacent-segment option assumes an existing distribution relationship transfers; the current-segment option assumes retention improves when onboarding changes. Neither premise is proven by the packet. Record the evidence that would strengthen or weaken each.
Do not create market size, return or adoption figures. AI may generate questions and counterarguments, but factual premises require attributed evidence.
Map shared dependencies
Two options may compete for or rely on the same capability.
Both fictional options require the same specialist review team, while only the adjacent option depends on the distribution partner. Show the shared review capacity as a constraint and the partner as an option-specific dependency.
Do not allocate people or money on behalf of owners. Ask whether sequencing could produce evidence before the shared capacity is committed, while leaving the actual allocation undecided.
Stress a premise failure
A strategy becomes clearer when one central assumption does not hold.
Assume the distribution partner cannot support the new segment this year. The adjacent option must pause or find a different route; the current-segment option is unaffected by that dependency. Record what remains valid instead of discarding the whole analysis.
Use the counterfactual to identify decision sensitivity, not to predict failure. A second scenario can test constrained review capacity without assigning probabilities that the evidence does not support.
Deliver an option and trigger map
The artifact should show what evidence changes the choice.
Include options, premises, supporting evidence, shared and unique dependencies, failure conditions and next evidence gates. Label illustrative sequencing as a proposal and leave authorization fields blank.
Review premise traceability and portfolio interaction. Programme management plans delivery after work is authorized; this page tests which strategic path deserves further evidence before that stage.
Compare learning paths before commitment
Sequencing can be valuable because an early step tests a shared premise.
Add a fictional limited discovery phase that tests whether the distribution relationship transfers without committing to the full adjacent-segment option. Define the observation sought, boundary, owner and stopping condition. Do not call it a pilot if no organization has authorized participation, and do not attach invented conversion or return targets.
Compare that path with immediately strengthening the current segment. The first may reduce uncertainty about the partner dependency; the second may use the constrained review team sooner. Record what each sequence teaches and which future choices it preserves, leaving the portfolio owner to decide whether the evidence value merits the operational load.
Sources and scope
NIST: AI risk management. Skills England: workplace AI foundations.
These references provide background, not validation or endorsement of this exercise. The case details, calculations and suggested review questions are original instructional material. Use them to discuss observable work, not to infer customer outcomes, professional credentials or performance in every setting. Before adapting the exercise, confirm the relevant facts, approved tools, data permissions and decision owners. If you change the case, revisit the expected answers and checks as well. These examples describe practice tasks, not a promise that a particular product includes the fictional features.
Sources: [1] [2]
Sources
- 1.AI RMF Core · NIST
- 2.AI foundation skills for work benchmark · Skills England