Before AI enters a product strategy as part of a bet, it must clear two separate questions, not one:
A strategy that only asks the second question (can we build it) without the first (should we) risks bolting AI onto problems it doesn't fit. Because AI solutions are typically probabilistic rather than deterministic, any AI-based bet also needs explicit guardrails — defined bounds for acceptable failure modes and unacceptable outputs — spelled out as part of the bet itself, not left implicit.
This mirrors the general discipline of Product Strategy as Explicit Choice-Making: just as any problem must justify why it's worth solving before resources are committed, AI's inclusion in a bet must independently justify both its fit and its readiness. See also ai-tool-vs-substitute-strategic-thinking for AI's role in producing the strategy, as opposed to its role as content within it.