Leaders can't afford to take a "wait and see" approach to adopting generative AI. They need a plan for applying it differently than others in the value chain, say the authors. In this article they introduce a framework for thinking about gen AI strategically and offer practical advice on how to apply gen AI to the tasks composing jobs. The framework focuses on two factors: the cost of errors and the type of knowledge required. If an error in carrying out a task would lead to serious harm, financial loss, or reputational damage, firms must be cautious about employing gen AI to perform it without human oversight. Tasks that rely on explicit data (information that can be captured and processed) are well suited for gen AI. But other tasks are fundamentally harder for it to perform because they involve not just retrieving information but also applying tacit knowledge: empathy, ethical reasoning, intuition, and contextual judgment. Placing the tasks in the appropriate quadrant makes it clear which ones gen AI can handle faster, cheaper, or better.