Course Catalog
AI Strategy
94-804
Units: 6
Description
AI strategy is not the task of deciding where an organization can use AI. It is the task of deciding where AI should change a consequential decision or workflow, how that intervention creates and captures value, what the organization must control, and what evidence should cause leaders to scale, wait, redesign, or stop. The course therefore begins with problem selection and value capture rather than with models, tools, or use-case catalogs.
Across eleven lectures, students move through a cumulative strategy arc: problem diagnosis, value creation and capture, learning advantage, firm boundaries, infrastructure and standardization, cognitive capital, production risk, strategic dependency, institutional context, capital allocation, and strategic coherence. Classical strategy enters at the point of use. Rumelt, Five Forces, ecosystems, network effects and switching costs, transaction cost economics, Value Chain and Wardley Mapping, VRIO, dynamic capabilities, real-options intuition, and scenario planning are used to answer specific AI decisions and are stress-tested where AI economics strain their assumptions.
The Real AI provides the conceptual anchor for the course. Its ideas about infrastructure and cognitive capital become especially important once an AI system enters production. Students examine when repeated use actually produces a learning advantage, which capabilities should be owned or rented, what should be standardized or protected from standardization, how expert interventions and outcomes become durable organizational experience, and when accumulated experience becomes difficult for competitors to reproduce.
The course also treats strategy as contingent on operating reality. Students analyze how failure and justified reliance change the automation boundary, how model, cloud, compute, vendor, and jurisdiction dependencies constrain future choices, and why the same AI capability can warrant different decisions in high-compliance, operations-driven, public-sector, and investor settings. The objective is not to produce a context-free AI playbook, but to identify the binding constraint that governs a particular organization and decision.
Six Friday StratLabs turn the lecture arc into a cumulative project. Teams repeatedly make a decision, receive a mid-session injection that changes the facts, revise their position, and produce a named artifact that becomes part of the final strategy. The sequence begins with Canvas 1 — Strategic Opportunity & Value Capture, then moves through a Learning Advantage Hypothesis, AI Architecture and Firm Boundary Map, Cognitive Capital Architecture, Authority and Dependency Map, and Final Revision Ledger. The final Friday is an open strategy clinic before executive defense.
The course culminates in capital allocation and coherence. Students must decide what to fund now, what to learn about before committing, what to postpone, what to shrink, and what to kill. Their final recommendation must show that the market position, architecture, learning system, human authority, dependency posture, and investment roadmap reinforce one another. The central question of the course is therefore: what makes an AI strategy valuable, defensible, operable, adaptable, and coherent for this particular organization?
Learning Outcomes
- Diagnose Problems Before Choosing AI: Identify recurring decision pain, size the opportunity, test non-AI alternatives, and distinguish documented facts, inferences, unknowns, and validation paths before proposing an AI intervention.
- Explain How AI Captures Value: Analyze industry structure, complements, ecosystems, switching costs, and AI-stack position to distinguish value creation from value capture and test whether claimed advantage can persist.
- Design Learning That Becomes Advantage: Determine when repeated AI use creates genuine learning effects, and design feedback, outcome, adjudication, and expertise mechanisms that convert deployment experience into cumulative organizational capability.
- Choose Boundaries and Build Infrastructure: Decide what to build, buy, rent, assemble, standardize, modularize, or specialize using transaction costs, lifecycle economics, reuse, strategic control, technical debt, and reversibility.
- Bound Authority Risk and Dependency: Separate failure, risk, and harm; calibrate human-machine authority and justified reliance; and convert model, cloud, vendor, compute, and jurisdiction dependencies into explicit fallback choices.
- Adapt Strategy to Institutional Context: Compare high-compliance, operations-driven, public-sector, and investor settings to identify the binding constraint, relevant definition of value, and operating conditions that change an AI strategy.
- Allocate Capital and Defend Coherence: Build a staged portfolio of AI investments, specify evidence gates and kill conditions, and defend a strategy whose market, architecture, learning, authority, dependency, and capital choices reinforce one another.
Prerequisites Description
Basic background of machine learning and AI.