A CEO once told me their company had spent roughly $20 million over three years building a custom AI product, only to watch a new general-purpose model make much of it replicable almost overnight. Their story reflects a broader shift. As capable AI becomes widely available, technical superiority alone is no longer a durable edge in healthcare.
In my new Harvard Business Review article, I draw on Roger Martin’s strategic choices framework to lay out five interconnected decisions that healthcare AI vendors and incumbents (health systems, payers, and life sciences companies) need to get right. These choices are iterative, and adjusting one usually means revisiting the others.
Strategic goals. Be honest about whether an AI investment strengthens your competitive position or simply improves efficiency. Both can be rational, but only the first is strategy. Knowing what not to pursue matters just as much.
Scope. Decide whether your value comes from a narrow, provable use case or a broader, orchestrating role. Healthcare’s regulatory and reimbursement demands reward focus, but narrow tools risk commoditization. Many successful companies start narrow and expand deliberately, once they have the credibility and capital to do so.
Moat. Identify where your defensibility actually lies and be realistic about its limits. Possibilities include proprietary data, regulatory clearance and reimbursement, distribution and trust, workflow integration, customer codevelopment, specialized proprietary technology, and a people-led approach to AI. Each has trade-offs.
Capabilities. Own what makes up your moat and build the rest on third-party and open-weight models. Too many organizations discover late that what they built is now available in foundation models. Just as important is a business model whose economics hold up over time.
Management systems. Organization, talent, and governance should fit the specific bets you’re making. Centralized AI leadership suits some companies and not others. Key decisions about where AI acts autonomously and where humans stay in the loop deserve oversight that includes voices for patients and clinicians.
The companies that win in healthcare AI will be those whose choices are internally consistent and add up to genuine, lasting advantage.
Read the full article on HBR.org