I have reviewed the AI pilot portfolios of two of the largest pharmaceutical companies in the world. Hundreds of pilots at each, spanning medical coding, drug discovery, patient services, compliance, and nearly every other part of the enterprise. Remove the company names and you could not tell the portfolios apart. The same use cases. The same provider names. Sometimes four or five AI partners contracted within a single business function, with similar pilots duplicated across business units of the same company. Most were created under a principle of autonomy over alignment. That is often the right way to spark experimentation. It is the wrong way to run year three of it.
The people who created that momentum deserve real credit. They saw the possibilities, fought for resources, built the pilots, and proved that the capabilities could matter. But the people best positioned to imagine and prove a capability should not be expected to scale it alone. At that point, the work has to widen. Business, technology, operations, compliance, sourcing, legal, and risk all need a place in the decisions. The innovation belongs to the people closest to the opportunity. Scaling it belongs to the enterprise.
This is not a fresh take, everyone is commenting on experimentation having a season. I have lived through these, at least one per decade. The patterns are the same, but these stakes are higher. Leading your organization through the change, that is the strategy.
Takeaway
Your list of AI pilots is not your differentiation. Your industry peers are holding nearly the same list. Differentiation begins with what happens next. Which pilots do you scale? Which do you stop? Where should capabilities be consolidated? Which provider dependencies have become material? Do the contracts and governance reflect the maturity and consequences of the deployment? That is the discipline organizations tend to fund too late: a portfolio view, honest stop and scale decisions, contract and dependency terms someone has actually stress tested, and governance that matches what the technology has become.