AI Skills Are Becoming a Commodity. Knowing the Work Never Will.

In The Wall Street Journal's CEO Brief, Lila MacLellan's The AI Change That Let Pfizer's Albert Bourla Sleep Better shares a candid admission from Pfizer's chief executive. Two years after rolling out generative AI across the company, Bourla was lying awake over the pace of results. AI had already taken time out of data reports and clinical trial planning, and Pfizer is starting 20 Phase 3 studies this year, double last year's number. The change was still slower than he wanted.

So he reversed a core decision. He had believed AI adoption needed to be driven from the center and led by technology experts. Over time he concluded that AI skills were becoming "commoditized," and a few months ago Pfizer moved to a distributed model. Leaders closest to each domain now make the AI decisions and own the results, with a central team in support. Resistance dropped, he said, and the pace picked up.

The Tool Was Never the Scarce Part

The article frames this as an operating model choice, centralized versus distributed. That framing is accurate and useful. Underneath it, though, is a bigger realization about what makes people valuable now. When everyone has access to the same tool, knowing how to use it stops being the differentiator. What can't be copied is the person who understands the work: the plant manager who knows which line struggles in the summer, the scientist who can tell which trial design will cost six extra months, the marketer who knows what a physician actually reads.

As the piece notes, the person who owns manufacturing is the natural choice to transform manufacturing. That sounds obvious. Yet many companies still run AI the way Pfizer first did. A central team picks the use cases, writes the playbook and rolls it out, and then leaders wonder why adoption stalls. People resist what is done to them far more than what they help design. When the person closest to the work leads the change, the change starts to fit the work instead of forcing the work to fit the change.

In 2019 I argued that individuals would increasingly define institutions, rather than the other way around. AI is accelerating that shift. The more capable the tools become, the more value moves to judgment, context and the lived knowledge each person carries. That knowledge is personal, and no central office can standardize it into a slide.

Keep the Center, Move the Ownership

Bourla didn't abandon the center, and that detail matters. Shared platforms, security standards and common training are what allow a company of Pfizer's size to move without chaos. That consistency is the floor that lets local experimentation scale. What changed is who owns the decision. The center makes progress possible. The people closest to the work make it real.

He also did something easy to overlook. He pared down his senior team, gave each leader broader responsibility and has the group meet far more often to learn what is working. Deeper in the organization, Pfizer formed what he calls clans. Engineers, for example, work in R&D, manufacturing and digital, and the clan connects them across those departments so ideas travel instead of getting stuck. That is a leader designing an environment where one person's expertise can find another's. Most organizations do the opposite. They silo the knowledge and centralize the decisions.

I wrote about a related risk in Don't Leave the Engine Room Behind: companies leaning on a small group of AI enthusiasts while the majority watches from the sidelines. Pfizer's shift is one answer to that risk. When every domain leader owns the change, AI becomes part of everyone's work instead of a specialist's project.

What Leaders Can Do This Week

Ask who owns AI in each part of your business. If the answer is the technology team, go to the leader who runs that function and ask what they would change first if the decision were theirs. Then give them the decision, along with the support to carry it out.

Find the person who knows the work best and has never been asked about AI. Their questions are often worth more than a vendor's answers. As I argued in Recompose the Role Around the Person, Not Just the Work, good redesign starts with knowing what each person uniquely brings.

Build one bridge across your silos. Choose a discipline that lives in several departments and bring those people together once a month to trade what is working and what isn't.

What I respect most in this story is Bourla's honesty. He doesn't claim to have cracked the code, only that he made choices he hopes are right. That is conviction at work, the earned capacity to act on what you believe before the outcome is certain, paired with the humility to change course when the evidence calls for it. Plenty of leaders would rather defend the original plan than admit it needed revising. The ones who move fastest are willing to revise it, and to hand the next decision to the people who know the work best.

When everyone has the same tools, the advantage belongs to the people who know the work and are trusted to lead it.

Want to explore these ideas further? Learn more about my work on leadership, identity and conviction at www.theglennllopis.com.