What Pfizer’s six-week journey from proof of concept to global implementation tells us about the next era of procurement transformation.
I spend a lot of time talking with procurement leaders who know their current technology architecture isn’t going to get them where they want to go.
The problem is that the alternative can feel even worse: replace the stack, clean up years of data, redesign the processes around it, manage another major change program—and wait years to see the result.
But those aren’t the only two options.
That was one of my biggest takeaways from a recent Art of Procurement conversation with Rick Raker of Pfizer and Patrick Reymann of IDC. We talked about legacy-system inertia, the rise of agentic AI, and how orchestration is changing the way enterprises can approach procurement transformation.
My advice during the conversation was simple: you don’t have to burn everything down in order to fix it.
Pfizer’s experience is a great example of what that can look like in practice.
Start with a problem you can actually solve
Pfizer didn’t begin by asking how to replace its entire procurement architecture.
One of its starting points was much more tangible: a repeatable 17-step review process for non-catalog requisitions. It was well understood, happened at significant scale, and created a clear opportunity for automation.
Pfizer automated that process through ORO Labs using agents and moved from proof of concept to global implementation in six weeks. Today, the workflow reaches 78 markets and supports roughly 250,000 transactions annually.
Rick has spent 24 years at Pfizer. As he put it during the webinar, he never expected to see something move from POC to global implementation that quickly.
The six-week timeline is impressive. But what stood out to me even more was what happened next.
“Success breeds more opportunity,” Rick said.
Once people saw what was possible, the conversation changed. Pfizer began exploring opportunities across self-service RFQs, services procurement, contingent labor, direct materials, travel and expense, and new ways to use contract and supplier data.
That’s a very different model for transformation. Instead of asking an organization to buy into a multiyear vision before it sees value, you solve something meaningful, demonstrate the outcome, and use that success to determine where to go next.
Orchestration changes where the complexity lives
There’s another important part of Pfizer’s story.
The complexity of a global enterprise doesn’t disappear because you introduce a better user experience or add AI. You still have ERPs, sourcing applications, contract systems, supplier data, policies, regulatory requirements and years of technology investment underneath it all.
What orchestration changes is who has to deal with that complexity.
Historically, we’ve put much of that burden on employees and suppliers. They’ve had to know which system to use, which buying channel applies, what information is required and what process comes next.
An orchestration layer can absorb much of that complexity and guide people based on what they’re actually trying to accomplish.
It also means the systems underneath don’t necessarily need to be replaced.
That matters enormously in an organization like Pfizer, where validated systems and controls serve an important purpose. As Rick explained, the goal isn’t to recreate capabilities that already work. It’s to complement them—making processes easier for employees and suppliers while allowing established systems to continue doing the jobs they do well.
That’s why I think orchestration can fundamentally change the economics and risk profile of procurement transformation. You can modernize the experience and introduce new capabilities without making every improvement dependent on another rip-and-replace.
AI makes that architecture even more important
Of course, no conversation about procurement transformation today gets very far without AI.
Patrick made an important point during our discussion: AI acts as a stress test for your architecture.
Enterprises have lived with fragmented data, inconsistent supplier records, point-to-point integrations and manual exception handling for years because people have learned how to compensate for those weaknesses.
Agents won’t compensate in the same way. They’ll scale whatever foundation we give them.
That doesn’t mean every organization needs to stop and fix everything before it can begin experimenting with AI. In my view, waiting for perfect data, perfect processes, or perfect architecture is another form of inertia.
It does mean we need to be deliberate.
Not every problem needs an agent. Sometimes deterministic automation is the better answer. Sometimes an agent can act autonomously. Sometimes governance, compliance or business judgment means a human needs to remain firmly in the loop.
Orchestration gives us a way to make those decisions in the context of the broader process rather than adding isolated AI capabilities wherever we can.
Transformation doesn’t have to be one big event
For procurement leaders looking at years of accumulated technology and process complexity, I think Pfizer’s experience offers a useful lesson.
Don’t let the imperfections in your current environment stop you from starting.
Find a real problem. Pick something where you can demonstrate meaningful value. Use orchestration to connect what already exists rather than assuming it all needs to be replaced. Put the appropriate controls around AI from the beginning. Then learn from the outcome and decide what comes next.
The goal isn’t another transformation program with a finish line three years from now.
It’s creating an architecture that allows procurement to keep transforming as the business, technology and opportunities around it change.
Rick said it best: success breeds more opportunity.
The faster we can turn good ideas into measurable outcomes, the faster we can start asking what else is possible.
You can watch the conversation here.
Matt Kapadia is the VP of Enterprise Solutions at ORO Labs
