Over 25 years in procurement, from buyer to CPO, I led or sponsored ten major proprietary procurement systems.
Sourcing platforms. Savings trackers. Supplier quality databases. Spend analytics. Global spend integration platforms.
At the time, organisations were often flying blind on their own data. These systems were the practical answer.
AI has changed the equation. Not because procurement no longer needs systems. It does. But the economics have shifted: faster prototypes, lower cost, broader access. The question for CPOs is no longer: "Can we afford to build this?"
It is: "What should our best people do with the time AI gives back?"
The most instructive cases filled different gaps: common savings definitions, validated performance tracking, shared supplier quality data. Each created value, but required months of development and maintenance.
The most complex case was SOGEFI. We unified spend visibility across 45 plants in six months. One company truth replaced 45 local truths. The hardest part was not technology. It was governance: which coding rules survive, which local practices disappear, which supplier relationships move from local preference to global strategy.
AI can accelerate data work. It cannot make those decisions for you.
The constraint has moved. No longer mainly technical. It is managerial: business rules, ownership and adoption.
Looking back, I see three categories
1. Tools AI can now make easier to build: savings trackers, supplier quality databases, faster and more flexible than before, but only if the business rules are well designed.
2. Platforms AI can accelerate, but not lead: global spend integration, category taxonomies, multi-ERP visibility. AI can classify, ut which plant surrenders its local coding? Which manager accepts a global contract over a long-standing local relationship? These are not data problems. They are political ones.
3. Work that remains deeply human: strategic negotiation, supplier relationships, category strategy, risk judgment. AI can prepare the brief. It cannot carry the accountability.
Many procurement functions are asking the wrong question.
The issue is not how much work can be automated. It is where the best people should be redeployed.
If that time is absorbed by reporting, internal process and administrative noise, most of the opportunity is wasted. If redirected toward supplier strategy and business partnering, procurement moves to another level.
Functions that treat AI as a cost-reduction tool will improve incrementally at work already commoditising. Those that treat it as a talent redeployment decision will create a much larger advantage.
AI does not change what good procurement looks like. It changes the cost of getting there.
The real move is not to automate faster. It is to decide, deliberately, where your best people go next.
Are you using AI to accelerate administration, or to move your best people closer to suppliers, negotiations and business decisions?