Title

AI in private equity: Procurement as the control point

Section
Insight
Summary

Written by Andrea Zanaboni

With new tools and vendors rapidly saturating the artificial intelligence market, losing control of where and how AI enters the portfolio poses significant risk for private equity (PE) firms. AI rhetoric commonly emphasises how inaction comes with a cost – but so does unconstrained experimentation, with AI pilots that don’t resolve or evolve into scalable capability ending up as distractions. So, where and how can PE firms toe this delicate balance?

Where and how AI should enter the portfolio

In PE, there are two main requirements for AI experimentation.

  1. The first is identifying an environment where tests can be conducted under real commercial pressure, outcomes are measurable, ownership is clear, and failure is containable and reversible. We argue procurement is one of the few areas that meet these criteria, making it an ideal starting point for AI usage. Procurement directly influences levers of value creation – cost, cash flow, and supply resilience – but because most AI use cases inform decisions rather than executing them, outputs are more visible and controllable and therefore lower-risk.

  2. The second is embedding frameworks that explicitly prevent drift and value leakage. Although it is clear that the greatest risk comes from falling behind the curve, it is also true that introducing AI haphazardly is dangerous – and so experimentation must occur with a solid structure that defines and measures a tool’s effectiveness. 

Experimentation with guardrails: A framework for getting started

The most successful AI initiatives in procurement share common traits. They:

  1. Start with pain points or tangible opportunities, not tools
  2. Distinguish tactical advantages from strategic ones
  3. Operate in narrow, time-bound sandboxes with an explicit scope
  4. Clearly define success and failure with metrics such as speed or accuracy 
  5. Pair the introduction of tools with capability-building, training teams to use outputs effectively and build their insights into sourcing strategies

Every experiment should be designed to reach a conclusion with explicit criteria on whether to stop or scale activity.

Avoiding pitfalls

When teams introduce AI into procurement, we see that failure rarely stems from technical issues. Common reasons initiatives stall include:

  1. Pilot purgatory, where tools neither conclude nor scale 
  2. Vendor over-reliance, where knowledge is not transferred to internal teams
  3. Tech-first thinking, where tools precede needs
  4. Governance ambiguity, where ownership is unclear

The path to success is treating procurement AI as a strategic capability – aligning leadership, building cross-functional ownership, and planning for scale from day one.

From first use case to institutional discipline

Leading with vision is important, but effective AI implementation in PE starts with rigorous control and governance centred around tangible, measurable results. Procurement is where PE firms can firmly embed this discipline from the outset in a real commercial environment.

The next step for operating partners and portfolio leaders is not necessarily to go all-in immediately, but to first experiment small and well, learn fast, and then scale deliberately.