Robot hand being controlled by AI

Artificial Intelligence

From pilots to scale: Navigating five stages of AI maturity in procurement

Published November 16, 2026 in Artificial Intelligence • 9 min read

AI agents are moving into sourcing, yet many procurement teams still test AI without a plan. We share a five-stage maturity model to map your journey and 10 questions you should ask to determine where your organization stands.

Rapid read:

  • Scaling AI in procurement requires clear ownership, trusted data, skills, and governance. More pilots alone won’t build organizational readiness.
  • Assess progress through five maturity stages – aligned, organized, data-ready, augmented, and autonomous – developing mindset, skillset, and toolset together.
  • Match autonomy to risk. Give agents defined access rights and spending limits, retain human oversight, and validate business value before expanding their responsibilities.

 

When the German software company SAP showcased the “autonomous enterprise” at its Connect conference in October 2026, procurement had a starring role. As one illustration of how AI assistants drive value, Novartis presented a pilot of a sourcing assistant, co-developed with SAP, designed to let AI agents take over the heavy manual work of tactical sourcing. The presenters promised 25% faster cycle times and 35% higher productivity with their sourcing agent.

These are impressive predictions, but at the same time, ProcureAbility’s October 2026 survey found that only 8% of senior procurement leaders had a defined plan for AI and 68% were testing it without one. As Dirk Holbach, former chief supply chain officer of Henkel Consumer Brands, characterized the landscape earlier this year: “We are still in the hype cycle. There is value – but not yet at scale.”

The impediment to scaling AI implementation in many spheres, and demonstrably in procurement, is a lack of organizational adoption. For any AI-aided transformation, organizations need an owner, quality data, guardrails, and models that work in daily use.

We see success when organizations move through five consecutive stages, from aligned to autonomous.

Five maturity stages of deploying AI in procurement

Pivoting an organization to adopt agentic sourcing and autonomous buying is a process, not a sudden leap. This transformation requires numerous foundations, including data and risk management, defined decision guardrails, a spend data platform, enterprise resource planning (ERP) integration, and operational platforms along with strict model governance. You can drive it with a small core team, but endless pilots will not move you to scale.

Instead, we see success when organizations move through five consecutive stages, from aligned to autonomous (Figure 1). Each of the first four stages ends with a clear gate to the next.

Figure 1: Five maturity stages, from pilots to scale

Mindset, skillset, and toolset move together

Moving toward AI maturity in procurement requires steady degrees of improvement across mindset, skillset (individual and organizational), and toolset (organizational data and technology). The first three stages of this transformation deliver little more than visibility, while savings leakage, price variance, and supply risk are tackled only at Stage 4. Value can start earlier where a use case’s own prerequisites are in place, but it does not scale before then.

Figure 2: The AI maturity model for procurement

The procurement transformation with AI draws on all three dimensions, from requiring a shared vision to upskilling and ensuring the right tools across the organization. “If production planning is still done in spreadsheets and entered manually into SAP, you’re not ready to scale AI,” Luca Nanni, who ran Philip Morris International’s supply chain before becoming its chief procurement officer, has said, while describing the holistic transformation he led there. “We have redesigned our operating model and accelerated the transformation of our processes, systems, data and capabilities altogether.”

Where does your organization stand?

As you prepare to modernize and transform procurement in your own organization, ask the following questions to bring your business leaders on board. You will notice that these questions focus on organizational challenges, not the tech, as these are far more likely to be a bottleneck than the technology.

10 questions for procurement leaders

  1. Ambition: What role does AI play in your procurement organization today, and what role should it play in one to two years?
  2. Leadership: How have your chief procurement officer (CPO) and leadership teams built their own understanding of AI, and how can you create opportunities for them to augment their expertise?
  3. Value: How do you decide which AI use cases come first, and how do you make sure their value reaches the P&L?
  4. Ownership: Who owns AI in procurement, and how do procurement, IT, data, and risk teams work together on it?
  5. Skills: Which AI and data skills have you built in-house, which do you buy, and how is the role of buyers changing?
  6. Governance: Which decisions about suppliers and spend would you never leave to AI, and how do you govern the models and agents you use?
  7. Data and security: How far do you trust your spend, supplier, and contract data, and how do you keep prices and bids safe with generative AI?
  8. Scaling: Which AI applications have moved from pilot into daily work, and what separated those that scaled from those that stalled?
  9. Autonomy: Where could AI agents negotiate or buy on their own, and what would have to be true before you let them?
  10. Competitiveness: Which foundations, systems, and data sources do you need to safeguard to stay ahead of the competition?

Early-stage teams pick use cases from vendor demos and estimate their value once, while more advanced ones rank a portfolio by value and validate savings with finance. For early-stage organizations, agents are a vendor promise, while advanced ones let agents run tail-spend events within limits and widen autonomy category by category.

On governance (Question 6), MIT research sorts decisions by ambiguity and risk, and a practical rule follows: agents fit routine, reversible decisions with limited downside, while high-stakes or irreversible ones stay with people. This was echoed by SAP’s chief security officer, Marielle Ehrmann, during SAP Connect: “The more autonomously an agent operates, the tighter the oversight applied.” And SAP apparently favors tighter controls, rewriting its API policy and warning that commercial models for agent access will evolve. Meanwhile, the EU AI Act now requires that people be told when they deal with an AI system – including when a supplier’s account manager is facing a negotiation agent. In this context, build the foundations first to address data and ownership questions instead of burning money building tools you won’t be able to use.

AI and digital transformation programs

Build the skills to lead in a digital world

Transform your career and business with next-generation AI and digital skills

Explore programs

Warning signals

Signs that AI is outrunning its foundations:

  • Agents ahead of data. Pilots run on unclassified spend, duplicate supplier records, or contracts that were never digitized.
  • Bids in public chatbots. Buyers paste prices and bids into public tools because no approved tool is available, risking disclosure of confidential data and, where personal data is involved, a GDPR breach.
  • Agents without governance or guardrails. No inventory of models and agents, no strict approval limits, no clear monitoring, and no protection against instructions hidden in a supplier’s invoice or bid. “The biggest threat to companies is not malicious AI; it’s unmanaged AI,” Ehrmann warned.

Map your journey: Push for organizational learning

  • Assess where you stand. Put the 10 questions to your leadership team and place the answers on Figure 2. Pay attention to where you stand on mindset, skillset, and toolset as a team, rather than on what the last decentralized pilot achieved – a pilot no one knows how to rebuild.
  • Choose your pace instead of rushing agent pilots. Another agent pilot will not lift maturity. In procurement, scale matters and so do supplier relationships, which is why ownership questions are important.
  • Give an agent its own card. If an agent does your “shopping,” give it its own credit card with a clear limit – Visa and Mastercard already issue dedicated credentials for AI agents. Buying agents need the same: their own access rights, spending limits by category, a named owner, and no right to change a supplier’s bank details.

Leaders at different levels in your organization need to spend time with AI themselves, individually, to explore how fast new models keep evolving, gauge their capabilities, and understand their remaining limitations. If you want to harness and multiply the benefits of deploying AI agents in procurement, you also need to close the feedback loop between those who build and those who maintain these models in your organization.

Authors

Ralf Seifert

Professor of Operations Management

Ralf W. Seifert is Professor of Operations Management at IMD and co-author of The Digital Supply Chain Challenge: Breaking Through. He directs IMD’s Strategic Supply Chain Leadership (SSCL) program, which addresses both traditional supply chain strategy and implementation issues as well as digitalization trends and the impact of new technologies.

Related

Learn Brain Circuits

Join us for daily exercises focusing on issues from team building to developing an actionable sustainability plan to personal development. Go on - they only take five minutes.
 
Read more 

Explore Leadership

What makes a great leader? Do you need charisma? How do you inspire your team? Our experts offer actionable insights through first-person narratives, behind-the-scenes interviews and The Help Desk.
 
Read more

Join Membership

Log in here to join in the conversation with the I by IMD community. Your subscription grants you access to the quarterly magazine plus daily articles, videos, podcasts and learning exercises.
 
Sign up

Log in or register to enjoy the full experience

Explore first person business intelligence from top minds curated for a global executive audience