Procright - How to Prepare Procurement for AI: 7 Practical Steps

AI has moved past the stage where procurement leaders can treat it as a distant innovation topic. It is now an operating model question.

The real issue is no longer whether AI will affect procurement. It is how quickly procurement teams can redesign their foundations, workflows, and talent model so they can use AI productively without creating new risks. For business professionals and procurement leaders, that means shifting the conversation away from hype and toward execution.

In a recent discussion with Hexion Chief Procurement Officer Gail Dematoera, one theme stood out: procurement should not bolt AI onto broken processes. Instead, teams need to rebuild key capabilities with AI in mind. That distinction matters. Organizations that rush into tools without addressing data quality, process fragmentation, or team readiness may automate inefficiency rather than improve performance.

This article distills the most useful ideas from that conversation and expands them into a practical framework for procurement leaders who want to make measurable progress now.

Key Takeaways

Start with foundations, not tools. Poor data, weak workflows, and unclear ownership will limit AI value.
Redesign procurement as a hybrid operating model. Keep humans close to strategy and relationships while using AI for analysis, pattern detection, and routine execution.
Train the whole team, not just a digital elite. AI readiness requires broad literacy across procurement, not isolated experimentation.
Use experimentation over long-range planning. In a fast-changing market, small use cases reveal more than rigid three-year roadmaps.
Be deliberate about build vs. buy. Buy commoditized capabilities; build where differentiated insight creates strategic advantage.
Keep humans accountable. AI can recommend, summarize, and structure, but judgment and risk ownership remain human responsibilities.
Treat curiosity as a core competency. Teams that stay open to new approaches will adapt faster than teams waiting for certainty.

Procurement’s AI Moment Is Bigger Than Automation

One of the most important insights from the discussion is that procurement’s AI journey is not only about efficiency. It is also about repositioning procurement as a strategic engine.

That view challenges the older model of procurement as a policy gatekeeper or process enforcer. In more advanced organizations, procurement is expected to do far more:

connect business needs with supplier capabilities
surface innovation from the supply base
improve working capital and value creation
identify risk earlier
enable faster, better-informed decisions

AI strengthens this strategic role, but only if procurement adopts it in service of better decision-making rather than novelty.

Dematoera described procurement as deeply embedded in two worlds: the business and the supplier ecosystem. That is exactly where AI can have outsized impact. Procurement sits at the intersection of internal demand signals, market intelligence, contractual data, supply constraints, and supplier innovation. Few functions have access to such a broad decision surface.

The implication is clear: procurement may be one of the best enterprise functions in which to operationalize AI, provided it gets the basics right.

Why Foundations Come Before AI

A common failure pattern in digital transformation is assuming a new technology can compensate for weak process design. In procurement, that often looks like this:

fragmented source-to-pay workflows
duplicate or incomplete supplier data
disconnected systems
inconsistent category intelligence
manual handoffs between teams
unclear governance

AI does not fix those conditions by itself. It can even magnify them.

Dematoera’s argument was straightforward: if you build AI on unstable foundations, you end up with a broken system at greater speed. That is a useful warning for any organization tempted to launch agents or copilots before clarifying its operating environment.

The core foundations highlighted in the discussion were:

Data

Data remains the first constraint. In many procurement environments, data exists across sourcing tools, ERP platforms, supplier systems, contract repositories, and external intelligence feeds. It is often, in her words, effectively "everywhere and nowhere."

For AI to produce reliable insights, procurement needs to improve:

data quality
data accessibility
process-to-data continuity
interoperability across tools
ownership and stewardship

This is not glamorous work, but it is essential. If supplier, spend, risk, and market data cannot flow cleanly through the procurement process, AI outputs will be limited or misleading.

Process Design

Source-to-pay processes often evolve over time through patches, exceptions, and local workarounds. AI creates an opportunity to rethink those flows more fundamentally.

That does not always mean starting from scratch. In practice, it often means:

decomposing major workflows into decision points
identifying repetitive, rules-based tasks
separating judgment-heavy work from automatable work
rebuilding workflows around clean handoffs between human and machine

This is process re-engineering, not just process digitization.

Talent and Mindset

Technology capability is only part of readiness. Procurement also needs people who can work effectively in AI-enabled environments.

Dematoera emphasized that the issue is not only technical skill. It is also mindset. Teams still need classic procurement capabilities such as:

negotiation
relationship management
commercial judgment
stakeholder influence

But they also need a more adaptive posture: curiosity, experimentation, and comfort with change.

That combination is especially important because AI changes how value is created in procurement. The winning teams will not be those who simply use more software. They will be the ones who pair procurement judgment with machine-scale analysis.

The 7 Practical Steps to Prepare Procurement for AI

1. Reframe procurement as a strategic business function

Before changing tools, change the narrative.

Procurement teams that see themselves primarily as compliance enforcers will struggle to create enthusiasm around AI. Teams that see themselves as value creators will approach AI differently. They will look for ways to improve decisions, connect supply market signals to business strategy, and generate commercial advantage.

This reframing matters because AI investments compete for executive attention and budget. If procurement cannot explain how AI supports business outcomes, adoption will remain tactical.

A useful internal test is this: can your team describe its contribution in terms of innovation, resilience, speed, cost, and decision quality, rather than only process adherence?

2. Audit the foundations before selecting solutions

Most procurement organizations should perform a baseline readiness assessment before scaling AI.

That assessment should cover:

data quality and accessibility
maturity of source-to-pay processes
current digital tool landscape
workflow bottlenecks
policy and governance constraints
team capability and AI familiarity

This step helps prevent a common mistake: buying advanced AI capabilities for an environment that cannot support them operationally.

If your supplier master is unreliable, your spend taxonomy is inconsistent, or your market data cannot connect to internal forecasts, that is the real work to prioritize.

3. Redesign the operating model around human-machine collaboration

One of the most useful parts of the discussion was the idea that procurement’s future operating model will have distinct layers.

Dematoera described a shift away from a purely people-centric model toward a structure that includes:

a human layer focused on business engagement, market understanding, and supplier relationships
an AI center of excellence to develop use cases and translate procurement needs into AI-enabled solutions
a digital capability center supporting more automated, touchless, or autonomous execution over time

This matters because many AI conversations stay trapped at the tool level. Operating model design forces leaders to answer more consequential questions:

Who owns experimentation?
Who governs agent behavior?
Which tasks remain human-led?
How are successful use cases scaled?
Where does procurement need dedicated AI product ownership?

Even if an organization does not establish formal new teams immediately, it should define these roles clearly.

4. Use experimentation as the primary implementation method

A standout point from the interview was a rejection of the traditional multi-year transformation plan for AI.

That does not mean strategy is unnecessary. It means procurement AI strategy should be tested through use cases, not treated like a static ERP roadmap.

This approach is more realistic because the market is evolving too fast for rigid planning assumptions to hold. New models, tools, and integration methods appear constantly. A team that waits for certainty will move too slowly.

A better path is:

select a high-value use case
test it quickly
measure utility, risk, and adoption
scale if successful
stop and move on if it fails

This also reduces organizational resistance. It is easier to build momentum around visible wins than around abstract transformation messaging.

Good early candidates for experimentation

The video mentioned market intelligence as one example. That is a strong starting point because it combines structured and unstructured data and has a clear path to commercial value.

Other likely early use cases may include:

supplier risk monitoring
benefits tracking and reporting
contract summarization
category intelligence synthesis
intake orchestration
low-complexity sourcing support

Not all of these were specified in detail in the video, but the broader principle was: begin where insight generation is repetitive and decision support is valuable.

5. Train every procurement professional, not just specialists

One of the more practical insights was the emphasis on broad team education. Rather than limiting AI exposure to a central innovation group, Hexion’s approach described in the discussion involves training the entire procurement function.

That is significant because many organizations still treat AI literacy as optional. In reality, procurement leaders should assume that baseline AI fluency will become part of normal professional competence.

Useful training should go beyond prompt-writing tips. It should include:

how AI models work at a practical level
how to assess output quality
how to build simple agents or workflows
dashboard and app prototyping
security and governance principles
use-case identification

This matters for adoption. People are more likely to use tools they helped shape or understand. If procurement staff can build or adapt lightweight solutions for their own work, adoption tends to become more organic.

For leaders, there is another benefit: training helps reveal hidden talent. Some of the best AI builders in procurement may not come from traditional digital roles. They may be category managers, analysts, or operations professionals with strong process insight and high curiosity.

6. Make build-vs.-buy decisions based on differentiation

The procurement technology market is crowded, especially in AI. Many vendors now offer highly specific, AI-labeled capabilities, while legacy platforms are embedding AI features into broader suites.

That creates confusion, but the discussion pointed to a useful framework: buy the commodity infrastructure, build the differentiators.

Buy when the capability is becoming standardized

Examples discussed or implied include:

spend analytics platforms
core orchestration tools
established sourcing infrastructure
baseline digital workflow capabilities

If multiple vendors can provide a robust capability at scale, building internally may not deliver enough advantage.

Build when insight and context are unique to your business

Market intelligence was used as a clear example. External data can be acquired, but the real value emerges when a company combines that data with internal forecasts, supplier knowledge, category strategies, and business context.

That layered intelligence can become a competitive advantage. It is harder for off-the-shelf products to replicate.

This distinction is especially important for technical decision-makers evaluating ROI. AI investments should not be judged only by whether a tool "works." They should be judged by whether the capability is:

differentiated
scalable
governable
economically justified

7. Put governance and accountability at the center

This may be the most important step of all.

The interview made a strong case that AI should remain human-led, not merely human-reviewed. That is a meaningful distinction. "Human in the loop" can become a superficial control if people are only validating outputs after the fact. "Human in the lead" means people remain responsible for the objective, context, and final decision.

That approach is well suited to procurement because the function routinely deals with:

supplier risk
legal exposure
confidentiality
commercial trade-offs
ethics and compliance
business continuity

An agent can generate options. It should not inherit unchecked authority simply because it is fast.

A useful line from the discussion reframed the issue: organizations would not hand a brand-new graduate unrestricted access to all enterprise data, so they should not do the equivalent with AI agents either.

That principle should guide governance design in areas such as:

data permissions
workflow approvals
auditability
exception handling
model monitoring
usage limits
escalation paths

AI governance in procurement is not bureaucracy for its own sake. It is the mechanism that makes scaled adoption safe enough to sustain.

The Skills Procurement Will Need Most in an AI-Enabled Future

Not everyone in procurement needs to become a builder. That was acknowledged clearly in the conversation, and it is an important point for leaders worried about unrealistic capability expectations.

Still, the future procurement team will need a somewhat different skills profile.

Curiosity

This came up repeatedly, and for good reason. Curiosity drives experimentation, continuous learning, and openness to new methods. In a market changing this quickly, curiosity is not a personality bonus. It is an operational advantage.

Judgment

AI can organize information and generate recommendations. It does not own consequences. Procurement professionals still need to evaluate trade-offs, detect when something seems wrong, and make context-sensitive calls.

Influence

As AI accelerates analysis, the ability to align stakeholders may become more, not less, important. Faster information only creates value if organizations can act on it.

Relationship management

Supplier and stakeholder relationships remain central. AI can support preparation and insight, but trust, negotiation dynamics, and strategic partnership still depend on people.

Humility

This was an especially sharp observation. Teams experimenting with AI will make mistakes. Leaders who cannot admit that, learn quickly, and course-correct will slow progress. Humility supports both innovation and control.

What This Means for Procurement Leaders Right Now

The conversation offered an encouraging reality check: most organizations are still figuring this out. Despite the market noise, there is no universal blueprint that has already solved procurement AI at scale.

That should reduce paralysis, not increase it.

Procurement leaders do not need perfect certainty before acting. But they do need discipline. The most effective next move is usually not a sweeping declaration that procurement is now "AI-first." It is a focused step that improves capability without increasing chaos.

For many teams, that means:

cleaning up a data domain
selecting one meaningful use case
training the team on practical AI basics
defining governance guardrails
testing whether a new workflow actually improves outcomes

In other words, progress comes from structured experimentation on top of improving fundamentals.

Conclusion

AI is giving procurement a rare opportunity to redesign itself - not only to automate work, but to become smarter, faster, and more strategic. The organizations that benefit most will not be the ones that buy the most tools. They will be the ones that align data, process, governance, and people around a clear operating model.

Dematoera’s perspective is useful because it avoids both extremes: blind optimism and defensive hesitation. Her approach suggests a more credible path forward for procurement leaders:

strengthen the basics
experiment aggressively but responsibly
train the whole function
keep humans accountable
build where differentiation matters

The organizations that move now will likely create an advantage that is difficult to close later. In procurement, that advantage will not come from AI alone. It will come from how well the function learns to apply AI to judgment, markets, suppliers, and execution at the same time.

Source: "AI with Purpose: Designing Procurement's Next Chapter W/ Gaël De Martelaere" - Art of Procurement, YouTube, Jul 27, 2026 - https://www.youtube.com/watch?v=9WFhFRzyX3Y

How to Prepare Procurement for AI: 7 Practical Steps