AI is changing the ambition of enterprise software. For years, procurement technology has focused on helping professionals gather information, identify patterns, and make better decisions. Increasingly, however, AI is being expected to participate directly in decision-making and execution.

This shift extends well beyond procurement. Across enterprise software, product teams are investing in copilots, intelligent agents, orchestration platforms, and new approaches to representing enterprise knowledge. While these technologies address different problems, they all point toward the same objective: enabling software to operate with a richer understanding of how the business actually works.

That prompted us to step back and ask a broader question:

What actually determines how far procurement autonomy can progress?

We believe the answer extends well beyond AI capability alone. As AI moves from supporting users toward participating in decision-making and execution, the amount and richness of organizational understanding required increases significantly.

AI Is Changing How Work Gets Done

AI is already creating meaningful value across procurement.

From contract analysis and spend classification to supplier discovery and risk identification, today’s capabilities are helping procurement professionals work faster and make better-informed decisions.

What’s more interesting, however, is how AI is changing the role software is expected to play.

Historically, procurement systems were designed to support human interpretation. People naturally supplied organizational context, historical perspective, and business judgment that rarely existed explicitly within software.

Increasingly, AI is expected to participate in decision-making and execution itself.

That changes the problem technology providers are trying to solve.

From Information to Decision-Making

Consider a familiar supply chain risk scenario.
A risk monitoring platform identifies that a critical supplier has exposure to a geopolitical disruption. The alert is accurate.
But determining the appropriate response requires considerably more than acknowledging the alert.

• How dependent is the business on this supplier?
• What inventory is available?
• Are qualified alternatives already approved?
• What contractual commitments exist?
• How has the supplier responded to previous disruptions?
• What level of risk is acceptable to the business?

These aren’t unusual procurement questions. They arise every day.

The important observation is that while the system successfully identifies the issue, determining the appropriate response still depends on understanding the enterprise in which the event occurs.

We don’t believe this pattern is unique to supply chain risk.

The same dynamic appears in sourcing, contracting, supplier onboarding, and many other procurement activities. As decisions become more consequential, they increasingly depend on a broader understanding of the enterprise.

A Different Way to Think About Autonomy

Rather than thinking only about what AI can do, we’ve found it useful to consider what AI needs in order to participate reliably in decision-making and execution. That shift in perspective changes how we evaluate procurement activities and where higher levels of autonomy are likely to emerge.

We’ve found it more useful to ask three different questions.

• How much context does this activity require to support reliable decision-making and execution?
• How much of that context is available to the system?
• What are the consequences of getting the decision or execution wrong?

Taken together, these questions provide a practical way to think about procurement autonomy.

Some procurement activities operate within relatively well-defined boundaries. Others require a much broader understanding of the enterprise, the external environment, and the business consequences of acting. As a result, they are unlikely to progress toward autonomy at the same pace—not because AI is less capable, but because the understanding required becomes substantially richer.

Viewed through this lens, procurement autonomy is better understood as a continuum than a destination.

Looking Ahead

Throughout this article we’ve deliberately used the word context without defining it too precisely.

That’s because we believe the more important question isn’t whether AI has access to more information, but whether it has enough understanding of the enterprise to participate reliably in decision-making and execution.

What kind of understanding does software actually need?

That naturally raises another question.

What do we actually mean by context, and what would it take for systems to operate within it rather than requiring people to reconstruct it for every important decision?

That’s where we’ll turn next.