Over the past few months, I’ve had a number of conversations with procurement leaders about AI and orchestration. Some of those happened during webinars and industry forums. Others came up in practical discussions with teams trying to modernize how procurement work actually gets done inside their organizations.
One exchange in particular stayed with me. It came during a webinar I conducted with SIG | ORG. A participant asked a series of questions that made me pause.
We’d been speaking in somewhat abstract terms about contracts, autonomy, and the business potential of applying AI agents to contract management. Then the questions came in. They weren’t about autonomy. They were about exposure:
One participant asked how they could confidently determine which agreement actually governs a purchase when a master agreement, multiple amendments, entity-specific pricing schedules, and regional variations all come into play.
Someone else described negotiating a telecom agreement that referenced more than twenty external links, each of which needed to be reviewed and validated because, as a procurement professional, their fiduciary responsibility requires understanding every obligation attached to the contract.
Here’s the truth. Many of us in the technology space—especially the orchestration space—are excited about the long-term potential of autonomous procurement agents.
Here at Tonkean we have spent a lot of time thinking about the potential for orchestration technologies to enable truly autonomous procurement agents. From a technological standpoint, the progress is real. Architectures that operate above enterprise systems (sometimes referred to as orchestration layers) can coordinate work across fragmented tools, teams, and processes.
Rather than being embedded inside a single application, these systems operate across the enterprise landscape and alongside the humans who are actually making decisions. By doing so, they can begin to capture something most enterprise systems miss entirely: the context behind how that decision was actually reached.
From a scientific perspective, this is closely related to what Sagi Eliyahu (CEO of Tonkean) describes as the context graph, a structured representation of how decisions, processes, policies, and systems interact across an organization.
Most enterprise platforms capture what happened: a contract was approved, a vendor was onboarded, a payment was issued. But procurement work rarely happens neatly inside a single system of record.
It happens in the spaces between them, through the interpretation of policies, conversations between stakeholders, exceptions granted because of business context, and in the many undocumented steps that define the day-to-day reality of enterprise operations.
If AI agents are ever going to take on meaningful autonomous responsibility in procurement environments, especially long-horizon processes like sourcing or contract management, they need to understand that context.
Capturing how work actually happens is what allows systems to eventually help guide it.
It’s natural that many of us in this space spend a lot of time talking about that future. The idea of more autonomous procurement operations is compelling, and the technology to support it is beginning to take shape.
But it’s also not where most procurement stakeholders are today.
That gap became very clear during the SIG webinar.
The questions asked were not abstract AI problems. They were everyday procurement problems.
Procurement professionals are not worried about agent autonomy. They are focused on diligence, risk management, and making informed decisions in environments where information is fragmented across systems, documents, and external references.
For technology providers, we need to meet procurement operators where they are. Discussions about the promise of tomorrow should flow from practical applications today. Without solving those operational challenges first, conversations about autonomy will naturally feel disconnected from day-to-day procurement work.
This gap between the promise of AI and the operational reality organization face came up again recently during an international finance executive networking event I attended.
In a discussion about AI adoption in enterprise software, a CFO asked a question that I suspect many organizations are quietly asking themselves right now.
Over the past decade, their company had invested heavily in finance and procurement technology. Now, almost every vendor in their stack was launching some form of bolt on AI capability.
The question was simple: How should they approach it?
It’s a fair question, because software vendors are commercial organizations. They need to innovate quickly and respond to market expectations. Adding AI capabilities to existing products is a logical and, in many ways, necessary step to not be left behind. But from an enterprise operating model perspective, something interesting happens when AI gets layered on top of an already fragmented technology landscape.
We end up applying AI to systems that remain fundamentally fragmented.
That doesn’t mean those capabilities are without value. In many cases, they can be a helpful way for organizations to begin exposing their teams to AI-assisted workflows.
But the deeper transformation will likely require a broader shift, one that connects data across systems, captures the operational context behind decisions, and enables proactive agents to operate across processes rather than inside individual applications.
Across the broader source-to-pay lifecycle, from supplier onboarding and sourcing to contracting, purchasing, and invoice management, procurement teams constantly operate across fragmented systems and incomplete context.
This is another gap. What many procurement teams are really asking for is not “AI agents” in the abstract. They are asking for operational continuity.
They want visibility into how policies are applied across processes. They want to understand why decisions were made. They want to reduce cycle times without compromising diligence.
They want technology that supports the way procurement professionals actually work.
This is where orchestration and context-aware systems begin to matter.
Not because they replace procurement professionals, but because they help connect the fragmented decisions and data points that define procurement work today.
Any discussion about the importance of orchestration, AI agents, etc., needs to start here: how these tools deliver needed solutions.
One reason procurement professionals’ may disengage from conversations about AI agents “changing the world” is because technology solutions often skip this step.
Some do it unintentionally. Others are simply trying to ride the wave of excitement around AI.
But procurement professionals are right to be skeptical. Before any technology becomes transformative, it first has to prove indispensable. Procurement leaders do not wake up thinking about agent autonomy.
But if orchestration technologies can help solve real-world operational challenges—by connecting systems, capturing context, and supporting procurement professionals in the decisions they already make—something interesting begins to happen.
Autonomy stops being a futuristic concept. It becomes the natural outcome of better operations.
And that, in my view, is where the real opportunity for procurement lies.
Want to learn more about the practical applications of process orchestration? Start here.
