Agentic Procurement Is Moving From Pilot Projects to Operational Control

For years, procurement automation largely meant digitizing approvals, centralizing sourcing events, or reducing paperwork inside fragmented enterprise systems. The newest wave of procurement AI is targeting something far more ambitious: replacing operational labor itself.

Across procurement software vendors, the language is shifting away from workflow automation and toward autonomous execution. Instead of helping buyers complete tasks faster, vendors increasingly claim their systems can negotiate with suppliers, evaluate sourcing options, orchestrate approvals, and execute transactions with limited human involvement.

The change reflects mounting pressure inside large enterprises. Procurement teams are being asked to manage growing supplier complexity, compliance demands, inflationary volatility, and cost pressure without proportional headcount growth. At the same time, many organizations remain burdened by fragmented ERP environments, disconnected intake systems, and supplier data spread across multiple platforms.

The result is a market now racing toward what vendors describe as “agentic procurement.” Whether enterprises are operationally ready for that transition remains far less clear.

From Workflow Automation to “AI Employees”

One of the clearest signals of the shift came from Gain, a company focused on autonomous procurement execution. In written responses to questions from The Supply Chainer, the company described a significant change in how enterprise procurement leaders are evaluating automation investments.

According to Michael Westman, Marketing Lead at Gain, procurement executives are increasingly moving away from the idea of automating isolated workflows and toward replacing operational roles entirely.

“Core operators are no longer asking how to digitize a sourcing event. They are asking how many negotiations a single AI employee can run in parallel, what oversight model that requires, and how to redeploy human buyers toward strategic suppliers and exception handling. Tail spend is where this is landing first because it is the workload that humans were never going to get to anyway. Indirect procurement is following close behind, driven by PO cycle time and compliance pressure rather than raw savings,” Westman wrote in the company’s response to The Supply Chainer.

Gain says its current deployments include autonomous negotiation systems and end-to-end indirect procurement agents capable of handling demand intake through supplier award. The company also claims some deployments have already generated measurable savings, including a reported 20 percent category price reduction in one consumer packaged goods environment.

Still, many procurement organizations remain cautious about how far autonomous execution can realistically scale. Exception handling, supplier disputes, regulatory controls, and ERP fragmentation continue to create operational friction that pure automation layers often struggle to manage cleanly.

ToolsGroup Positions Decion as a Step Toward Agentic Autonomy in Planning

Supply chain planning systems, long designed for periodic replanning under relatively stable conditions, are facing growing pressure in an environment defined by persistent volatility. The gap between planning and execution has become a central operational bottleneck. Many organizations still rely on static forecasts and manual interventions that struggle to keep pace with rapid shifts in demand, supply disruptions, and cost pressures.

ToolsGroup’s launch of DecionTM reflects one response to this tension. Positioned as an agentic AI platform, it seeks to move beyond traditional automation toward continuous, autonomous decision-making. Built on probabilistic modeling and multi-objective optimization, the system is designed to sense changing conditions in real time, evaluate risks across multiple scenarios, recommend actions, and execute approved decisions within defined human guardrails.

Key elements include an agentic digital twin, inventory-aware demand shaping, and cross-enterprise orchestration capabilities. The stated goal is to shift planners from routine execution toward higher-level orchestration, targeting improved service levels, lower inventory carrying costs, and greater resilience without adding headcount.

While the industry is still testing the practical boundaries of agentic autonomy in planning, Decion represents a clear example of the broader move from batch-based optimization to continuously steering supply chain performance.

The Orchestration Battle

Much of the current competition in procurement AI is increasingly centered on orchestration rather than simple automation.

That trend is particularly visible at Coupa, which has spent the past year aggressively expanding its AI and orchestration stack through acquisitions and platform launches. The company recently announced acquisitions of both Rossum and Tonkean while simultaneously launching Coupa Compose, a platform designed to build and orchestrate AI agents across procurement, finance, and supply chain workflows.

The broader strategic goal appears increasingly clear: becoming the operational control layer connecting buyers, suppliers, enterprise systems, and autonomous agents.

In a written statement provided to The Supply Chainer, Coupa described the next stage of procurement AI as less about isolated copilots and more about coordinating large-scale autonomous workflows across enterprise environments.

According to the company, the combination of orchestration infrastructure, intake management, and agent-to-agent coordination is designed to reduce manual handoffs that still dominate many procurement processes today.

Coupa stated that Tonkean’s orchestration framework enables “complex workflows and replaces manual handoffs, reducing cycle times by 50% and saving operations teams over 30 hours per week.” The company also emphasized that its architecture is built around integrating AI agents directly into existing enterprise systems rather than requiring organizations to replace core infrastructure.

That positioning reflects a broader market reality. Many enterprises already operate highly customized procurement environments layered across ERP systems, supplier portals, compliance tools, and finance platforms accumulated over years of acquisitions and regional deployments. Replacing those systems outright is rarely operationally feasible.

Instead, orchestration is emerging as the next battleground: connecting fragmented systems well enough for AI agents to act across them reliably.

Alex Saric, CMO Ivalua: "many so-called unified procurement platforms still operate on siloed architectures underneath a shared interface"

“Too many organizations believe layering AI on top of fragmented systems will solve the underlying problems.

Autonomy Still Depends on Foundations

The operational challenge, however, is that many procurement environments remain structurally fragmented beneath the AI layer.

That concern surfaced repeatedly in responses provided by Ivalua, which focuses on unified source-to-pay platforms for enterprise procurement operations.

In responses submitted in writing to The Supply Chainer, the company argued that many organizations are underestimating how heavily autonomous systems depend on unified data and operational consistency.

According to Alex Saric, Chief Marketing Officer at Ivalua, many so-called unified procurement platforms still operate on siloed architectures underneath a shared interface.

“Too many organizations believe layering AI on top of fragmented systems will solve the underlying problems. But when data remains fragmented and systems disjointed, AI often amplifies existing issues rather than resolving them. Supplier adoption also remains a major challenge. Even after platform deployments, suppliers frequently bypass systems or avoid onboarding requirements, limiting visibility and reducing the expected return on investment from automation initiatives,” Saric wrote.

The warning highlights a growing disconnect inside the procurement software market. Vendors are increasingly promoting autonomous execution capabilities while many enterprise procurement environments still struggle with inconsistent supplier master data, disconnected approval chains, and incomplete workflow standardization.

That does not necessarily mean agentic procurement will fail. But it likely means adoption will progress unevenly across categories and processes rather than through immediate end-to-end autonomy.

Tail spend, indirect procurement, invoice handling, and repetitive sourcing workflows appear to be moving first because they involve lower operational risk and higher volumes of manual work. More strategic procurement functions still require human oversight, supplier relationship management, and exception-based decision-making that AI systems remain less equipped to handle independently.

The larger shift, however, is already underway. Procurement automation is no longer being positioned primarily as productivity software. Vendors increasingly see it as a labor model.