Descartes Launches Agent Control Plane to Coordinate AI Agents Across Supply Chain Workflows

Supply chain software vendor Descartes Systems Group has launched a platform called the Agent Control Plane, bringing AI agents into the flow of logistics and supply chain work. According to Supply & Demand Chain Executive, the platform combines the operationally ready data and connectivity of the Descartes Global Logistics Network (GLN) with capabilities drawn from Descartes' other products, helping agents coordinate work across applications and trading partners.

The launch was announced at the Descartes Innovation Forum. In the keynote demonstration, the agents did not stay at the level of chatbots or better search — they executed parts of a representative end-to-end supply chain process.

The demonstration was unusually specific: an order moved through four representative companies, 15 Descartes products and 28 individual process steps. Agents handled compliance checks, information gathering, disruption detection and replanning behind the scenes, while the underlying Descartes applications kept doing the operational work they were built for.

Descartes frames the industry's persistent problem as the “swivel chair tax.” A shipment moves physically from origin to destination, but the information needed to manage it still travels through transportation systems, compliance applications, carrier portals, email inboxes, messaging platforms, spreadsheets and people. The applications themselves are often perfectly capable; the problem is everything that happens between them.

The most telling part of the demo may be what the agents did not do. When a compliance issue appeared, the agent gathered the relevant information and escalated the decision. Later, when a disruption created a chance to rebook transportation at an 8% saving, the alternative carrier carried a trust score of just 32 — the agent surfaced the option, and the human rejected it. Descartes' summary of the philosophy: autonomy is a dial, not a switch.

In practice, that means agents can detect a disruption, assess downstream impact, investigate alternatives and synchronize information across participating companies, delivering the problem and a proposed resolution together rather than a bare alert. It shifts the role of supply chain software, which for decades has largely waited for people to operate it.

Descartes insists this is not merely conceptual. The company says it has already executed roughly 3.25 million agent operations and is opening an early-access program for the Agent Control Plane. Governance is being built alongside the model: agents have identities, actions are logged and attributable, and activity can be reviewed and replayed.

For enterprises, that replayable governance layer may matter as much as the AI itself, since different decisions, risks and business contexts call for different levels of human oversight. What to watch next: the pace of the early-access rollout, which concrete scenarios those 3.25 million operations covered, and how trust scores and human overrides actually get used once agents move from stage demos into customer production environments.