Automated Negotiation for Future Supply Chains
agent systems. XMPro— the Australian Industrial AI vendor whose MAGS framework is embedded in its iBOS suite— was named a sample vendor by Gartner in three relevant categories( Composite AI, MAGS and Expert AI Agents) [ 32 ].
Cross-boundary coordination. MAGS deployments achieve their efficiency gains by globally optimizing over agents within a single organization, where the deploying organization can specify all utility functions. Coordination across organizations is structurally different: each side ' s utility function is private, the policies attached to those utilities differ, and full disclosure is generally not legitimate. Without a mechanism for principled exchange of partial information, cross-organizational decisions tend to escalate to humans, which removes much of the speed advantage that MAGS provides internally. Automated negotiation is one mechanism for closing this gap. It enables two organizations ' agent teams to converge on a binding agreement while each side preserves the privacy of its preferences, the integrity of its policies, and the audit trail its compliance function requires.
Three complementary components. The DTC testbed brings together three components that, in combination, address the cross-boundary problem. Digital twins help address the evaluation problem in automated negotiation by letting agents run counterfactual simulations to score candidate offers against real-time operational data. Automated negotiation helps address the cross-boundary coordination problem in MAGS by providing a principled protocol for convergence under private preferences. MAGS in turn provides the orchestration substrate that lets digital twins and negotiation agents participate in broader organizational decision-making. We do not claim this triangle is the only useful arrangement, but it captures one coherent integration pattern that the testbed is designed to evaluate.
The NEC – XMPro testbed. The Digital Twin Consortium announced its Digital Twin Testbed Program on 22 May 2025 [ 9 ], with the Automated Negotiation with Digital Twins and MAGS testbed led by NEC in collaboration with XMPro. The testbed will study the integration of negotiation AI with multi-agent systems and digital twins, and aims to produce a reusable framework for confidential, policy-preserving orchestration of interorganizational decision-making among enterprise teams of agents.
Worked example. In one illustrative configuration on the testbed, a supplier wishing to delay deliveries and a customer requiring steady supply searched a joint outcome space of 531 candidate schedules on the seller side and 441 on the buyer side. A fair win – win agreement was reached in approximately two minutes, with the seller realizing 91 % of its theoretical utility and the buyer realizing 81 %. The same configurable architecture supports a wide variety of supply-chain applications and is standards-compliant: the approach works with almost any multiagent generative AI system, running NEC ' s Negotiation AI engine on top of XMPro ' s industrial multi-agent layer.
The implication for inter-organizational coordination is the following. MAGS provides a substrate for reasoning and orchestration; the NAP provides a substrate for contracting. The combination, in principle, lets a buyer ' s MAGS team negotiate with a seller ' s MAGS team while each side preserves its own institutional policies, confidential utility weights and regulatory constraints. Pieter van Schalkwyk, CEO of XMPro and lead author of the
EDM Association – Journal of Innovation 109