RELEX Solutions said on November 2 that it has added new agentic operating capabilities to RELEX Open, the company’s platform for connecting systems and coordinating AI-driven workflows across the value chain. In a company announcement, RELEX said the additions are meant to let agents and AI tools in supply-chain planning, merchandising, and operations take account of priorities and constraints elsewhere in the business, rather than optimizing a single function in isolation.
“Giving every function its own AI agent does not create a better-run business if their decisions don’t work in sync,” said Mikko Kärkkäinen, group chief executive officer and co-founder of RELEX Solutions. “Our experience supporting hundreds of retailers, wholesalers, and manufacturers has given us the domain expertise and AI foundation to help companies move toward more adaptive, coordinated operations.”
RELEX Open, the company said, already opens RELEX capabilities to external systems, agents, partners, and customer teams through APIs, the Model Context Protocol, plugins, and other open standards. The new agentic layer, it said, now comprises three connected capabilities: a production-ready foundation of built-in agents and planning AI on a unified operational data model, covering forecasting, replenishment, merchandising, pricing, and operations; tools for customers and partners to extend RELEX with their own data, logic, user-interface plugins, agents, and skills; and an orchestration layer in which agents across functions can share context, identify priority actions when performance changes, and trace the decisions behind a recommendation. Underpinning those capabilities, RELEX said, is a shared knowledge graph that maps how processes, systems, teams, decisions, and performance indicators relate. The company said it applies a “responsible-by-design” approach, with security, transparency, explainability, and human oversight over how agents access information, make recommendations, and act.
Steve Rowen, managing partner at RSR Research, said in the same announcement that as retailers and manufacturers adopt more AI capabilities, “the next challenge is making sure those systems can work together to support better business decisions,” and that RELEX’s approach of “building orchestration on established planning capabilities and domain knowledge is a good direction.” That is an analyst comment, not a measured customer result.
The release also described several customer collaborations. Sprouts Farmers Market is working with RELEX to refine store replenishment with feedback from store teams. “What has been different about working with RELEX is the opportunity for our store teams to help shape the solution from the beginning,” said Steve Huesing, vice president of inventory management at Sprouts Farmers Market. RELEX also said a large European food retailer, already using RELEX planning at scale, worked with forward-deployed engineers on use cases including a supplier-planning challenge, and that the first capability was developed within days and ready in production within weeks; that a U.S. grocer moved fresh ordering and inventory capabilities onto RELEX and went live in stores within one quarter; and that a major European convenience retailer piloted RELEX Dynamic Markdown across store formats, increasing adoption of AI-generated markdown recommendations while reducing manual overrides and waste. Those accounts come from RELEX. Several of the companies are unnamed, and the release does not present controlled measurements that isolate the new orchestration layer from earlier planning software.
What it means for manufacturers
The useful caution in this announcement is one midsize manufacturers already feel: a better forecast that ignores production capacity, or a replenishment suggestion that ignores a promotion, can make the next decision worse. Coordinating those trade-offs still depends on clean item, inventory, and demand records, and on a person deciding which recommendation to accept. That is the same foundation discussed in AI demand forecasting software and in AI inventory management software.
A practical next step is to take one recent planning disagreement — a stockout during a promotion, or a replenishment that overshot after demand dropped — and write down which system made the call, which constraint was missed, and who could have overridden it. Our recent note on Blue Yonder’s AI-native order management covers a related problem of connecting a promise to what the operation can actually do.
Questions about forecast-and-replenishment handoffs, promotion overrides, or where an assistant belongs in a midsize planning process?
Contact usSources
Additional sources
- The PR Newswire posting of the same RELEX announcement (Nov. 2, 2026), which carries the same product description, executive and analyst quotations, and customer examples.
Photo: Walmart Corporate, Wikimedia Commons, CC BY 2.0 (illustrative)
Try the Statadventure live tools
Inventory tools for US manufacturers: reorder point, safety stock, EOQ and landed cost, with the inputs and assumptions shown.
Open the live tools