The first cohort of the Kinaxis Maestro Remix Agent Hackathon has officially concluded, and its three award-winning submissions offer an early look at how agentic AI could reshape supply chain planning.
Across the cohort, 18 teams and approximately 100 participants explored how AI agents can help planners investigate issues, interpret complex data, and move from insight to action. Each team brought a different perspective on where agents could create value within Maestro.
The three winning teams stood out for different reasons. Together, their submissions reveal three important signals about what will define successful agentic AI.
Signal 1: Practicality and completeness matter
The overall winner, a leading innovator in robotic automation, developed a well-defined agent concept that connected a clear business objective with a practical end-to-end workflow.
The team’s submission went beyond demonstrating what an agent could produce. They considered how the experience would work in practice, including how the agent would coordinate specialized tasks, retain relevant context, direct users to the appropriate resources, and manage confirmations before making changes.
Rather than presenting AI as a standalone capability, the team showed how an agent could support a broader planning process, from interpreting the user’s request and coordinating analysis, to recommending actions and helping the user execute them.
The takeaway is clear: Strong agent concepts must do more than generate answers. They need to fit into the way planners work, connect the steps required to reach a decision, and account for the practical realities of implementation.
Signal 2: Business impact is what makes AI credible
Incora received the Best Business Impact award for addressing a familiar but complex planning challenge: determining whether a part is truly at risk and understanding the downstream consequences.
The team developed an AI advisor that orchestrated four specialized agents to validate part risk, prioritize affected customer orders, classify available inventory, and investigate component-level root causes.
By bringing together data, business rules, and investigative workflows, the concept showed how agents could reduce the manual effort required to gather and interpret information from multiple sources. More importantly, it connected the technology directly to a business outcome: helping planners respond to supply issues faster and more consistently.
That connection is what makes AI credible. The value does not come from using an agent for its own sake, but in applying it to a meaningful decision and demonstrating how it could improve speed, focus, or planning performance.
Signal 3: Innovation comes from rethinking familiar work
Sanofi received the Most Innovative Agent award for demonstrating a creative approach to multi-agent analysis and decision support.
Its Net Demand Orchestrator used coordinated agents to compare scenarios across multiple worksheets, identify changes in demand, and uncover the drivers behind those changes. The concept reimagined work that planners already perform by showing how agents could coordinate different analyses and bring the most relevant findings together.
Innovation does not always require inventing an entirely new process. It can come from reconsidering familiar planning tasks and asking how they could become more connected, intelligent and explainable.
The community is shaping what comes next
These three winners represent different strengths: practical completeness, measurable business value, and creative potential. Together, they show that the future of agentic AI in supply chain planning will be shaped by concepts that combine all three.
Remix gives the Kinaxis community a space to test those possibilities, challenge assumptions and explore what agents could become within Kinaxis Maestro.
The next Remix cohort will bring together Kinaxis partners, introducing builders with new areas of expertise and perspectives. As each cohort contributes new ideas, agentic AI’s potential will continue to expand.
Explore more highlights from the first cohort here.
