The Kinexions Maestro Remix Agent Hackathon is entering the final stretch of the build phase, and the story is no longer only about what AI agents can do.
It is about how teams are deciding where agents can create real value in supply chain planning.
Across Remix, teams are building agents for challenges in forecasting, inventory, supply, and more. The use cases vary, but one thing has stood out throughout the build phase: Participants are not just experimenting with technology. They are making design choices grounded in the realities of planning work.
That is where some of the most interesting learning is happening.
What decision should the agent support?
Some teams are starting with a simple question: What the decision does a planner needs to make?
In supply chain planning, the hard part is often not finding more data. It is understanding what changed, what matters, and what action should come next. These teams are thinking about agents as decision-support tools that help planners evaluate options, compare tradeoffs, and move from information to action with more clarity.
The question is not simply, “What can the agent tell me?”
It is, “What decision can this agent help me make better?”
That distinction matters because it changes how the agent is designed. It shifts the focus from producing an output to supporting a planning moment. It helps teams think about the user, the workflow, the tradeoff, and the action that follows.
What context does the agent need?
Other teams are thinking carefully about the signals an agent needs to interpret.
An agent is only useful if it has the right context for the problem it is trying to solve. That context might include lead times, customer priorities, or other signals that shape the decision at hand.
But context is not just about giving the agent more information. It is about giving it the right information in a way that supports a useful output.
The same signal can mean different things depending on timing, constraints, and commitments. A demand increase, a late purchase order, or an inventory imbalance rarely tells the full story on its own. The value comes from connecting that signal to the planning reality around it.
Where should the human stay in control?
Another important design choice is where the agent should stop and the planner should decide.
Agents can synthesize information, surface risks, compare scenarios, and recommend next steps. But planners still need to validate assumptions, interpret tradeoffs, and make decisions.
The strongest ideas do not remove human judgment from planning. They give planners better support for the moments where judgment matters most.
A new design conversation
The real innovation in Remix is not only the agents being built. It is the design conversation happening around them.
Teams are not just asking for features. They are asking sharper questions about the work itself: where decisions slow down, where context gets lost, where signals are hard to interpret, and where better support could make a meaningful difference.
As the build phase comes to a close, the submissions will be exciting to see. But the thinking behind them may be just as important.
What makes Remix powerful is not only that teams are building agents. It is that they are learning how to shape agentic AI around the work they know best.
Stay tuned as Remix teams continue to advance their builds and prepare to show what focused, practical AI agents can do in Maestro.
