Four logistics use cases, stock re-routing at port, dynamic network selection, container utilisation and warehouse capacity, ran as agents that recommend and apps where planners decide, with the accepted decisions written to the execution and planning systems.
Goods waited at the dock for transfer, primary distribution centres were congested by hub-and-spoke shipments, containers left under-used and warehouse overflows were planned late.
Each use case pairs a model with a decision app. The model recommends: re-route this stock from the port to that warehouse, ship direct and skip the hub, consolidate these orders into a full container, add flexible warehouse space in week 12. The app shows the recommendation with its lead-time impact, its proposed date, the days-of-supply comparison and the financial impact, and the planner accepts, changes or rejects it; accepted transfer requests are written to the planning and execution systems.
Catalogue entries that match the apps in this story.
Every recommendation is an instance with its own process ID; the planner's decision and the write-back acknowledgement are logged; the board shows demand against supply, days-of-supply deviation at the 3PLs and the effect of the accepted decisions.
Reviewed in the app, with potential cash release above $500K.
$3M loss on sales avoided, 11 percent lead-time reduction, service level up 2 basis points.
Container utilisation up 23 percent; about €300K a year saved on warehouse capacity planning.
Slot for the customer’s own words: supplied and approved by the reference contact before publication. The story above is written in Lima’s voice from the engagement material; the customer’s feedback goes here in theirs.
Build the decision app before the agent is trusted to act: the log of accepted and rejected recommendations is what earns the agent its autonomy.
Under NDA, with the account owner's approval, we arrange a conversation with a customer whose processes resemble yours.