Logistics Reply Launches Five Warehouse AI Agents With Task-Specific Authority

Logistics Reply is giving warehouse operators five ready-made AI agents—and a framework for deciding how much power to hand them. Announced on October 5, 2026, the LEA AI Agent Authority Model accompanies agents available that day in LEA Dynamic Intelligence. The company’s stated goal is task-appropriate authority, not maximum autonomy: permission should depend on operational context and risk.

Readiness and permission are separate decisions

Despite its name, the Authority Model is a decision framework, not a new AI model. It brings together two dimensions: an organization’s readiness to use AI and the authority an agent should receive for a particular task. Those decisions are linked, but not interchangeable.

The framework pairs four stages of organizational AI maturity, ranging from early adoption to mature, with five authority levels: Inform, Recommend, Act, Coordinate and Governed Autonomy. Logistics Reply says customers should choose authority according to the use case, its context and risk, rather than simply an agent’s ability to act.

Authority can then grow as operational evidence and trust develop, the company says. That makes the framework a way to decide both where agents can work now and when their responsibilities should increase—not a blanket instruction to automate every warehouse decision.

AI maturity is organizational; agent authority is contextual.

Enrico Nebuloni, Executive Partner at Reply

Five tasks, different action boundaries

The five agents show how those distinctions translate into warehouse work. Logistics Reply describes the following capabilities, with delegated authority and human oversight matched to each task:

• Out of Stock investigates why stock is unavailable, distinguishing real shortages from temporary issues. Its role is diagnostic: the company presents it as a way to reduce delays and manual checks by identifying the cause of an apparent stock problem.
• Labor Distribution identifies bottlenecks and estimates the effort needed to meet cut-off times. It recommends moving workers to rebalance workloads in real time. The described output is a workforce recommendation, rather than an explicitly stated power to carry out that reassignment itself.
• ABC Rebalancer recalculates item classifications using movement data and produces a reclassification report. After approval, it writes the updated classes into the warehouse management system’s item records. Here, the company specifies an approval step between preparing a change and applying it to operational data.
• Dock Scheduling lets planners and carriers find and book dock-door slots through ordinary conversation. Its guided dialogue takes scheduling rules into account. Unlike an agent that only recommends an action, its described capabilities include completing a booking.
• Lost & Found starts when a task runs late. It can use camera input to examine the surroundings and identify causes of delay, including an item blocking an autonomous mobile robot’s route. The announced capability is detecting the problem, not removing the obstruction.

Pre-built agents still have distinct connections

Logistics Reply says the five agents address recurring tasks without requiring custom development. But they are not presented as having one uniform connection to warehouse systems: each has its own data contract, defining its data requirements, and its own integration approach.

The surrounding LEA Reply platform combines warehouse management, planning and optimization with connections to automation and robotics. Logistics Reply describes it as a common foundation for people, processes and different automation technologies. Its intended settings include both highly manual warehouses and increasingly automated operations, rather than only facilities already dominated by robots.

For customers needing different workflows, LEA Dynamic Intelligence also provides an agent builder for creating and deploying tailored agents. The release therefore offers both ready-made warehouse functions and a customization path, with the authority framework intended to guide permissions across those operational choices.