AI is moving into execution systems

Many AI conversations in supply chain focus on insight, forecasting or orchestration. Manhattan’s session at DELIVER Europe 2026 focused on a more operational question: how can AI help teams execute work faster inside the systems they already use?

Raphaël Hervé positioned agentic AI as a tool for accelerating execution across warehouse management, order management, transportation and store operations.

The focus was not on replacing execution systems, but on embedding AI into them so users can act faster on real-time data.

Real-time data changes what AI can do

Execution environments depend on live operational information: where inventory is, which orders are at risk, what labour is available, which shipments are late and what exceptions need action.

The session highlighted why this matters. If AI is working inside the execution system, it can use the same real-time data and permission structure as the user. That helps maintain control while allowing agents to support operational decisions.

Security was also a key theme. AI should not expose data that users are not authorised to access, and customer operational data should remain protected inside the relevant environment.

Agents can reduce exception-handling time

One of the warehouse examples focused on wave execution.

When orders are not selected during a wave, teams often need to investigate why: inventory may be unavailable, blocked, not yet put away or missing key data. That investigation can consume valuable time and delay fulfilment.

An AI agent can help surface the reason for the exception, show impacted orders or lines and suggest the next action. This turns what can be a manual diagnostic process into a faster, more guided workflow.

AI can support labour and transportation workflows

The session also explored labour and transportation use cases.

In labour management, agents can help managers understand which areas are running ahead or behind, where staffing gaps exist and which associates could be reassigned. In transportation, autonomous agents can support freight invoice processing by reading documents, comparing them to expected charges and surfacing exceptions.

The practical value is time. AI can reduce repetitive investigation and administration so operational teams can focus on decision-making and service performance.

What this means for the DELIVER community

For retailers, brands and logistics operators, the opportunity is to apply AI where execution complexity already creates friction.

The strongest use cases are likely to be specific, operational and measurable: reducing order deselections, improving labour allocation, identifying shipments at risk, processing invoices or supporting customer service teams.

Agentic AI will only create value if it is trusted, controlled and embedded into the workflows teams use every day. For supply chain leaders, the next step is to identify where operational decisions are slowed by manual investigation and where agents could help teams act faster.