In manufacturing, the administration around production decides delivery performance and cost: suppliers confirm different dates, fault reports end up as free text in the shift log, complaints need data from five systems. Agents bring this information together; they change nothing in machine control.
Industry 05/103 example agentsAutonomy level L3–L4
Example agents
Three agents that take over work here
Examples from typical workflows, not client references. Which agents pay off for you is shown by assessing your own processes.
01
The order confirmation agent
Purchasing · materials planning
Matches supplier order confirmations against purchase orders on price, quantity and date, accepts deviations within tolerance, alerts planners to critical date shifts and chases missing confirmations.
Deviations within tolerance at L4; price changes and critical dates at L3.
Systems
SAP S/4HANA MMEDI · supplier portalEmailMRP
Guardrails
No price change without approval, tolerances per material group, no order cancellations.
Human in the loop
Buyers decide on price deviations; planners assess shifted dates against production.
What to measure
Confirmation rate · time to resolution · shortages from late date changes
02
The maintenance agent
Maintenance · plant
Turns alarms and free-text shift log entries into structured maintenance notifications, assigns them to the asset, finds similar past faults and their fixes, checks spare parts stock and prepares the work order with a parts list.
The agent creates notifications at L4; work orders and parts orders stay at L3.
Systems
SAP PM · EAMMES · control room (read-only)Spare parts storeKnowledge base
Guardrails
Read-only access to machine data, no intervention in controls, orders only up to the approval limit.
Human in the loop
Maintenance supervisors prioritise and release orders; safety-relevant faults go to people immediately.
What to measure
Time from fault to work order · repeat faults · unplanned downtime
03
The complaints agent
Quality · customer complaints
Records customer complaints, gathers batch, inspection and delivery data from the systems involved, drafts the 8D report up to the root-cause hypothesis and tracks the deadlines towards the customer.
Autonomy level
L0
L1
L2
L3
L4
Start L3 · Assisted agents
Data gathering and drafting at L3; quality owns root cause and actions.
Systems
SAP QMCAQMESCRM · email
Guardrails
No statement on root cause to the customer, no credit notes, every statement with its data source.
Human in the loop
Quality engineers confirm root cause and actions and sign the report.
What to measure
Time to first 8D status · deadline adherence · repeat complaints
Framework
What to factor in for this industry
Machinery: Since the Digital Omnibus, AI as a safety component of machinery takes its own route via the Machinery Regulation (EU) 2023/1230, which applies from 20 January 2027. The examples here do not touch machine controls.
NIS2: Manufacturers of certain products, such as machinery, vehicles, electrical equipment or medical devices, are among the sectors covered. Agents then need the same controls as any other account with system access.
Works council: When agents touch performance or behaviour data of employees, for example in the shift log, plan for co-determination; the audit trail shows which data the agent actually uses.
Note, not legal advice: whether a specific use falls under these rules is for you and your legal counsel to decide. We provide the technical facts for it, from the agent profile to the audit trail.
