The industrial AI & maintenance glossary.

MAINTENANCE MANAGEMENT Maintenance software & strategies

CMMS (Computerized Maintenance Management System)

Software that manages maintenance: work orders, asset hierarchy, preventive schedules, spare parts and intervention history. It is the system of record for what was done to which machine — but the documents, procedures and production context usually live in other systems.

GMAO — “CMMS” in French

GMAO (Gestion de Maintenance Assistée par Ordinateur) is the French term for a CMMS. Same role: managing work orders, assets and preventive plans. In English, the term you’ll search for is “CMMS”.

Preventive maintenance

Maintenance performed on a fixed schedule — by calendar time or usage (hours, cycles) — to prevent failure before it happens. Planned, not triggered by the machine’s condition.

Predictive maintenance

Maintenance triggered by the actual condition of equipment: data crosses a threshold and signals an impending failure, so you intervene just in time rather than too early or too late. In French, “maintenance prédictive” (and often “prévisionnelle”) refers to this condition- and data-driven approach. Crucially, much of the signal already exists in your CMMS, SCADA and MES — connecting it is often a faster first step than a sensor rollout.

OEE / TRS (Overall Equipment Effectiveness)

OEE — TRS (Taux de Rendement Synthétique) in French — measures how effectively a machine runs: Availability × Performance × Quality. A drop in OEE is a question (“why?”) whose answer usually spans production, maintenance and quality systems at once.

FACTORY SYSTEMS The systems on the floor

ERP (Enterprise Resource Planning)

The business backbone: orders, inventory, finance and production planning. In maintenance terms it holds the fabrication order for a machine — the planned side of what the CMMS records as work.

MES (Manufacturing Execution System)

Tracks production execution on the floor: lots, batches, quality events and genealogy. When a bad lot appears, the MES holds the “what happened”; tracing the “why” usually means linking it to machine parameters and procedures elsewhere.

SCADA / PLC

SCADA (Supervisory Control and Data Acquisition) supervises and controls industrial processes in real time; a PLC (Programmable Logic Controller — “API” in French) is the controller on the machine. They raise the alarms and error codes — often cryptic until matched to the right manual passage.

AI & DATA The AI vocabulary

Knowledge Graph

A model of the entities in your plant — assets, work orders, production orders, documents, events — and the relationships between them. Unlike flat tables, it lets a question travel across systems, which is what makes cross-system root cause possible.

OPC-UA & MQTT

Two industrial communication standards. OPC-UA (Open Platform Communications Unified Architecture) is a secure, vendor-neutral standard for exchanging machine data; MQTT is a lightweight publish/subscribe protocol common for telemetry. Both let a system read live data from SCADA/PLC without disrupting it.

AI agent (in maintenance)

Here, a specialised AI that reads and answers questions over one data source — documents, CMMS, ERP, SCADA, MES — fluent in that source’s shape. One agent per source, joined by a Knowledge Graph, beats a single generalist model that flattens the domains it doesn’t understand.