This Glossary is a structured compendium of the most used terms, protocols, and technologies in Industry 4.0, production monitoring, and advanced industrial maintenance.
The terms that define iGromi OS, in the order the data travels through them: Sources deliver signals, the Second Brain organises them, Lobes teach it a domain and the Orchestrator coordinates the Agents. These are the official definitions.
The product category iGromi belongs to. It describes an infrastructure able to connect the Sources of an operation, organise their context, keep operational knowledge over time and let AI agents and models understand a plant and act on it. It names the category, not an internal part of the product.
iGromi's main product. It connects the Sources of an operation, organises their information in the Second Brain and lets you install Lobes, coordinate Agents and choose the Swappable Cognitive Core used to reason about the plant.
Systems, devices or services that deliver signals or facts from the operation: PLCs, sensors, cameras, meters, machines, ERP, SCADA, MES, CMMS, APIs and databases. A Source does not need to understand the full context of the plant — iGromi OS connects it and turns its signals into information the Second Brain can use.
The shared graph where iGromi organises the entities, relations, events, situations and knowledge of an operation. It keeps context over time, relates information coming from different Sources and gives Lobes, Agents and the Orchestrator a common understanding. It is not a transactional database and it does not replace the customer's systems: those keep their own data and teach it through events.
A Lobe packages the knowledge of a domain —ontology, rules, tools and agents— and installs into the Second Brain under a fixed contract of identity, permissions and shared memory. It ships in the .lobe format. A Lobe does not own the transactional data of an operation: it may come with a sovereign module, with its own database and backend, that publishes facts to the Second Brain through events.
The elements a Lobe declares it contains: knowledge, ontology, rules, MCP tools, agents, protocols, connectors, automations, views or policies. All of them are optional. What makes a package a .lobe is not its contents but its contract of identity, permissions and integration with the shared memory.
Software capabilities that observe, query or reason over the Second Brain to assist people, analyse situations, produce recommendations or prepare actions. An Agent may belong to a Lobe or use shared platform capabilities. It should not be confused with the Swappable Cognitive Core: the Agent defines a responsibility or behaviour, and the Swappable Cognitive Core is the model it reasons with.
The component that coordinates Agents, tools, permissions and workflows. It receives an intent, queries the Second Brain, identifies the available capabilities and coordinates the response or the action. It distinguishes three operational levels: Query, getting information without changing the operation; Propose, recommending an action without executing it; and Execute, performing a previously authorised and audited action.
The AI model iGromi uses to interpret language, reason, summarise, call tools or coordinate Agents. It is swappable because the customer chooses between local models running at the Edge or external providers over API —GPT, Claude, Gemini or others—. Identity, memory, permissions, tools and knowledge belong to iGromi OS and the Second Brain, not to the model in use. That is why replacing it does not cost a migration.
The catalogue where customers, integrators, manufacturers and developers discover, install, update and distribute Lobes. It turns reusable operational knowledge into installable packages for iGromi OS: process knowledge, maintenance protocols, industrial connectors, specialised agents, industry configurations, safety rules or operational views. Everything installable uses the same .lobe format.
The set of tools, contracts, documentation and validators to build, test, version and publish Lobes compatible with iGromi OS. It defines how to declare identity, version, Components, permissions, dependencies, extensions, capabilities and integration with the Second Brain.
The discipline that coordinates an organization's activities to extract maximum value from its physical assets across their life cycle, balancing cost, risk and performance.
The set of work orders still pending execution in a plant. A backlog that grows steadily signals that the maintenance team is overwhelmed or that data to prioritize correctly is missing.
Software that centralizes work orders, failure history, spare-part inventory and preventive maintenance plans for a plant's assets.
Intervention performed after the equipment has already failed. It has the lowest planning cost but the highest impact on unplanned downtime — the goal of any reliability programme is to shrink its share versus preventive and predictive work.
A dynamic virtual representation of a physical system. iGromi builds an Operational Digital Twin of your plant, showing the real-time state of every motor, conveyor and sensor in a web interface.
A distributed computing architecture where data is processed close to its source (the machine) instead of being shipped to the cloud. This guarantees near-zero latency, data security and continuous operation even without internet.
Software that manages finance, purchasing, sales and human resources at corporate level. It operates at a different abstraction level than an MES: it plans the business, it does not execute production in real time.
The network of instruments, sensors and devices connected to industrial computers. Unlike consumer IoT, it focuses on mission-critical operation, low latency and high reliability for process control.
A family of international standards defining the requirements for a physical asset management system. It establishes how to align maintenance and equipment life cycle with the organization's strategic objectives.
A maintenance metric measuring the average time elapsed between one mechanical failure and the next. A high MTBF indicates high equipment reliability.
The average time it takes the maintenance team to diagnose and fix a failure and return the asset to production. Reducing MTTR is the primary goal of early-warning systems.
The grandfather of industrial protocols. Created in 1979, it remains the de facto standard for connecting PLCs and sensors. iGromi supports both Modbus RTU (serial) and Modbus TCP (Ethernet).
Software sitting between the ERP and the shop floor (PLCs, SCADA): it executes production orders, tracks work-in-progress inventory and captures each machine's performance in real time.
The global standard for measuring manufacturing productivity. It calculates the percentage of production time that is genuinely productive, combining three factors: Availability (is the machine running?), Performance (is it running at full speed?) and Quality (is it producing good parts?).
A cross-platform machine-to-machine communication protocol. It is the modern Industry 4.0 standard thanks to its built-in security and its ability to handle complex data structures, not just bits and bytes.
A rugged industrial computer designed to control manufacturing processes. It is the machine's local brain. Common brands include Siemens, Allen-Bradley, Omron and Schneider Electric.
A maintenance strategy based on calendar or usage (operating hours, cycles): the asset is serviced at fixed intervals whether or not there are signs of wear. It reduces unplanned failures but can trigger unnecessary part replacement.
The practice of modernizing legacy machinery by adding sensors and IoT connectivity without altering its mechanics or original PLC. It digitizes 1980s/90s plants for a fraction of the cost of buying new equipment.
A methodology that analyses how each asset can fail and prioritizes maintenance strategies (preventive, predictive or corrective) according to the real operational impact of each failure mode.
The probability that an asset performs its function under normal operating conditions for a given period without failing. It is the central indicator of any predictive maintenance strategy.
A supervision and control system that remotely monitors and operates industrial processes, collecting data from PLCs and sensors in real time. Unlike an MES, a SCADA does not manage production orders or connect to the business layer.
A management philosophy where maintenance stops being the exclusive responsibility of a technical team and involves the entire plant staff, machine operators included, to maximize availability and eliminate losses.