Glossary
This glossary explains the terms used in this documentation and across the Manufacturing Co-Intelligence portfolio. Terms are grouped by solution. Terms that apply to several solutions are listed under Portfolio.
The Smart Operations Toolkit and Semantic Stack documentation contain additional terms for installation, operation, and data modeling. See the SOT glossary and the Semantic Stack glossary.
Portfolio
Section titled “Portfolio”| Term | Definition |
|---|---|
| Manufacturing Co-Intelligence® | The Bosch portfolio for intelligent collaboration between humans, machines, and Agentic AI. It consists of three solutions: Agentic AI, the Smart Operations Toolkit, and the Semantic Stack. |
| Agentic AI | AI that pursues a goal on its own: it plans steps, uses tools, and acts, while people stay in control. Also the name of the solution in this portfolio that provides AI agents for manufacturing. See Agentic AI. |
| AI agent | Software that uses a large language model to understand a request, decide on the next steps, call tools, and return a result. Also called agent. |
| Performance Domain | A cross-functional area that a manufacturer must master to improve operational performance. The solutions of the portfolio are mapped to eight domains: Shopfloor Management, Value Stream Optimization, Value Stream Digitalization, Plant Maintenance, Data-Driven Quality, Data Analytics, Regulatory Compliance, and Product Lifecycle. |
| Bosch Production System (BPS) | The Bosch approach to lean production. KPI calculations and shopfloor routines in the portfolio follow BPS practices. |
| Shopfloor | The production area where people, machines, and material come together to manufacture products. |
| Shopfloor management | Leading production directly on the shopfloor, with regular routines, KPIs, and immediate reaction to deviations. |
| Human in the loop | A design principle: people review and approve decisions before an agent or system acts. See Human in the loop. |
| Tenant | An isolated environment for one customer organization. Users, data, and configuration of one tenant are not visible to other tenants. All three solutions use tenants. |
| Tenant ID | The unique identifier of a tenant. Include it in support requests. |
| Role | A set of permissions. Users receive roles, usually through groups, and can then use the functions the roles allow. See Roles and permissions. |
| Group | A set of users that share the same roles. Assign roles to groups instead of individual users. |
| KPI | Key performance indicator, such as overall equipment effectiveness (OEE), part count, cycle time, or downtime. |
| OEE | Overall equipment effectiveness: the share of planned production time that is truly productive, combining availability, performance, and quality. |
| Downtime | A planned or unplanned period in which a machine or line does not produce. |
| Deviation | A difference from the expected state, for example a machine fault, a downtime, or a process value outside its limits. |
| Digital twin | A virtual representation of a real asset, product, or process, kept in sync with it. See Semantic Stack. |
| Large language model (LLM) | An AI model trained on large amounts of text that understands and generates language. AI agents use LLMs to reason and to formulate answers. |
| Prompt | The instruction or question that is sent to a language model or agent. |
Agentic AI
Section titled “Agentic AI”Agents and solutions
Section titled “Agents and solutions”| Term | Definition |
|---|---|
| Shopfloor Agent | An agent that helps shopfloor teams report problems by voice or text and get immediate, actionable solutions from a cross-plant knowledge base. Called Shopfloor AI Agent in Agents. |
| Smart Maintenance Agent | An agent that aligns maintenance plans with real production conditions, turns recurring unplanned maintenance into planned activities, and answers questions about maintenance, procedures, and spare parts. See Agents. |
| Talk to Your Data AI Agent | An agent that answers questions about manufacturing data in natural language. It is powered by the Semantic Stack and queries digital twin data. |
| Workforce Scheduling AI Agent | An agent that monitors production needs and worker availability and adjusts schedules proactively. See Agents. |
| Curated agent | An agent that Bosch provides and maintains, ready to use. See Agents. |
| Chat UI | The web interface in which users talk to agents, rate answers, and approve actions. See Basic operations. |
| Agentic Flows | The low-code editor to build automated workflows and data ingestion pipelines from deployed agents and your systems. See Agentic Flows. |
Building and publishing agents
Section titled “Building and publishing agents”| Term | Definition |
|---|---|
| Agent Registry | The place where you build, configure, and publish agents from reusable components. See Agent Registry. |
| Component | A reusable building block in the Agent Registry, for example an agent, team, model, tool, or context. See Components and tools. |
| Team | Several agents that work together on a request. |
| Handoff | A transition from one agent to another, or back to the user. |
| Termination | A condition that ends a team conversation, for example a maximum number of messages. |
| Context | Instructions and guidelines for an agent, such as tone of voice or rules to follow. |
| Model | The language model an agent uses. |
| Tool | A function that an agent can call to get information or perform an action. |
| Built-in tool | A tool provided by the service, such as date and time, calculator, or knowledge search. |
| Toolbox | A set of tools generated from the OpenAPI description of a REST API. |
| Workbench | A connection to a remote tool server that uses the Model Context Protocol (MCP). |
| External tool | A tool that runs in your own application instead of in the service. See External tools. |
| Environment | A named set of settings, such as endpoints or timeouts, that differ between stages. See Environments and secrets. |
| Secret | A credential, such as an API key, stored securely per tenant. Secret values cannot be read back. |
| Deployment | A published selection of agents and teams, combined with environments, guardrails, and the roles allowed to use them. See Deployments. |
| Guardrail | A rule in plain language that checks requests or answers during a conversation and blocks them if they break the rule. Input guardrails check requests, output guardrails check answers. See Guardrails. |
| Policy | A rule that checks how agents are configured, for example which tools may be combined with which data. Violations are reported when components or deployments change. See Policies. |
| Metadata | Key-value attributes of components that policies evaluate. See Metadata. |
| Approval | A confirmation that users give before an agent performs a sensitive action. See Feedback and approvals. |
Integration, knowledge, and access
Section titled “Integration, knowledge, and access”| Term | Definition |
|---|---|
| Agent Runtime | The API to talk to deployed agents and teams from your application. See Integration Guide. |
| Chat completions API | The default, OpenAI-compatible API of the Agent Runtime, with or without streaming. |
| Responses API | An alternative streaming API that delivers reasoning, tool activity, and citations as separate items. |
| Session | The stored state of a conversation, identified by a chat ID. See Sessions. |
| Streaming | Delivering an answer in small parts while it is generated, so users see text immediately. |
| Knowledge search | Semantic search over your own documents, which agents use to ground their answers. See Knowledge and Search. |
| Semantic Query Engine | The API to manage datasources, ingest documents, and run semantic searches. |
| Datasource | A named collection of documents, for example all product manuals. Access is controlled per datasource. |
| Document | A text that you load into a datasource, together with its metadata. |
| Node | A searchable piece of a document. Long documents are split into several nodes. |
| Ingestion | Loading documents into a datasource so that they can be searched. See Ingesting documents. |
| Semantic search | Search by meaning instead of exact words. Hybrid search combines semantic and keyword search. |
| Source | A document that an agent used for its answer. Sources are shown with the answer. |
| Feedback | A rating of an agent answer by a user. See Feedback and approvals. |
| Service role | A role defined by the service that allows administrative functions, such as service-operator. |
| Access role | A role that you define to control who may use a specific agent or search a specific datasource. |
Smart Operations Toolkit
Section titled “Smart Operations Toolkit”| Term | Definition |
|---|---|
| Smart Operations Toolkit (SOT) | A set of shopfloor applications that combines lean management principles of the Bosch Production System with IoT, installed in your own environment. All modules work on the same central data. See Smart Operations Toolkit. |
| Nexeed Industrial Application System | The industrial IoT platform on which the Smart Operations Toolkit modules are built. |
| Module | A licensed application of the Smart Operations Toolkit, such as Shift Book or Condition Monitoring. |
| Condition Monitoring | Monitors machine measurements, events, and counters, and triggers actions, such as maintenance tickets or notifications, when rules detect deviations. Also offered as a solution package. |
| Process Monitoring & Anomaly Detection | A solution package that visualizes process curves, such as pressure, temperature, or force, in real time and uses self-learning AI to detect anomalies. Based on the Process Quality module. |
| Process Quality | Monitors production process data in real time and visualizes measured values as process curves to detect deviations early. |
| Shopfloor Performance Insights | A solution package with pre-configured KPI dashboards to find bottlenecks and their root causes. |
| Digital Daily Leadership Routine | A solution package that supports daily shopfloor meetings with structured templates, integrated KPIs, and tracked measures. Based on the Operational Routines module. |
| Andon Live | Shows the current part count against the target and machine faults on one or more lines, and triggers requests for assistance. |
| KPI Reporting | Provides reports on production and machine KPIs, such as OEE, part count, cycle times, and downtimes. |
| Shift Book | Plans and documents shifts, downtimes, and shift handovers. |
| Operational Routines | Presents safety, quality, cost, and other production KPIs in standardized routines, such as a daily morning meeting. |
| Deviation Processor | Shows deviations detected by other modules, such as downtimes and faults, and reacts to them automatically or manually. |
| Quick reaction | An automatic reaction of the Deviation Processor that forwards matching deviations to a reaction system, such as Ticket Management. |
| Ticket Management | Manages tickets for tasks and problems, created manually or automatically from deviations, and assigns them to resolving groups. |
| Notification Service | Records events from modules and notifies subscribed users by email or web push. |
| Information Router | Receives messages from machines, controllers, and PLCs on a single endpoint and forwards them to the configured modules. |
| Master Data Management (MDM) | Central management of master data that several modules use: production structure, equipment, error definitions, processes, and materials. |
| Device Portal | Manages and updates the IoT devices of one or more plants through a web interface. |
| AI Services | Provides machine learning capabilities for the modules, such as anomaly detection, pattern recognition, and predictive analytics. |
| Web Portal | The central entry point to all licensed modules, with dashboards and user profile settings. |
| Shopfloor Management Administration | Manages targets, downtime causes, and configuration for Andon Live, KPI Reporting, Operational Routines, and Shift Book. |
| Nelson rules | Statistical rules that detect non-random patterns or out-of-control conditions in process data. Available in Condition Monitoring. |
| Service pack (SP) | A release that delivers fixes and security updates for a major Smart Operations Toolkit release. |
Semantic Stack
Section titled “Semantic Stack”| Term | Definition |
|---|---|
| Bosch Semantic Stack | A semantic data foundation for product lifecycle data. It is organized in four layers: Semantic Data Models, Semantic Data, Digital Twins and Data Products, and Solutions. See Semantic Stack. |
| Semantic Data Models Layer | The layer that establishes a shared understanding of data through Aspect Models. |
| Semantic Data Layer | The layer that harmonizes and stores data according to Aspect Models, for example in the Semantic Data Lakehouse. |
| Digital Twins and Data Products Layer | The layer that links data to products and assets through digital twins, for example in the Digital Twin Registry. |
| Solutions Layer | The layer with applications that create value from product data, such as Battery Passport. |
| Workspace | The web entry point to the Semantic Stack applications available for your tenant. |
| Aspect | A domain-specific view on the data of a digital twin, provided through an API and described by an Aspect Model. |
| Aspect Model | A formal, machine-readable description of the data that an aspect provides. Aspect Models follow SAMM. |
| Semantic Aspect Meta Model (SAMM) | The open specification for writing Aspect Models, part of the Eclipse Semantic Modeling Framework (ESMF). Formerly known as BAMM. |
| Aspect Model Editor | An open-source desktop tool to create and edit Aspect Models visually. |
| Aspect Model Catalog | The central place to store, version, and share Aspect Models. |
| Aspect on Demand | Turns data from the data lake into ready-to-use aspect APIs with a guided wizard, without coding. |
| Data Browser | Creates dashboards and charts from harmonized data through drag and drop. |
| Semantic Data Lakehouse | Managed, isolated data storage and processing that organizes data in raw, harmonized, and refined layers according to Aspect Models. |
| Digital Twin Registry | The central registry of all digital twins and their aspects, including how to access them. Certified by Catena-X. |
| Asset Administration Shell (AAS) | The standardized digital representation of an asset, defined by the Industrial Digital Twin Association (IDTA). Digital twins in the Digital Twin Registry are Asset Administration Shells. |
| Submodel | A part of an Asset Administration Shell that describes one aspect of an asset. |
| Asset | A physical object, such as a machine, sensor, or product, that is represented by one or more digital twins. |
| Product Type Twin | A digital twin of a product type, identified by its part number. |
| Product Instance Twin | A digital twin of a single product, identified by its serial number. |
| Local identifier | Key-value pairs that uniquely identify an asset within a tenant, such as a serial number. |
| Battery Passport | A solution that creates and manages digital battery passports as required by the EU Battery Regulation, with separate views for the public, interested parties, and authorities. |
| Product Passport | A solution, announced as coming soon, that creates Digital Product Passports for the EU Ecodesign for Sustainable Products Regulation (ESPR). |
| Catena-X | An open data space for the automotive industry for standardized, sovereign data exchange along the value chain. |
| ”As engineered”, “as produced”, “as operated” data | Product data from engineering, from manufacturing, and from use in the field. |
| Ontology | A formal description of the concepts of a domain and the relationships between them. |
| Knowledge graph | An ontology together with the data that instantiates its concepts. |
| SingleKey ID | The identity provider recommended for signing in to the Semantic Stack workspace. |
Abbreviations
Section titled “Abbreviations”| Abbreviation | Meaning |
|---|---|
| AAS | Asset Administration Shell |
| AI | Artificial intelligence |
| API | Application programming interface |
| BPS | Bosch Production System |
| DPP | Digital Product Passport |
| EDC | Eclipse Dataspace Connector, used for data exchange in Catena-X |
| ESMF | Eclipse Semantic Modeling Framework |
| ESPR | Ecodesign for Sustainable Products Regulation |
| HITL | Human in the loop |
| IDTA | Industrial Digital Twin Association |
| IoT | Internet of Things |
| KPI | Key performance indicator |
| LLM | Large language model |
| MCI | Manufacturing Co-Intelligence |
| MCP | Model Context Protocol |
| MDM | Master Data Management |
| MES | Manufacturing execution system |
| OEE | Overall equipment effectiveness |
| PLC | Programmable logic controller |
| RDF | Resource Description Framework |
| SAMM | Semantic Aspect Meta Model |
| SOT | Smart Operations Toolkit |
| SP | Service pack |
| TTL | Turtle, a text format for RDF data and Aspect Models |
© Robert Bosch Manufacturing Solutions GmbH, all rights reserved. Corporate Information Legal notice Data protection policy