Boomi Professional key terms

The core Boomi design artifact that defines an integration flow. It contains shapes connected by paths that move data from a start shape through transformations and logic to an end shape.

A single step inside a process, such as a connector, map, decision, or branch. Each shape performs one action and passes documents to the next shape along a configured path.

The runtime route a document takes through a process, determined by which paths and shapes execute. Tracing document flow helps diagnose where data was transformed, split, or dropped.

A process shape that evaluates a condition against document data or process properties and routes execution down one of several possible paths based on the result.

A shape that splits a single incoming flow into multiple parallel paths so several operations can run on the same document. Results from each branch can later be combined or processed independently.

A shape that wraps a segment of a process so that errors raised inside it are caught and handled by a separate error path instead of failing the whole execution.

A named variable scoped to a process that can be set and read by shapes at runtime. It is commonly used to pass values, flags, or counters between steps.

A process property whose value is loaded at runtime rather than at design time, often from an environment component. It lets one deployed process behave differently per environment without edits.

A reusable component that lets a process read from or write to a specific system, protocol, or service, such as a database, file server, or cloud application. It handles connection and operation details.

A stored configuration holding the credentials, URL, and settings needed to reach a target system. Connectors reference it so credentials are managed in one place rather than inside each process.

A defined action within a connector, such as a query, get, create, or update, along with its request and response profiles. Processes call operations to interact with the target system.

The lightweight runtime engine that executes deployed processes and hosts connectors. Atoms can run on-premises or in a cloud-hosted environment and are the basic unit of execution.

A clustered runtime made of two or more nodes that share workload and provide high availability. If one node fails, others continue processing so integrations keep running.

An Atom hosted and managed by the platform vendor rather than the customer. It reduces infrastructure work but may limit access to on-premises systems without additional networking.

A runtime component that sits on a local network and brokers requests between cloud-hosted processes and on-premises systems. It avoids exposing internal systems directly to the internet.

A named configuration set that supplies environment-specific values, such as URLs or credentials, to deployed processes. It allows one packaged process to run correctly across test and production.

A versioned, deployable snapshot of a process or related components. Packaging freezes the design so the same artifact can be promoted through environments predictably.

The act of installing a packaged component onto a runtime and attaching the correct environment configuration. It makes the integration available to execute on that Atom or Molecule.

A structured definition of a data format, such as XML, JSON, flat file, or database, describing its elements, types, and hierarchy. Profiles are the basis for mapping and validation.

A single node within a profile, representing a field, record, or group with a data type and optionality. Elements define what data a profile can carry.

A transformation component that defines how source profile elements correspond to target profile elements, including functions and default values. It converts data from one structure to another.

A built-in or custom operation used inside a map to transform values, such as concatenation, lookup, date formatting, or scripting. Functions let one source value become a derived target value.

The use of embedded scripts, such as JavaScript or Groovy, inside a map or process to perform logic that standard functions cannot express. It adds flexibility at the cost of maintainability.

Metadata attached to a document as it flows through a process, such as its source, tracking ID, or custom values. Properties travel with the document and can be read by later shapes.

A shape that runs a configured data processing step, such as combining, splitting, or sorting documents, without calling an external system. It shapes document sets before they reach connectors.

The operation of merging multiple documents into a single document, often to batch records before sending them to a target system. It reduces call volume and can improve throughput.

A design artifact that exposes a process or data operation as a REST or SOAP endpoint. It defines resources, methods, and request and response handling for external consumers.

A single method and path defined within an API component, such as GET /orders or POST /customers. Each operation specifies how requests are validated and which process handles them.

The runtime layer that receives API requests, enforces policies such as authentication and rate limits, and forwards traffic to the backing process or service. It centralizes control over API traffic.

An API configuration that forwards requests to an existing backend service rather than a process, optionally applying policies along the way. It lets teams govern third-party or legacy APIs without rebuilding them.

A policy that caps how many requests a client may make in a time window, protecting backend systems from overload. It is commonly enforced at the API gateway layer.

A configurable rule applied to API traffic, covering concerns such as authentication, throttling, transformation, or logging. Policies are attached to APIs and enforced at runtime.

An authorization framework in which a client obtains a token from an authorization server and presents it to access a protected resource. It avoids sharing user credentials directly with each API.

An AI-driven component that perceives input, reasons about a goal, and chooses actions such as calling tools or APIs to complete a task. Agents can chain steps with limited human intervention.

The mechanism by which a model requests that an external function or API be invoked with specific arguments, then uses the returned result to continue reasoning. It connects AI reasoning to real systems.

The coordination of multiple agents or steps so that tasks are routed, sequenced, and combined into a coherent outcome. It manages handoffs, retries, and shared context between agents.

A machine learning model trained on large text corpora to understand and generate language. In integration platforms it powers tasks such as classification, extraction, and natural-language interfaces.

The instruction and context supplied to a language model for a single request. Prompt design strongly influences output quality, format, and reliability in automated workflows.

A numeric vector representation of text or other content that captures semantic meaning. Embeddings enable similarity search, clustering, and retrieval by comparing vector distances.

A data store optimized for saving embeddings and performing fast similarity searches. It underpins retrieval features that let models ground answers in an organization's own content.

A pattern where relevant documents are retrieved, often via vector search, and inserted into a model prompt so answers are grounded in specific source content. It reduces unsupported claims.