Glossary

Plain-English definitions of the enterprise integration, BPM and AI terms we work with every day.

AI Agent

An AI agent is a goal-driven system that uses a language model to plan, make decisions and take actions through tools, often with limited autonomy.

An AI agent goes beyond answering a single prompt: it pursues a goal by planning steps, calling tools and APIs, retrieving information, and adapting based on results — sometimes coordinating with other specialist agents. Production-grade agents need clear architecture, scoped tool permissions, observability and human-in-the-loop checkpoints. Connected to enterprise systems via APIs, middleware and MCP, agents can automate content pipelines, intelligence hubs and multi-step workflows.

API-Led Connectivity

API-led connectivity is an integration approach that exposes data and capabilities through reusable, layered APIs (system, process and experience).

API-led connectivity organises integration into three tiers of reusable APIs: system APIs unlock data from underlying systems, process APIs orchestrate and combine that data into business logic, and experience APIs tailor it for specific channels or apps. Each layer is independently governed and reusable, which speeds delivery and reduces duplicated point-to-point work. It is the dominant model behind modern integration platforms and the composable enterprise.

AS2

AS2 is a secure internet transport protocol for exchanging EDI and other business data with encryption, digital signatures and delivery receipts.

Applicability Statement 2 (AS2) carries business documents over HTTP/S with S/MIME encryption and digital signatures, plus signed Message Disposition Notifications that act as non-repudiable delivery receipts. It became a de facto standard for B2B EDI exchange — notably mandated by large retailers — because it provides security and proof of delivery over the public internet without a dedicated value-added network.

Business Process Management (BPM)

BPM is the discipline of modelling, automating, executing and continuously improving an organisation's business processes.

Business Process Management treats work as end-to-end processes that can be designed, measured and optimised. A BPM platform lets teams model a workflow (often in BPMN notation), automate the routing of tasks and data between people and systems, handle exceptions, and monitor performance against SLAs. Done well, BPM turns manual, undocumented handoffs into auditable, repeatable, continuously improved workflows. It pairs naturally with integration, which supplies the data and system connectivity processes depend on.

Camunda BPM

Camunda is an open-source process automation platform that executes BPMN workflows and DMN decisions for developer-centric process orchestration.

Camunda is a widely used process automation and orchestration engine that runs standards-based BPMN process models and DMN decision tables. It gives developers a code-friendly way to automate end-to-end processes, coordinate microservices and human tasks, and gain real-time visibility into running workflows. It is a common choice for business process automation in integration projects where transparency, exception handling and auditability matter.

Electronic Data Interchange (EDI)

EDI is the structured, standardised electronic exchange of business documents — such as orders and invoices — between trading partners.

Electronic Data Interchange replaces paper and email with machine-readable documents in agreed standards (EDIFACT, ANSI X12, etc.), letting partners' systems exchange purchase orders, invoices, shipping notices and more without manual rekeying. EDI is foundational to supply chain, retail, logistics and healthcare. Onboarding partners involves mapping their formats to internal data and enforcing compliance, often over transports like AS2.

Enterprise Application Integration (EAI)

EAI is the practice of connecting an organisation's separate software applications so they share data and processes as one coordinated system.

Enterprise Application Integration connects systems such as ERP, CRM, SCM, legacy databases and cloud SaaS that were never designed to work together. Rather than rebuilding applications, EAI adds an integration layer that moves and transforms data between them — through messaging, APIs, events or file transfer. The goal is to eliminate data silos, automate cross-system processes and give the business a single, consistent view of its operations.

Enterprise Service Bus (ESB)

An ESB is a middleware architecture that routes, transforms and mediates messages between applications through a central communication backbone.

An Enterprise Service Bus provides a shared backbone for integration: applications connect once to the bus instead of point-to-point to every other system. The ESB handles message routing, protocol and format transformation, and orchestration of services. It reduces the brittle 'spaghetti' of direct connections and centralises monitoring and governance. ESBs are a common implementation of service-oriented architecture, though many organisations now complement or replace them with API gateways and event streaming.

ETL (Extract, Transform, Load)

ETL is a data integration process that extracts data from sources, transforms it into a target format, and loads it into a destination system.

Extract, Transform, Load moves data from operational systems into a warehouse, lake or another application. Extraction pulls raw data; transformation cleans, validates, reshapes and enriches it; loading writes it to the destination. ETL traditionally runs in scheduled batches, while the related ELT pattern loads first and transforms in the target. Both are core to analytics, reporting and feeding data to machine-learning pipelines.

Event-Driven Architecture (EDA)

EDA is a design pattern where systems communicate by producing and reacting to events rather than calling each other directly.

In an Event-Driven Architecture, a change of state — an order placed, a payment cleared — is published as an event that any interested system can consume asynchronously. Producers and consumers are decoupled: they need not know about each other or be available at the same time. This enables real-time responsiveness, elastic scaling and resilience. Event brokers and streaming platforms such as Apache Kafka or message queues carry the events between services.

iPaaS (Integration Platform as a Service)

iPaaS is a cloud-hosted platform that provides tools to build, deploy and manage integrations between applications and data without managing infrastructure.

Integration Platform as a Service delivers integration capability — connectors, data mapping, workflow, API management and monitoring — as a managed cloud service. Teams build integrations in the platform rather than installing and operating middleware themselves. iPaaS suits hybrid landscapes that mix SaaS and on-premises systems, and accelerates delivery with prebuilt connectors, though enterprises weigh it against self-managed platforms for high-volume, mission-critical flows.

Large Language Model (LLM)

An LLM is an AI model trained on vast text data to understand and generate natural language, powering chat, summarisation and reasoning tasks.

Large Language Models are neural networks trained on extensive text corpora to predict and generate language. They underpin conversational assistants, document processing, summarisation, classification and code generation. In the enterprise, LLMs are most valuable when integrated with reliable data and tools — through retrieval-augmented generation, function calling and protocols like MCP — so their output is grounded, governed and connected to real systems rather than used in isolation.

Message-Oriented Middleware (MOM)

MOM is infrastructure that lets applications communicate by sending and receiving messages through queues or topics, asynchronously and reliably.

Message-Oriented Middleware decouples applications in time and space: a sender places a message on a queue or topic and continues working, while the receiver consumes it when ready. The middleware guarantees delivery, ordering and buffering even if one side is temporarily unavailable. This underpins both classic integration and event-driven systems. Common implementations include JMS providers, IBM MQ, RabbitMQ and Apache Kafka.

Middleware

Middleware is software that sits between applications and operating systems to enable communication, data management and integration across them.

Middleware is the connective software layer that lets distributed applications and systems work together — handling messaging, data transformation, transactions, authentication and orchestration so individual applications don't have to. Integration middleware (ESBs, message brokers, integration platforms) is the category most relevant to enterprise integration, providing the plumbing that moves and reshapes data reliably between heterogeneous systems.

MLOps (Machine Learning Operations)

MLOps applies DevOps-style practices to machine learning — versioning, automated training, deployment, monitoring and retraining of models in production.

MLOps is the set of practices and tooling that take machine-learning models from notebooks to reliable production systems. It covers data and model versioning, reproducible training pipelines, automated deployment, and ongoing monitoring for performance and data drift so models can be retrained before they degrade. MLOps is what keeps AI solutions accurate, maintainable and trustworthy once they are integrated into business processes.

Model Context Protocol (MCP)

MCP is an open standard that lets AI models connect to external tools and data sources through a consistent, auditable interface.

The Model Context Protocol standardises how AI agents and applications communicate with external tools, data and services. Instead of bespoke glue code for each integration, systems expose capabilities through MCP servers that any MCP-aware model can use over a common interface. For the enterprise this means agents can reach CRM, ERP and analytics through standardised, governed, auditable connections — bringing integration discipline to AI agent development.

Process Mining

Process mining analyses event-log data from systems to discover how processes actually run, revealing bottlenecks and deviations from the intended design.

Process mining reconstructs real process behaviour from the event logs that enterprise systems already produce. Algorithms reveal the actual paths cases take — including variants, rework loops and bottlenecks — versus the designed 'happy path'. This evidence base lets organisations target improvements, predict SLA breaches and validate that automation delivers the intended outcomes. It is a natural complement to BPM and to AI-enhanced process intelligence.

Retrieval-Augmented Generation (RAG)

RAG is a technique that grounds a language model's answers in retrieved, authoritative documents rather than relying solely on its training data.

Retrieval-Augmented Generation improves the accuracy and currency of large language models by retrieving relevant content — from a knowledge base, document store or database — and supplying it to the model as context when generating an answer. This reduces hallucination, lets responses cite up-to-date and proprietary sources, and keeps sensitive data in systems you control. Agentic RAG extends this with planning, multi-source routing and iterative refinement.

Service-Oriented Architecture (SOA)

SOA is an architectural style that structures software as reusable, loosely coupled services accessed over a network through defined interfaces.

Service-Oriented Architecture decomposes capability into discrete services — each with a contract describing what it does — that other systems call without knowing the internal implementation. Loose coupling means services can evolve independently, and reuse reduces duplication across the enterprise. SOA is often realised with an enterprise service bus for mediation. Modern API-led and microservice approaches share its core principle of well-defined, reusable, independently deployable services.

Software AG webMethods

webMethods is an enterprise integration platform providing an ESB, API management, B2B/EDI, managed file transfer and process automation.

Software AG webMethods is a mature, comprehensive integration suite used by large enterprises for application integration (ESB), API management, B2B and EDI partner integration, managed file transfer and BPM. Its strengths are breadth of capability on a single platform, proven scale and vendor support; the trade-offs are licensing cost and a specialist skill set. KONDEVS has delivered webMethods integration and BPM projects across European finance, rail, insurance and logistics.