Skip to main content

Industry reference architectures

This page provides industry-specific reference architectures that illustrate how data can be ingested, processed, and consumed within the Celonis Platform. By modeling the data landscapes most commonly found in different industry, these references can help you understand how Celonis fits into your existing environment.

Important

Any references to third-party products or services do not constitute Celonis Product Documentation nor do they create any contractual obligations. This material is for informational purposes only and is subject to change without notice.

Celonis does not warrant the availability, accuracy, reliability, completeness, or usefulness of any information regarding the subject of third-party services or systems.

Architecture layers explained

This section explains the structure of the reference architectures, and provides a high-level overview of elements that make up the layers. The diagrams on this page should be read from the Data Sources bottom layer to the top layer Consumers, reflecting how data flows through the platform. This perspective makes it easier to understand how raw data is ingested and transformed into insights and actions within the Celonis Platform.

Example reference architecture diagram showing four horizontal layers—Data Source, Enterprise Data Lake, Semantic Layer, and Consumers—with an additional vertical AI Infrastructure layer. Each layer includes the corresponding Celonis Platform components.

Data Source

Example diagram of the Data Source layer in industry-specific reference architectures, showing Celonis-supported extraction patterns and the common data sources that use each pattern.

The Celonis Platform is system-agnostic when it comes to data ingestion, allowing you to work with the data sources you already use. A wide range of extractors are available to help ingest your data, including pre-built native extractors, customizable templates, and options to build your own extractors tailored to specific needs. The Celonis Platform also supports secure authentication and transport methods for extracting data, ensuring flexibility without compromising security.

In each industry-specific reference , you will find examples of data sources commonly used in the specific industry, along with their most common extraction patterns. The following sections explain the extraction patterns Celonis supports.

Celonis Native Extractors

Celonis Native Extractors are pre-built connectors developed by Celonis to integrate directly with various enterprise systems. These extractors are tailored for specific source systems and help streamline the data ingestion process:

  • Cloud-based applications: Celonis offers native extractors for a wide range of cloud-based applications, including Oracle EBS, Coupa, Salesforce, SAP Ariba, SAP S/4HANA Public Cloud, ServiceNow, and others. These extractors support various features like pseudonymization, parallelization of requests, table configuration (renaming, joins, filtering), column selection, data type casting, and extractor execution configurations (batch size, partitioning).

  • SAP ECC and S/4HANA: For SAP systems, Celonis provides a dedicated (on-premise) SAP extraction client. This client acts as middleware between the Celonis Platform and SAP, monitoring tables for changes, and supporting real-time extractions for continuous data flow.

    When records are updated, their IDs are logged, and only these updated records are extracted in subsequent loads. The process involves the Celonis Platform defining extraction tasks, the SAP extraction client fetching these requests, and the Celonis RFC module within SAP extracting data to CSV files, which are then converted to parquet and uploaded to Celonis.

JDBC Extractors

For connecting to SQL databases, the Celonis Platform utilizes JDBC Extractors. This method is suitable for a variety of applications, including those from providers like Infor LN, BlueYonder, IBM AS/400, QAD, and PEGA. There are two primary connection types:

  • Direct connections: Using a direct connection enables the Celonis Platform to access your database without additional infrastructure. You do not need to install, patch, or maintain on-premises extractors, which reduces complexity and simplifies operations.

  • Uplink connection via on-premise extractors: When direct access is restricted or not preferred, you can install and use an on-premise extractor to poll job requests from the Celonis Platform, execute SQL queries against your database to retrieve the required data, and then securely send it back to Celonis.

    Note

    Data security is maintained through HTTPS-encrypted communication via TLS 1.2, in-transit encryption (often managed by the database provider's JDBC driver), and data pseudonymization performed by the extractor.

You can further customize these connections using custom JDBC strings for advanced settings like SSL and certificates, or by providing custom JDBC drivers for uplink connections. You can download the JDBC extractor from the Celonis Download Portal.

Extractor Builder

The Extractor Builder is a versatile tool that allows you to leverage and extend pre-built extractors, or create custom extractors to connect source systems to the Celonis Platform. It is effective for integrating with systems that expose REST and OData APIs and return JSON and XML responses. Systems commonly connected via Extractor Builder include SAP IBP, SAP SuccessFactors, and Workday.

You can use the Extractor Builder in two primary ways:

  • Customizing pre-built extractors: Celonis provides many pre-built extractors that can be customized via the Extractor Builder.

  • Creating custom extractors: If a pre-built extractor is not available, you can build custom connectors. This involves defining the extractor connection parameters (such as API URLs, placeholders, and default values), setting up authentication methods, and defining API endpoints (including requests, responses, headers, and error handling).

Tip

Extractor Builder configurations are portable and can be exported or imported as JSON files. This makes it easy to replicate and standardize configurations across data pools and teams, while also serving as a reliable backup mechanism.

Indirect Data Warehouse and ETL Push-Based Data Ingestion

The indirect data ingestion method, specifically through the Standard Data Ingestion API, provides a mechanism for pushing real-time data to the Celonis Platform from your existing systems and applications. It is often used as a secondary or alternative way to access data for various source systems. Key characteristics of this ingestion method include:

  • AWS S3 compatibility: The Standard Data Ingestion API is compatible with AWS S3, automatically processing files upon arrival.

  • File format: It exclusively uses uncompressed Parquet files. Any other file formats must be converted to Parquet before ingestion.

  • Delta loads: When primary keys are defined, the API always performs delta loads, ensuring only updated information is processed.

  • File uploads: Supports both single and multiple file uploads simultaneously, processed using a first-in, first-out (FIFO) method.

  • Data types: Accommodates both flat and nested data structures.

Note

* This extraction pattern can be used as an alternative for sources whose data can be output in, or converted into, Parquet format.

Enterprise Data Lake

Example diagram of the Enterprise Data Lake layer in industry-specific reference architectures, showing common data lake technologies and the patterns used to process data into the Celonis Process Landing Zone.

The Enterprise Data Lake serves as the central repository for raw or minimally processed data within your architecture. It supports high-volume data storage and acts as a foundation for data transformation into the Semantic Layer. Key components and methodologies in this layer include:

  • Zero data copy: Celonis can establish zero-copy connectivity with platforms like Microsoft Fabric, allowing data to be shared with the Celonis Platform without physical duplication. This ensures zero latency for data updates, embeds Celonis as a native workload, and preserves data governance by keeping data in its original location. After establishing access, Celonis registers table metadata to create a read-only data connection to your data lakehouses.

  • Data platforms (JDBC): Celonis can establish robust connections to your data platforms like Amazon Athena, Databricks, Snowflake, and Google BigQuery for scalable data storage and processing.

  • Process Landing Zone: This area within the Celonis Platform is designated for staging and preparing data before it undergoes further transformation into the Process Intelligence Graph.

Semantic Layer

Example diagram of the Semantic Layer in industry-specific reference architectures, comparing common third-party semantic layer technologies on the left with the Celonis Process Intelligence Graph.

The Semantic Layer is where your data is curated and transformed into an intelligent, actionable format for process intelligence. At its core is the Process Intelligence Graph, which enables object-centric process mining.

Object-centric process mining moves beyond traditional case-centric approaches by modeling the relationships between various objects (e.g., orders, invoices, products) and events that occur across your business processes. This approach helps creates a digital twin of your organization, offering a flexible and realistic view of complex and interconnected operations. It also allows you to analyze processes from any perspective without the need to re-extract or re-transform data.

Additionally, Celonis provides pre-built object types, event types, and relationships for common business processes like Accounts Payable, Accounts Receivable, Order Management, Procurement, and Inventory Management. You can also extend these or create custom object types, event types, and perspectives to complete the digital twin of your business based on your needs.

The curated and harmonized process data in the Process Intelligence Graph integrates on the same level as your other Semantic Layers in various modern data platforms such as Databricks, Snowflake, Google BigQuery, Microsoft Fabric (or OneLake), and Amazon Athena.

AI Infrastructure

Celonis offers an AI Infrastructure layer that serves as the core AI infrastructure for agents and agentic AI, encompassing capabilities such as LLMs and RAG. These platform-integrated features are designed to deliver functionalities at the Consumer layer, which you can then utilize.

Additionally, you can bring your own AI infrastructure (BYOM), and Celonis can integrate with it in a seamless manner. This flexible approach ensures your organization can leverage both Celonis-native AI and your own AI investments in a unified way.

Example diagram of the AI Infrastructure in an industry-specific reference architecture, showing common AI technologies and their associated models.

Consumers

Example diagram of the Consumers layer in an industry-specific reference architecture, with common third-party consumers on the left and Celonis consumers on the right, connected by arrows indicating bi-directional integration.

The Consumers layer leverages the intelligence generated in the Semantic Layer, and delivers it to your users through various interfaces and capabilities. These include Celonis Platform features such as context-specific Process Copilots, AI Annotators, Model-Based Alignment, Process Discovery, Machine Learning Workbench, and Automation (Action Flows), and Orchestration Engine.

The Celonis Platform integrates seamlessly with popular business intelligence platforms like Power BI, automation platforms like Power Platform, digital workplace experiences like Microsoft Teams, and agentic platforms for autonomous workflows such as Copilot Studio and Azure AI Foundry. This seamless integration is achieved through specialized APIs (Intelligence APIs, Tools, and Chat APIs, etc.), allowing third-party components to directly consume the Process Intelligence Graph.

Industry-specific references

Different industries rely on distinct combinations of business applications, data systems, and integration patterns. To support these unique landscapes, Celonis provides industry-specific reference architectures. Each architecture illustrates how the Celonis Platform connects with the typical data sources and operational systems found in that industry, such as ERPs, CRMs, and supply chain platforms.

The following sections present these architectures with diagrams that highlight how Celonis integrates into each industry environment.

Automotive reference

Diagram of an Automotive-specific reference architecture with four horizontal layers—Data Source, Enterprise Data Lake, Semantic Layer, and Consumers—and an additional vertical AI Infrastructure layer. Each layer includes the Celonis Platform components it comprises.

Note

This reference architecture is provided as an example, based on commonly used technologies in this industry. If one of your systems is not shown, Celonis offers multiple extraction methods that can be tailored to your environment. For more information on extraction patterns, see Data Sources layer.

Banking reference

Diagram of a Banking-specific reference architecture with four horizontal layers—Data Source, Enterprise Data Lake, Semantic Layer, and Consumers—and an additional vertical AI Infrastructure layer. Each layer includes the Celonis Platform components it comprises.

Note

This reference architecture is provided as an example, based on commonly used technologies in this industry. If one of your systems is not shown, Celonis offers multiple extraction methods that can be tailored to your environment. For more information on extraction patterns, see Data Sources layer.

Consumer Package Goods reference

Diagram of a CPG-specific reference architecture with four horizontal layers—Data Source, Enterprise Data Lake, Semantic Layer, and Consumers—and an additional vertical AI Infrastructure layer. Each layer includes the Celonis Platform components it comprises.

Note

This reference architecture is provided as an example, based on commonly used technologies in this industry. If one of your systems is not shown, Celonis offers multiple extraction methods that can be tailored to your environment. For more information on extraction patterns, see Data Sources layer.

Energy reference

Diagram of an Energy-specific reference architecture with four horizontal layers—Data Source, Enterprise Data Lake, Semantic Layer, and Consumers—and an additional vertical AI Infrastructure layer. Each layer includes the Celonis Platform components it comprises.

Note

This reference architecture is provided as an example, based on commonly used technologies in this industry. If one of your systems is not shown, Celonis offers multiple extraction methods that can be tailored to your environment. For more information on extraction patterns, see Data Sources layer.

Life Sciences reference

Diagram of a Life sciences-specific reference architecture with four horizontal layers—Data Source, Enterprise Data Lake, Semantic Layer, and Consumers—and an additional vertical AI Infrastructure layer. Each layer includes the Celonis Platform components it comprises.

Note

This reference architecture is provided as an example, based on commonly used technologies in this industry. If one of your systems is not shown, Celonis offers multiple extraction methods that can be tailored to your environment. For more information on extraction patterns, see Data Sources layer.

Manufacturing reference

Diagram of a Manufacturing-specific reference architecture with four horizontal layers—Data Source, Enterprise Data Lake, Semantic Layer, and Consumers—and an additional vertical AI Infrastructure layer. Each layer includes the Celonis Platform components it comprises.

Note

This reference architecture is provided as an example, based on commonly used technologies in this industry. If one of your systems is not shown, Celonis offers multiple extraction methods that can be tailored to your environment. For more information on extraction patterns, see Data Sources layer.

Retail reference

Diagram of a Retail-specific reference architecture with four horizontal layers—Data Source, Enterprise Data Lake, Semantic Layer, and Consumers—and an additional vertical AI Infrastructure layer. Each layer includes the Celonis Platform components it comprises.

Note

This reference architecture is provided as an example, based on commonly used technologies in this industry. If one of your systems is not shown, Celonis offers multiple extraction methods that can be tailored to your environment. For more information on extraction patterns, see Data Sources layer.