Platform reference
Celonis
Architecture
Five-layer stack from data source to consumer — data flows top-to-bottom, value flows bottom-to-top.
01
Data Sources
SAP ECC / S/4HANA
Salesforce
Oracle Fusion
Snowflake
Databricks
Amazon S3
Microsoft Fabric
Google BigQuery
Kafka
REST APIs
SAP Ariba
ServiceNow
Workday
Coupa
02
Data Integration
Extractors / OPC
On-premise clients bridge SAP and legacy systems to the cloud platform securely via RFC
ETL Engine
SQL-based transformation engine (Vertica syntax) with delta and full-load modes
Replication Cockpit
Real-time CDC from SAP — near-zero latency event streaming for operational processes
Data Pools
Logical containers holding connections, data jobs, transformations and scheduling
03
Core Process Engine
Database
SaolaDB
In-memory columnar store combining relational and graph techniques, purpose-built for process mining at scale
columnar · in-memory · OLAP
Query Language
PQL
Process Query Language — SQL-inspired but process-native. Co-designed with SaolaDB for optimal execution
180+ operators
04
Semantic Layer — Process Intelligence Graph
Knowledge Models
Defines KPIs, filters, records and calculated attributes as reusable business logic
Objects & Events
Object-centric data model (OCDM) for multi-object process mining beyond case-centric logs
PI Graph
Process digital twin — the real-time operational model of the enterprise's running processes
Zero-copy connectors
Fabric, Databricks, Snowflake — data stays in place, metadata shared to Celonis via Delta Sharing
05
Consumers
Studio
Build views, dashboards, and process apps with no-code components driven by PQL
Action Flows
Low-code automation engine with 1,000+ connectors — trigger changes in SAP, Salesforce, and more
Orchestration Engine
Multi-step workflows combining Action Flows, human approvals, and AI tasks
Process Copilots
AI chat agents grounded in the PI Graph — answer process questions in natural language
AgentC
Feed PI context to Copilot Studio, Amazon Bedrock, Agentforce, and CrewAI via Intelligence API
ML Workbench
Python / SaolaPy environment — run ML models directly against SaolaDB without data export
Insight Explorer
AI-generated statistical insights — automatically surfaces subgroup deviations and anomalies
Power BI / Teams
Intelligence APIs expose KPIs and process metrics to external BI and collaboration tools