The argument Celonis is making is structural. AI models are increasingly being deployed in enterprise operations, but they reason against training data rather than against how a specific business actually runs. They do not know your process bottlenecks, your approval chains, or which system is the source of truth for a given data type. The result is AI agents that can hallucinate confident-sounding decisions that contradict operational reality.
The Celonis Context Model is positioned as the answer: a dynamic, real-time digital twin of business operations built on process data pulled from across an organisation's systems, applications, and devices. The CCM is designed to be open — any data source, model, or agent — and to learn continuously as operations change, rather than requiring periodic manual updates.
Ikigai Labs, the acquisition announced alongside, brings specific capabilities that complement this. Founded on research from MIT, the company has built tools for decision intelligence: modelling future-state scenarios, predicting where processes will break before they do, and running simulation and forecasting exercises. Devavrat Shah, Ikigai's co-founder, becomes Chief Scientist for Enterprise AI at Celonis.
The timing reflects where a number of large enterprises now find themselves: past the proof-of-concept phase with AI, confronting the question of whether and when to trust AI agents with operational decisions that have material consequences. Celonis has built its business on process mining — identifying inefficiencies in how work actually moves through an organisation — and the CCM is a direct extension of that into the AI orchestration layer.
Celonis is offering briefings with co-CEO Alex Rinke and Devavrat Shah for further detail.
