Enterprise AI does not have a model problem as much as it has an operating-context problem. Companies can deploy copilots, agents, and LLMs, but those systems often lack a reliable view of how work actually moves across ERP, CRM, supply chain, finance, service, and line-of-business systems.
Process mining and intelligence platform Celonis is aiming directly at that gap. The company launched the Celonis Context Model and acquired Ikigai Labs, an AI decision-intelligence company of which capabilities include planning, simulation, and forecasting. Together, the moves position Celonis less as a process-mining vendor that explains what happened and more as a provider of operational context for AI systems that need to reason, recommend, and eventually act.
That makes the announcement relevant for ERP leaders. ERP Today recently covered why AI agents need more than direct access to ERP data before they can be trusted inside core workflows. Celonis is arguing for one of the missing layers in that architecture: a living model of business operations that gives AI agents process context before they touch execution.
Context Is the New AI Control Point
Celonis describes the Context Model as a dynamic, real-time digital twin of operations that translates how a business works into a form AI can use.
The idea is straightforward but difficult to execute. AI systems need to understand not only records and fields, but process flow, dependencies, business rules, bottlenecks, historical patterns, exceptions, and likely outcomes. In ERP terms, that means an agent should not only see an invoice, purchase order, shipment, or customer record. It should understand where that item sits in the broader process and what could happen if action is taken.
That is where Celonis’ process-mining background gives it an advantage. The company already works by extracting event data from systems of record and mapping how processes run in practice. CCM extends that into an enterprise AI context layer that Celonis says can combine process data, business knowledge, operational intelligence, and decision intelligence.
For ERP buyers, the important question is whether that context layer becomes part of the control architecture for agents. A finance agent, procurement assistant, or supply chain copilot should not be allowed to act only because it can access an API. It needs to understand the workflow, constraints, approvals, exceptions, and downstream risk attached to that action.
Analysis
What this means: AI agents need operational context before they can earn trust. Celonis’ Context Model shows how process intelligence may become a control layer between ERP systems and AI agents. ERP leaders should evaluate whether agentic AI tools understand process flow, constraints, exceptions, and approvals before allowing them to recommend or trigger actions.
Ikigai Adds Foresight Layer
Ikigai Labs gives Celonis a stronger decision-intelligence story. Celonis says Ikigai brings capabilities in planning, simulation, forecasting, causal inference, and large-scale simulation. The acquisition gives Celonis exclusive rights to MIT-owned patents that Ikigai had licensed, with MIT becoming a Celonis shareholder as part of the agreement.
That matters because process intelligence has often been strongest at explaining the past and diagnosing the present. Enterprise AI needs more than hindsight. It needs to test possible actions before recommending them.
Simulation does not remove the need for human approval, but it does give teams a stronger basis for deciding when automation should act, pause, escalate, or recommend alternatives.
Analysis
What this means: Simulation will become part of enterprise AI governance. Ikigai Labs gives Celonis planning, forecasting, and simulation capabilities that can help teams test likely outcomes before automating decisions. CIOs and process owners should prioritize AI use cases where “what-if” analysis can reduce operational risk before agents act in live workflows.
Sponsor Industry‑Grade Research
Process Mining and Agent Governance
Celonis is entering a crowded conversation about enterprise AI context. Cloud vendors, ERP providers, data platforms, and AI orchestration tools are all trying to define the context layer for agents. Celonis’ argument is context cannot come only from documents, databases, or metadata. It has to come from how work actually happens across systems.
As companies move from AI pilots into production workflows, agents recommend process changes or trigger actions will need a much stronger evidence base. They need to know which process variant is running, which policy applies, which exceptions matter, and which systems will be affected next.
For CIOs and enterprise architects, that reframes process mining. It is not only a diagnostic tool for transformation programs. It can become part of the infrastructure that helps determine whether an AI recommendation is relevant, safe, and executable.
Still, Celonis has to prove how broadly its model can scale across real enterprise landscapes. Context is only as strong as the systems, processes, and data it covers. If process data is incomplete, inconsistent, or poorly governed, the context layer can create false confidence.
ERP leaders should evaluate Celonis against specific process scenarios rather than generic AI posturing. Good pilots should start with measurable workflows such as invoice exceptions, claims routing, procurement bottlenecks, inventory rebalancing, service scheduling, or order delays. The test is whether Context Model can connect data, process logic, decision rules, and simulation in a way that improves outcomes without weakening controls.
Analysis
What this means: Process mining is moving from transformation insight to AI infrastructure. Celonis is positioning process intelligence as a foundation for reliable enterprise AI, not only as a tool for finding bottlenecks. ERP teams should treat process coverage, data quality, and governance as prerequisites for scaling context-aware AI across finance, supply chain, service, and operations.
