AI is only as effective as the context it has access to
In the long effort to make artificial intelligence genuinely useful inside complex organizations, Celonis has identified a quiet but consequential failure: AI systems that see data without understanding the living context of a business. The company's newly unveiled Context Model, paired with its acquisition of decision intelligence firm Ikigai Labs, represents an attempt to close that gap — building a dynamic bridge between the messy reality of how enterprises actually operate and the structured language machines can reason from. It is, at its core, a wager that AI's value lies not in raw processing power, but in the quality of understanding it is given.
- Enterprise AI deployments are quietly failing because systems process data without grasping the operational context that gives that data meaning.
- Celonis is disrupting its own category by introducing the Context Model — a real-time digital twin that translates the full complexity of business operations into AI-readable intelligence.
- The acquisition of Ikigai Labs injects forecasting, simulation, and scenario-planning muscle, turning passive process awareness into active decision support.
- An MIT AI professor joining as chief scientist signals that the ambition here extends well beyond product feature — this is a bid to define the foundational layer of enterprise AI.
- With integrations already spanning AWS, Databricks, Microsoft Fabric, Oracle, and major ERP platforms, the Context Model is positioned to sit atop infrastructure companies already rely on.
- The full picture is expected to come into focus at Celonis's Next conference on May 19, as the Ikigai deal moves toward regulatory closure.
Celonis has built its identity around process intelligence — understanding how businesses actually run, not just how they are designed to run. Its latest move names a problem the industry has largely left unspoken: enterprise AI systems operate with an operational blind spot, seeing data but missing the context that makes data meaningful. The result is AI investment that underdelivers, not because the technology is weak, but because it is working from an incomplete picture.
The company's answer is the Celonis Context Model, a dynamic digital twin that assembles a real-time representation of an organization's operations — drawing from every process, application, device, and interaction — and renders it in a form AI can actually reason from. Rather than forcing businesses to simplify themselves for machines, the CCM translates in the other direction. President Carsten Thoma put it plainly: AI is only as effective as the context it can access, and until now, no system could provide that context completely.
To sharpen the capability further, Celonis announced the acquisition of Ikigai Labs, whose technology specializes in AI-driven decision intelligence — planning, simulation, and forecasting at scale. Where Celonis encodes how processes work, Ikigai models how they might fail and what to do about it. Ikigai cofounder and MIT professor Devavrat Shah, who will become Celonis's chief scientist for enterprise AI, called the combination the most complete operational representation of business reality yet assembled.
The deal awaits standard approvals, with deeper details promised at the Celonis Next conference on May 19. The Context Model will sit atop an integration layer already spanning AWS, Databricks, Microsoft Fabric, Oracle, and major ERP and CRM platforms — making it less a new system to adopt than a new layer of understanding placed over infrastructure organizations already depend on.
Celonis, the process intelligence company, has unveiled a new technical layer designed to solve a fundamental problem in enterprise artificial intelligence: the blind spot. When companies deploy AI systems to run their operations, those systems often lack a complete picture of how the business actually works. They see data, but not context. They process information, but not meaning. Celonis calls this gap the operational blind spot, and it's the reason many AI investments fail to deliver the returns companies expect.
The company's answer is the Celonis Context Model, or CCM. Rather than asking businesses to reshape themselves around what AI can understand, the CCM translates the other direction—it takes the messy, interconnected reality of how a company operates and renders it in a language artificial intelligence can comprehend. The system builds a dynamic digital twin of operations in real time, pulling data from every process, every application, every device, every interaction across the enterprise. It then layers in business knowledge and operational intelligence, creating what Celonis describes as a unified foundation for AI to actually work from.
Carsten Thoma, the company's president, framed the problem simply: AI is only as effective as the context it has access to. Until now, he said, no organization could give its AI systems a truly complete, dynamic model of how their business actually functions. The CCM changes that equation.
To strengthen the capability, Celonis announced it has signed a definitive agreement to acquire Ikigai Labs, a company focused on AI-driven decision intelligence. Ikigai brings proven technology for working with large-scale structured data, along with capabilities in planning, simulation, and forecasting. The combination is meant to be powerful: Celonis has encoded how business processes work; Ikigai has built systems that can model future scenarios, predict where processes will fail, and help organizations make decisions with confidence. Devavrat Shah, Ikigai's cofounder and an MIT professor of artificial intelligence, will serve as chief scientist for enterprise AI at Celonis. He described the pairing as offering "the most complete operational representation of business reality."
The acquisition is expected to close soon, pending standard regulatory and procedural approvals. Celonis plans to share more details about both the deal and the Context Model at its Next conference on May 19, and at the upcoming Process Intelligence Day event.
The Celonis platform itself already connects to much of the enterprise technology stack—it integrates without copying data from AWS, Databricks, and Microsoft Fabric, with Snowflake coming soon. It has preconfigured connectors to major systems of record like Oracle, as well as leading ERP and CRM platforms. The Context Model will sit atop this infrastructure, translating between the world of business and the world of machines, so that when AI makes a decision, it's making it from a place of actual understanding.
Citas Notables
AI is only as effective as the context it has access to. Organizations must equip their AI with a holistic, dynamic model reflecting how the business actually works.— Carsten Thoma, president of Celonis
Ikigai Labs has proven foundation model technology for large-scale structured data; Celonis has encoded business processes. Together, we offer the most complete operational representation of business reality.— Devavrat Shah, cofounder of Ikigai Labs and MIT professor of artificial intelligence
La Conversación del Hearth Otra perspectiva de la historia
So the problem Celonis is solving—is it that AI systems don't have enough data, or that they have data but can't interpret it?
It's the second one. They have plenty of data. The problem is that data by itself doesn't tell you what's actually happening in the business. You need context—the rules, the relationships, the way things really work.
And the Context Model provides that by building a digital twin. But why is that so hard to do?
Because a business isn't a single system. It's Oracle talking to Salesforce talking to a spreadsheet someone maintains in accounting. It's formal processes and workarounds. It's what the handbook says and what people actually do. Pulling all of that together and translating it into something an AI can reason about—that's the hard part.
So Ikigai Labs is being acquired to add what, exactly?
The ability to look forward. Celonis understands what's happening now. Ikigai brings simulation and forecasting—the ability to say, if we change this, what breaks? What should we do differently?
That sounds like it could be genuinely useful.
It could be. If it works. The real test is whether organizations can actually use this to make better decisions, not just better predictions.