First PI, then AI: Celonis Puts Processes in the Right Context

AI projects fail if you don’t understand how your organization works. Celonis places process intelligence (PI) first.

Alex Rinke, one of the three founders and co-CEO of Celonis, immediately sets the tone during his opening speech. “AI is the dominant theme at every conference and in every boardroom. It is the only technology for which you are sure to get a budget. But something is missing, which means investments are not yet paying off. How is it that we cannot make the most influential technology of our time work for us?”

Celonis is now trying to answer that question. The rise of the German company reads like a Silicon Valley story. In 2011, the company originated as a spin-off from the Technical University of Munich and quickly grew into one of Germany’s fastest-growing startups. The connection to its hometown remains strong: Munich hosts the Celosphere event every year.

No AI without PI

After the hype of recent years, 2025 threatens to be the year of disappointment. MIT concludes that 95 percent of AI projects fail, and that is not the only alarming statistic. The explanation, according to Dan Brown, Chief Product Officer at Celonis, is not hard to find. “You cannot automate what you don’t understand yourself. How can you expect AI to change your organization if you don’t know what you want to change?”

Although numbers never lie, Celonis does not advocate abandoning all AI projects; on the contrary. “If you miss the first wave, you won’t benefit from the second wave either,” says Carsten Thoma, Chairman of Celonis. Celosphere’s slogan, ‘No AI without PI,’ leaves nothing to the imagination.

Celonis aims to make a difference with Deutsche Gründlichkeit. Thoma: “Enterprise AI is completely different from ChatGPT. It is not a standalone technology, but a framework where all elements must interact. People have the reflex to follow the loudest voice in the room. Let’s not jump hysterically on the hype and take a step back to see what’s truly important.”

Free the Process

The Celosphere event has a second slogan: ‘free the process‘. The slogan can be read everywhere on the exhibition floor, and even a ‘process march’ with much fanfare is organized. In sports and business operations, we often hear that we must trust the process, but according to Celonis, that is impossible if the process is not visible. Thoma: “AI must understand how your organization operates and what’s in your people’s minds. If the picture is not complete, you’re already starting on a bad foundation.”

This describes what Celonis essentially means by process intelligence. Celonis creates a ‘digital twin’ of your company. That has been the company’s specialty for fourteen years. Celonis started in process mining, which allows companies to create an ‘X-ray’ of their processes to identify pain points. The focus gradually shifted to process intelligence to map processes in real-time and proactively.

This provides companies with the right context, a word constantly heard during Celosphere, to orchestrate AI models. Rinke: “Your processes provide the best description of how your company is structured. We take the data from your systems and enrich it with your company’s unique context to reflect reality. This way, AI can be implemented in the right place. A successful enterprise is able to improve what it does every day and adapt to changes to innovate freely.”

Celonis takes the word ‘freeing’ very literally. It constantly emphasizes the openness of the ecosystem. Collaborations with Microsoft and Databricks, among others, to exchange data bi-directionally via a ‘zero-copy’ principle, highlight this. Thoma: “Your organization’s footprint is everywhere. If vendors shield their environment, customers suffer. Silos don’t give you answers.”

If vendors shield their environment, customers suffer.

Carsten Thoma, Chairman Celosphere

From As-is to To-Be

In his keynote, Brown provides a cross-section of the Celonis platform, which is built on three layers. The bottom layer is formed by Data Core (not to be confused with the company Datacore, ed.).“Data is the beginning of everything we do,” says Brown. “But the heart is the Process Intelligence Graph, where we add the context needed for AI to work. We hold up a system-agnostic mirror to your company.”

In the third layer, it is then up to the customers to work with that information. Divya Krishnan, VP Product Management, takes the floor. “You must build on a foundation. After an analysis, we let you create a blueprint to orchestrate AI with your processes and together with your people. From ‘as-is,’ you evolve to ‘to-be.’ There is only one thing better than solving problems, and that is being able to avoid them.”

Good Marriage

No major product announcements are to be found at Celosphere this year. Celonis says it is building on what was announced last year and wants to let customers speak. Furthermore, it paves the way for the AI agent train to roll in, as the launch of the Model Context Protocol server makes clear. “But we deliberately don’t call ourselves an AI company,” notes Eugenio Cassano, VP Strategy & Innovation. “We maintain our open mind that makes us unique.”

Process intelligence and AI will increasingly move towards each other, emphasizes Christian Tabois, who leads Celonis in the Benelux. “There is a movement towards automation. That is a logical next step: analysis provides insight into what goes wrong, but then you must also do something with it. Technology fails if people don’t use it. As we grow horizontally and vertically, we will encounter new parties. What we do feeds what the major hyperscalers deliver. We see a happy marriage there to accelerate the agentic roadmap.”

Will 2026 be the year of return on AI? If it’s up to Celonis, yes, but on the condition that AI tools are placed in the right context. First PI, then AI: that order is the key to success for Celonis.