At Celosphere 2025, I spoke with Wil van der Aalst — Chief Scientist at Celonis and one of the original creators of process mining — about how process mining has evolved from an academic field into a core enabler of AI-powered business transformation.
He traced the technology’s origins from the late 1990s to its current role as the backbone of “process intelligence” — a discipline that fuses data, process context, and machine learning to enable prediction, automation, and continuous improvement. He also described how process mining provides the guardrails for AI, powering digital twins, intelligent agents, and autonomous decision-making across the enterprise.
From proof-of-value pilots to full-scale deployments, van der Aalst’s vision highlights how enterprises can build transparent, intelligent systems that learn and optimize in real time.
Core Takeaways
-
- The Evolution from Process Mining to Process Intelligence: What began as an academic effort to combine data and processes has matured into an enterprise-wide discipline. Modern process intelligence uses predictive analytics and machine learning to uncover inefficiencies and guide transformation.
-
- AI’s Relationship on Process Context: “Data without process context is blind,” said van der Aalst. For AI to produce meaningful results, it must understand the processes behind the data. Process mining provides that structure, linking data to business outcomes.
-
- Scaling Unlocks Real ROI: True value emerges when process mining is scaled across multiple workflows and departments. Like any enterprise software, the return on investment grows exponentially when applied broadly.
-
- Guardrails for Autonomous Agents: As AI agents take on operational decisions, process mining ensures accountability through real-time monitoring and conformance checking — providing the “control layer” for responsible automation.
Key Quotes
Combining Data and Processes Was Always the Goal
“In the late nineties, when we started working on process mining, there were no companies doing this and very little research, even though it was logical to combine data and processes. Process mining brought these two worlds together. Today, we use the term ‘process intelligence,’ but my original definition already included predictive and analytical capabilities. It’s not a new concept — it’s an evolution of what we envisioned more than 20 years ago.”
“The reason we now emphasize ‘process intelligence’ is to highlight the link to AI and machine learning. These technologies are changing the game completely, but their foundation is still process mining — understanding how your organization truly operates.”
AI Without Process Insight Is Just Guesswork
“Organizations have many systems and many tables of data. Process mining sits as a layer on top, giving you an end-to-end view. Without that, AI is blind. If you try to solve a delivery problem with AI, it might generate text unrelated to your actual issue because it doesn’t know whether the root cause lies in logistics, procurement, or production. Only by exposing the full process can AI do something meaningful.”
“It’s naive to think AI can generate accurate recommendations without understanding the processes behind the data. Process mining provides the structure that lets AI make sense of enterprise reality.”
Scaling Process Intelligence Multiplies Value
“The first step is often a short proof of value — four weeks of focused work that demonstrates potential. But the real value comes from scaling. You don’t buy Excel for one spreadsheet; you use it for many. The same goes for process mining. Companies like Mercedes and BMW are successful because they apply it across the entire organization. That’s when return on investment truly accelerates.”
“Once you build the expertise and infrastructure, each new process added increases the value exponentially. Scaling is the difference between a pilot project and enterprise transformation.”
Process Mining Is the Guardrail for Autonomous Agents
“As AI agents start making more decisions, process mining becomes the governance layer. It shows where problems occur and lets you decide when to deploy agents or make organizational changes. You can then monitor those agents continuously through conformance checking — defining how a process should run and spotting deviations automatically.”
“I don’t believe in agents that just do things without oversight. Process mining ensures accountability. It allows organizations to innovate with AI while keeping visibility, control, and trust intact.”
