Celonis wants to become the ‘Wikipedia of process intelligence’ with new platform

Celonis co-CEO (and co-founder) Alex Rinke took to the stage at the vendor’s annual user event in Munich this week and said that Celonis is “building the Wikipedia of process intelligence”. The intention behind that comment being, specifically, that Celonis hopes to become a foundational layer for enterprises wanting to deploy artificial intelligence (particularly generative AI). Why? It’s well known that OpenAI used Wikipedia as a source for its training data when building ChatGPT. And now Celonis similarly wants to become the common data model for enterprises seeking training data for their AI models, which it argues is hard when systems speak different languages and data is trapped in silos. 

Rinke took to the stage and said: 

There's no common system. There's no connective tissue that connects these different things. And that’s why it’s so hard for AI, because it confuses the AI.  

We are now introducing a common layer that we are building. We are putting all this process knowledge in one place. We're bringing it together in the product, in one place. And we're powering it with common definitions and over time we're going to extend into a system as well. So this is the process knowledge. 

Think of this as building the Wikipedia of process intelligence. 

The product being launched is what Celonis is calling its Process Intelligence Graph, which brings together process data, process knowledge, and also aims to take advantage of the knowledge trapped in the Celonis ecosystem to add further value. 

The Process Intelligence Graph goes well beyond process mining, making use of Celonis’ recent advancements in object centric process mining, which as diginomica has previously explained, takes process mining from two dimensional to three dimensional. Essentially, it allows people to map beyond single processes and to not have to go back to the data every time they need another view of that process. The best way to think about it is that whilst process mining is very effective in following a single object through an organization (such as an ‘order), it fails to take into account other objects that impact its progress (such as production, shipments, procurement etc). 

The Graph, now, adds in machine learning and deep process knowledge gathered from thousands of customer deployments over the past decade, which then creates a “system-agnostic, enriched digital twin of the business”. Celonis hopes too, that in the future its ecosystem’s knowledge, from partners and system integrators, will be added to the Process Intelligence Graph, to further build out this platform intelligence. The aim is that this provides the foundational layer for automation, system improvement, and ultimately AI deployments. 

Rinke added: 

The Process Intelligence Graph is the modern equivalent of the Rosetta stone - it’s the connective tissue that’s been missing in modern enterprises. By bringing together the building blocks of data, knowledge, and the power and scale of our ecosystem, we are building the foundation our customers need to make processes and entire value chains work for everyone. 

This level of cross-process, cross-system intelligence can only be achieved with the Celonis platform and the Process Intelligence Graph at its heart. The PI Graph enables technologies like AI and automation, feeding them the data they need for how processes actually run, contextualized with the knowledge of why they run the way they do, and how they can be improved.

Diving deeper

diginomica got the chance to speak to Divya Krishnan, VP of Product Marketing at Celonis, to get a better understanding of the vendor’s thinking behind the Process Intelligence Graph and what this means for customers. We have spoken to Krishnan previously about her - and the company’s - ambitions to make Celonis more intuitive, open and intelligent. 

Krishnan said that Celonis brought process mining out of academia and applied it to the enterprise with its technology. The focus has been on generating high quality insights around how processes are running, why they run a certain way, and where potential improvement lies. Since then, it has advanced this thinking to include actionable insights - putting in place alerts, automations, triggers and pre-built dashboards and apps to help customers realize more value from the data. 

Considering the opportunity of the Process Intelligence Graph, Krishnan said: 

The next level for us now is this unified, agnostic, digital twin, that's enriched with process improvement knowledge that comes from our decade-plus of process mining and implementation. 

What's really needed is this holistic end-to-end understanding of: how did the processes actually run? How did they get executed? These underlying systems also have a different means of engagement. They don't play well together. They have their own vocabulary for how they record information about activities and events that are taking place. And they weren't built to facilitate easy information sharing, from one tool or one system to another. 

But for organizations to really be able to improve process performance, there is this need for a shared language to be able to go in and achieve the enterprise goal. 

Providing the example of the relationship between marketing and sales, Krishnan said that when looking at the end-to-end lead-to-cash funnel, the story told by marketing systems is going to be very different from the point of view of the sales systems. She added: 

What you really want to know end-to-end is: How is it running? What is the value that I should be focusing on? And how can I go and capture it? That's what we're really looking to do here with the Process Intelligence Graph. 

As noted above, the first building block of this shared language is process data, which is the object centric data model. The way Krishnan talks to customers about this is that organizations should think about their business as a city - and that an object centric data model allows you to capture exactly what’s happening: the movement of every taxi, train, bike, person. Or in the context of a business, order, sale, person, invoice, etc. 

Looking at the role of process knowledge as the next layer, Krishnan added: 

The next building block of the shared language is this process improvement knowledge. And that, if we go back to my business as a city example, is the knowledge of: this is a one way street, that’s why cars are moving only in one direction, this is where our traffic light is, that's why the traffic is slowing down and speeding up, this is an alleyway and it's a shortcut. 

That has been built up by us [Celonis] and by partners and customers that we've worked with over the last decade. It's now productized within the platform and really embedded to drive significantly faster time to value. 

And finally, Celonis will look to its partners to make the Process Intelligence Graph even more intelligent. She added :

The third piece of this is the extension of both the data and the knowledge, by our ecosystem. So we have partners who are defining additional objects and event definitions, who are adding to the process improvement knowledge with their own benchmarks, their own improvement opportunities that they’ve found, and all that really comes together as this process intelligence graph. 

So it's the unification of the data model. And this process intelligence layer that really provides that. 

My take

It has been an interesting first morning with Celonis at its event in Munich. There’s a few elements to this. Firstly, it makes sense for Celonis to want to enable customers to realize more value from its platform more quickly. The Process Intellgience Graph aims to do this in a number of ways, by both looking cross-organization, but also through actionable insights and automation. The key being that with the right data, organizations may be more willing to change and identify where things aren’t working. But looking down the line, there’s another interesting element to this. Celonis believes that it could provide the governance foundation for AI models in the enterprise. By having object defined process knowledge at their fingertips, organizations may be able to better control and understand what data they are putting into an LLM, for example. Hence, aiming to becoming the ‘Wikipedia of process intelligence’. We will have more on that later this week, following an interview with Celonis’ AI expert.