Phil Fersht — CEO, HFS Research[00:11]
So good afternoon, Alex. Great to catch up with you again. You know, there’s so much going on this year. I feel like it’s a dichotomy in our industry between excitement and anxiety. But it’d be great to hear a bit more from you. We’ve heard from you a lot in the past with your growth of Celonis and your infiltration into workflow platforms and process intelligence. But the shiny new object that seemed to hit the whole industry flat-footed last November was GenAI, and there’s been a huge amount of activity and noise around this, obviously, since then. I’d love to hear a bit more from your perspective: did this surprise you? And maybe a little about, is this something we should be enthusiastic and excited about, or is this something we should be a little scared of?
Alexander Rinke — Co-Founder and Co-CEO, Celonis[01:03]
Well, first of all, I think it surprised all of us in terms of the adoption cycle with ChatGPT. I mean, we played around with those things before and we incorporated a lot of AI into our product for years. But I think what ChatGPT really did, one, it took the GPT model. GPT-3 was much better than GPT-2. I still remember playing around with GPT-2, being quite impressed, but GPT-3 was obviously much better, and then 3.5 and then 4. And obviously, they created a UI for that everybody could use. So then everybody started — it really started, I think, to jumpstart the adoption cycle of generative AI for both consumers but also businesses. So that certainly came as a surprise to many of us. Not as a sort of surprise — OK, we never thought this would happen — but I think how it happened came as a surprise. And I think generally we should be excited about it.
Phil Fersht — CEO, HFS Research[01:52]
But where do you think we’re going to see some initial immediate value and maybe longer-term value with this?
Alexander Rinke — Co-Founder and Co-CEO, Celonis[01:52]
Yeah, I think the first layer that is really gonna change pretty dramatically is the user interface, right? So what humans engage with, how they engage with computers. We for a long time have had different paradigms for automating user interfaces, making them easier to use, making workflows and automation better. And a lot of that will just be generated on the fly, according to process context and intelligence. And that’s really exciting because you can generate entire workflows on the fly. You can generate apps on the fly. You can automate much more. Think about user interfaces that allow us access to insights. We’ll just be able to have conversations about insights, which will really make them accessible to many, many more people. And we already have this working at Celonis, and taking the next step. In the first versions we see with users, it’s really quite exciting, and a real breakthrough in terms of transforming how we can provide insights to people and how many people we can provide it to. So I think really the whole space of automation, and how software companies build user interfaces, which are very related, is going to be transformed. And it’s gonna unlock productivity. It’s going to make people better. It’s going to accelerate human accomplishment and have a real impact on businesses.
Phil Fersht — CEO, HFS Research[03:38]
So do you see we’ve finally reached an inflection point where enterprises have no choice but to make a lot of difficult, maybe complex changes to their infrastructures, their data, their processes to actually exploit this?
Alexander Rinke — Co-Founder and Co-CEO, Celonis[03:58]
If you take current infrastructures without Celonis and then the AI models, or the LLMs more specifically — because what we can really provide is a unified and horizontal process data language and taxonomy. They can be exposed to these AI models. And that is really exciting because you can just plug in the AI model and they can, in one language, in one system, get context about many, many processes in the company and even how those processes relate to each other. And that is a huge unlock, right? Because you think about AI — AI is really good at generating code. It’s much better at generating code than it is at many other problems. Why? Because code already has a structured grammar and language for writing computer programs. And when you give it that constrained space of how to express things, then it gets much better. And you give it a lot of examples, it gets really good at helping you to solve more complex problems. And the same is true for Celonis plus AI and business processes, because we can provide that common process intelligence layer. So we are working with a large bank and one of our big SI partners at the moment. And the bank really wants to build an LLM-powered customer service interface. So a customer chatbot, basically, that’s very intelligent. And it kind of really explains well what I tried to describe earlier with the process intelligence layer, right? Because now the chatbot needs to know a lot of things about the process. Just GPT is not good enough, because it can answer everything that’s on the internet, but it can’t answer, where’s your order? Because it doesn’t know where the order is. It can’t answer when the order is going to arrive. It can answer how do I open my bank account here, but it can’t answer those questions because it lacks the context of that specific business. So now they could try to sort of harmonize data from many different systems, feed them into the chatbot, and try to explain them all this process context, or they already have Celonis deployed across many of those processes along the customer journey. We already know where the orders are and how to open accounts, and also how well we served that specific customer in the past. And we just expose this process context to the front office. And that way, the chatbot is implemented much faster, and it’s much better because it has all this process context. It doesn’t just know how the process works, it also knows how long it takes, how long it took for that specific customer, where the current requests are stuck, who’s usually working with them. So it has a lot more context about the business processes in that company. And that becomes a critical enabler of making these things really work beyond science projects within large businesses.
Phil Fersht — CEO, HFS Research[07:05]
How’s GenAI changing the types of companies that you’re needing to partner with, when we see firms coming into play like, obviously, Llama, Hugging Face, some of these businesses?
Alexander Rinke — Co-Founder and Co-CEO, Celonis[07:05]
Yeah, 100%. I mean, and ultimately, we’re going to enable whatever the customer wants to use. There are different reasons to choose any one of those models that you just mentioned. I think the interesting thing for us is then there’s a whole host of other companies that build on top of that infrastructure, that either train models for specific purposes or fine-tune them, or build AI and AI-native apps for specific use cases.
Phil Fersht — CEO, HFS Research[07:31]
So, I’m gonna put you on the spot a little here, but you know, how big is this? Is this like as big as the internet? Is this more like the steam engine? Is this something, maybe once the noise dies down, a good tool for getting a bit smarter and slicker at things? What’s your overall assessment for the general impact of GenAI?
Alexander Rinke — Co-Founder and Co-CEO, Celonis[08:04]
So I’m gonna be controversial, and I don’t think anybody could give like a 100-year view here because we don’t know how much this is gonna evolve. And obviously, if we really evolve into AGI, that might change all of those predictions. But I think in the current form, obviously foreseeing some advancements, I think it’s somewhere around like the internet. Maybe a little bit different, right? But it’s bigger than mobile, for sure. You could argue it’s as big as the internet or as big as mobile. It has some similarities to mobile because it changes the user interface, but it does that in a more fundamental way. So I think it’s going to be more comparable to the internet. I don’t think it’s the steam engine.
Phil Fersht — CEO, HFS Research[08:39]
How does this, in a nutshell, impact do you think your world of process intelligence? Is this gonna really move you to a different place, or do you think this is a more gradual step change as you start to absorb its potential?
Alexander Rinke — Co-Founder and Co-CEO, Celonis[09:18]
Well, I think it creates this new layer. First of all, we are embedding AI, many GenAI — and we have embedded AI models for many years in many steps of how we build our product and in our core products. So it just accelerates the speed of innovation and time to value within our core system. That’s really important. The second thing is it really puts process intelligence at the heart, as the engagement layer to make AI work in your enterprise. That is really important. When you look at the current GenAI implementations, there’s some really exciting momentum, but there’s also a lot of science projects, and they don’t really drive business value yet. In order to drive business value, you need to quickly be able to connect these AI models with process context and process intelligence. So it really accelerates — and we believe very, very quickly and very deeply — the importance of a process intelligence layer in your enterprise. In order to make AI work, you need AI that knows how your business flows, and that’s what Celonis enables.
Phil Fersht — CEO, HFS Research[10:31]
Wonderful. Well, I really appreciate your thoughts on this, Alex. It’s always good catching up with you. Thank you. Bye bye. Take care.
