Bootstrapping a Decacorn on $15k with Celonis CEO Alexander Rinke

Thanks for having me, and good to see everybody. It's a packed house here. Is it always like that? Yeah, pretty much.

Okay, so I think one of my predecessors here on stage said, like, who doesn't know vector databases? So I'm going to do it the other way around. Who knows process mining? Who's heard of it? Well, that's more than I would have expected, but still less than 50% of the people in the room. So, process mining—look, obviously every business is running a lot of processes, right? You're selling to customers, you're invoicing customers, you're serving customers, you're responding to product requests, you're building products.

All those things can be expressed as a process. So our model for looking at businesses is they run on this really complex set of interconnected processes, right? And what process mining does, and process intelligence, is use all the data that's generated because, obviously, through digitization, a lot of those processes are based on actual IT systems, right? So you don't book your invoices on paper anymore. You don't send faxes around anymore.

You have IT systems to do all that, to keep a record of all your customers, to keep a record of all your suppliers, of all your orders, right? And Celonis will be like a huge vacuum cleaner, soak up all that data, and then show you how your company operates very visually, right? What are the main process flows? Where are you deviating? Where are you disappointing customers? Where are you leaking time and money? And it really helps you, in a data-driven way, to run your company and optimize it.

And we do this, obviously, we started with very big companies, right? Obviously, startups wouldn't be our target audience usually, but we work with enterprises, I would say north of 1,000 employees. We are big users of Celonis internally. And that includes half of the biggest 200 companies in the world. So we have a lot of customers from different industries. It doesn't really matter which industry. Processes are really an important problem everywhere. And it's important to make people successful because it really sucks to be part of a bad process.

It's important to make businesses successful, right? Some of our customers—we just had, at an event here in New York, Pfizer on stage. And Pfizer last year saved more than $200 million using Celonis, right? So there's huge economic value in what we do. But there's also sustainability value, right? So we have a big coffee retailer, and they look at how much emissions they produce through their processes, through their supply chain, right? You're thinking about having thousands of stores around the country, bringing the coffee there.

I mean, there's a lot of emission factors that you generate, and they're reducing that using Celonis. So there's a top-line impact, a bottom-line impact, a people impact, and also a green-line impact. So, process mining. Super great.

Right. When we started Celonis, we didn't even know what a startup or venture capitalist was. So it was very different than all of you guys who are here in the room, who have a huge head start in terms of experience and exposure. We found out about process mining; one of my co-founders found out about it in a university lecture. And back then, I think all the people that had heard of process mining would have fit into a corner of this room, right?

Like, not even the whole room. So in a sense, we got very lucky that we got early exposure to this, even though the technology had been around for almost 10 years, and the research had been around for almost 10 years. At the same time, what you do in Germany is a bit different than in the U.S. When you go to university or college, you often have a job on the side, right? So we actually all had a job where we worked a day a week, a couple of days a week, doing IT service in an IT service organization.

Right? So tickets, incidents, my printer doesn't work, I don't know how to reset my password, that kind of stuff. And there is where we realized that the processes in this organization had a lot of optimization potential. At the same time, we heard about process mining, and we didn't find any tools to do it. So we went online, we searched for process mining software, and we didn't find any software. So we decided, let's just build our own software. And then we decided, let's start a company that focuses on that.

We had no funding, right? So for the first six years, all the funding we had was $15,000. So when I hear some startups now raising, like, $50 million seed rounds, I just had a VC email me, "Hey, we have this AI company. They're raising a $200 million seed round," right? I'm like, well, that's impressive. So our seed round was $15,000, because when you start a company in Germany, you need to put in $15,000. You can't do it like in the U.S., where you can incorporate for, like, $500.

So we had $15,000. I borrowed my part from my co-founder because he had done a better job managing his finances, and we got started.

So it's obviously a very constrained environment, right? The good news is it didn't take us long to build the first product. And I think that's a lesson, right, that has also been taught by many people. But I think we really felt it. Like, the first product we built in three weeks. And I still remember, I was there, I was contributing a little bit. And then Martin, who is our key engineer, he's our CTO, was writing most of the code.

And then Basti was bringing us beer and pizza, and we were literally in the office for three weeks. And we had our first prototype. It was not very sophisticated, but kind of did the job. You could do demos. And then we went out and we tried to meet with customers. We used every one of our friends who did an internship in some company to get an intro and get a meeting. And then in the first year, we found, I think, five or six customers that were sort of willing to pay us $20,000 for a pilot or $30,000 for a POC, not huge projects, but real money.

So we had some money to hire our first employee and buy our first booth at a conference. I mean, the first conferences we went to, we couldn't even afford the booth. We just went there and talked to people, right? A bit awkward, like, hi, I'm the guy with a business card and we don't have a booth. But it was very scrappy at the beginning. But we sort of worked our way there through finding customers. And then those customers, as they were happy, they would sign the first license deal.

And that way, we sort of funded the company for the first, I think, almost six years.

So it was not the CEO; it was the co-founder, Hasso Plattner. I'm in Frankfurt. Martin kicked me out of the engineering team after the three-week prototype. He's like, "You're not good enough, right? You go figure out something else to do." So then I was responsible for sales. I was out in the country every day. I think in one year, I drove over 100,000 kilometers just to see customers in Germany.

So I was out a lot seeing customers, and I was just at a customer meeting when someone called me that Hasso Plattner was giving this speech at an event in Berlin, where I grew up, very close to where I actually grew up. So I just went there and stalked him, basically, after he was done giving a speech. And then ultimately he made some intros, and that led to the first partnership we had.

Well, I think first of all, partnerships often don't work. Like 90% of them don't work, right? And we also tried many, and many didn't work. I think the first thing that a company has to figure out right out of the gate is product-market-people fit, right? People talk about product-market fit. It's not just about product-market fit; it's also your team, right? You need people that fit that opportunity. And before you have that product-market fit, I don't think you really need to think a lot about hiring a sales team, building a partnership, going into new markets, or doing anything fancy like that.

And I think one of the challenges that I've noticed is that a lot of companies fake this product-market fit, right? They basically put it on a slide: "Oh, we have this great product-market fit," because they want to raise the next funding round. And some of the VCs might not be as sophisticated as you guys are, and they kind of believe it. They throw a lot of money at the company, but the real substance isn't there, right?

And then it's going to bite you back later, right? So how do you really find product-market fit? You need a product that you can offer at a reasonable cost, right, that actually solves a problem for customers that they're actually willing to pay for and that they're actually willing to renew, right? So they are actually going to get value continuously from this product. Sounds very easy. It's very hard to do. Sometimes it takes years, and you can't skip it, right?

Because if you try to skip it, it's going to bite you back later. So I think once a company has solid product-market fit, which can take years, right, depending on the market you target, the complexity of your product, I think one of the benefits of bootstrapping, it gave us a lot of time because we didn't have to show any VC the progress we made this quarter or all the exciting—we didn't have to excite anybody, right? Except for customers. So it gave us a lot of time in almost like a protected environment.

Being in Germany, not knowing—I mean, we were basically the only enterprise software startup that I knew of, right? And I think there was a benefit. There's lots of downsides, but there was a benefit. And I would not recommend you thinking about partnerships or scaling go-to-market or hiring salespeople or even raising a lot of money, to be honest, before you feel really confident that you have a product that solves a problem that you can scale, that you can also renew customers, you can find enough customers, not too nichey.

And actually, the other thing is, like, your product doesn't just need to be a little bit better than the alternative, right? It needs to actually—because there's a lot of inertia. So it needs to be, like, so good that it energizes people beyond their level of inertia, right? And if you can do that, you've got something. You can think about partnerships, different strategies to scale, go-to-market, et cetera. If you try to fake that or skip that, I think you're not going to succeed.

So we did Germany. It was a big market, right? We had probably, I think, close to 100 customers or so. So if you have 100 customers, especially in big enterprise, right, they're serious companies. You've got product-market fit. So we thought about how to scale go-to-market. And then we said, hey, what is really the benefit if we are in Germany and we are also in France?

I think you're from France, right?

Yeah. So no offense, but it's like, we are in Germany and we are also in France. Or we are in Germany and we are also in the Netherlands or Switzerland. So we thought, well, that's not very interesting. If we really want to tackle an international global market, we need to do the US. So we went from Germany, and our second market was the US. And actually, our first Celonaut here—so we call our team Celonauts—here in the US is actually here, Mark.

So I came over with Mark. Where's he? Oh, there he is. So if you want to talk to him as well. So both of us came over. I spent sort of a portion of my time here. He moved full-time, and we started attacking the US market. Because we said, well, if you can do two markets, then you can do multiple, but the US is really the big nut to crack. And since then, I've moved here full-time. And the US, if you're a European company, is a really tough market to crack, and you need to put a lot of focus into that.

You can't do 20 other things at the same time.

So the legal headquarters of the company is Germany, but the second headquarters—we call it our second headquarters—its biggest individual market is the US now.

Right. A lot of process in there, right? I mean, as a German company, I guess we have to stay on brand. And so the Process Intelligence Graph is something we introduced at our last user conference and something we are very, very excited about. It's really the foundation for our future. And the Process Intelligence Graph—before, we always looked at individual processes. It was called case-centric. So you pick a case: I want to track the order, and then you can track the order.

I want to track the Jira issue, and then you can track the Jira issue. I want to track the customer service request, track the customer service request. Obviously, all those things are interrelated, right? Those processes are all connected. One of our customers is GE Healthcare. If GE Healthcare wants to ship an MRI machine, they have to build the MRI machine. They have to ship the MRI machine. They have to make sure there's electricity at the hospital.

These are huge machines. They have to make sure the building is ready. I mean, there's lots of processes that are interconnected. So the Process Intelligence Graph basically marries process mining and a lot of the theory behind graph databases and graph technology, and establishes a foundation that allows you to represent all of a company's processes in one connected data model and data structure that is also completely independent from the source system. So you could have Oracle, SAP, Salesforce—the Process Intelligence Graph, or we call it the PIG.

Now I have a lot of pigs in my office—not real pigs, but little pig statues. So it's our PIG. And the PIG can bring in data from any system, any number of processes, and establish a knowledge graph, effectively, for this process data, and establishes a common language for representing all your process data in your company. And it drives a huge amount of insights because you cannot just understand what's going on in individual processes.

You can also understand how interconnected processes work and how they operate. And we believe it's big for generative AI in enterprises. And it's really a new foundation that we've built.

100%, yes, 100%. We basically switch on the lights for you. It's always a huge aha moment. And we show you where the friction is in your company without you telling us anything. If you tell us what your systems are—you use Salesforce, SAP—boom, the Process Intelligence Graph will show you: here's where your customers get stuck. Here's where you can generate more cash by billing people faster. Here's where your customer service is screwing up. We show you all the opportunities to optimize.

Yeah, so it's based on—basically, first of all, this process data, you can think of it as objects and events. So objects are things like invoices, customer service complaints, orders, customers, people, et cetera. And then events are things like, okay, a new customer puts in an order, or we ship an MRI, right? So we are very good at finding these objects and events in databases and application logs and all sorts of representations. And then we bring it together and we see, okay, where do things get delayed?

Where do we have a lot of loops? So it's some generic understanding that we can get. And then we also have a wealth of knowledge. So we have this knowledge layer of, hey, we know that it's not a good idea to pay the same invoice twice, right? It's not a good idea, right? Paying the same invoice twice. We know it's not a good idea to leave your customer hanging when they have a new call for five days, right?

We know it's not good to promise your customer—like, you've always bought something on Amazon, and you get really annoyed when they promise you, "We're going to show up on Wednesday," and then they eventually show up on Friday. We know that not being on time for delivery is a really bad idea. So we have this knowledge built in. You can load a complete process that we've never seen before; we'll still give you some really strong insights, but you might need to use our toolchain to explore the process.

If it's a process that we've seen before, we're going to automate some of the insight generation for you.

So there's usually a center of excellence, some people that get really deep, right? And they analyze. They serve sort of as a hub for the adoption of Celonis within the company. But ultimately, this observability can be used by anybody in the company, right? So there's business users that access dashboards and monitor their processes on an ongoing basis. And we can even bring the insight to you. So if you use Salesforce, we can show insights in Salesforce, right?

So you don't even need to go to the Celonis UI. We can show a trigger, or, hey, you should focus on this customer, or you have a problem here, proactively to people as well.

Yeah, and generative AI is this incredibly exciting technology where someone said this recently, it's not something that's just being invented. It's also something that's being discovered, right? We don't even know what the potential is going to be. So there's two things we know that we are very excited about, and there's a lot of things we don't know that we probably will be excited about in a year from now. So the things we know is, one, people will interface with software in natural language, right?

So we have, for example, invested a lot in building a natural language interface for Celonis. So instead of going to a dashboard and figuring out your process flow and finding, oh, this is delayed here, you can just ask your Celonis Copilot, "What are the top five issues in my process? Why are we not paying our suppliers on time? Why are our customers not happy?" And we will give you answers back in natural language, right? And we can even generate dashboards for you automatically. Again, it just lowers the bar so you can talk to your software, right?

And that is pretty clear, right? It doesn't work perfectly yet. There's a lot of issues you have to solve technically, but it works pretty well, actually. And we are constantly surprised, week over week, by the progress we are making. The second topic is that we believe generative AI will also automate processes in a company, right? It will automate things that are manual today. And we think that Celonis is very important because we have all the context of the process, right?

So we know where customers usually get stuck. We even know in real time, right? If you have five different systems, we know across those systems where this order is. And AI is incredibly input-sensitive, right? So the better the input, the better the AI. So we believe also that Celonis can be an engagement or translation layer to basically, if you have a really smart LLM that knows nothing about your company, instead of trying to find a lot of data sources and make sense of the data, it can just interact with Celonis, which has this common language and knowledge graph for process data, and it will know a lot more about the company.

It will be able to automate better and write better recommendations, et cetera. So we think those are two things that we see pretty clearly, and we're very excited about. But there's a lot more that we probably don't know.

Yeah, so Symbio basically captures a lot of knowledge about a company that's not in the systems. So, for example, Symbio would capture information like which risks you have in a company, right? Because you need to show that to the regulators. So you put it in Symbio, but it's not really in your systems. Or it would capture information about your organizational chart, things like that. And it basically extends our process intelligence graph that is focused on process data with a lot of conceptual knowledge about the company.

Yeah, I think that's really tough. And I think that a lot of enterprise companies sort of overpivoted on AI, but then couldn't really show traction, right? Dot AI, right? And these are the only two letters that matter in the alphabet, right? A and I. And there were people saying, like, you need to tell people that you've always been a generative AI company, right? And you were, like, way ahead of this and all that stuff. Hold on a second.

Like, let's find out what actually sticks and what's tangible, what's possible today. Because also, especially if you have some scale, you can't lead your customers down science projects that don't lead to an outcome. So we tried to really figure out what works. We had a team already around AI called Celonis AI, and we obviously expanded that team, and we really figured out how to provide tangible things to customers. And at the beginning it was like, oh, you're not doing enough. You're not talking enough about this.

And I think now it's, hey, you actually have some tangible things to show, where I think a lot of other companies tried to sort of AI-wash their offerings, but didn't have a lot of actual tangible things to demonstrate.

Right. So who's in enterprise software broadly? Okay, it's a lot of people. So, I mean, first of all, one thing I've learned is that the go-to-market is highly dependent on your product and also the target segment, right? So a lot of people, I think, think about go-to-market in generic ways, but it's something very specific for the company. So first of all, it obviously matters a lot whether you sell to big enterprises or whether you sell to sort of SMB, right, which is more of a product-led motion.

Big enterprises can be product-led motions. Usually, product-led motions start with smaller companies, but not for what we do, because you have to connect to all these data sources, you have to get executive sponsorship, et cetera. So then, if you have an enterprise go-to-market like we do, which I also know most about, it's going to be very reference-based, very use-case-based, right? So even though you have a platform that can theoretically address anything, you need to build some core use cases that you know who you target, right?

I target the head of shared services, for example. For us, a lot of big companies have shared service centers, and that's usually a great place for us, right? Because they have a lot of process work. They have to optimize, they have to save money every year. We found that it's a good sweet spot for us. There are others as well. And you build these hero use cases or killer apps on top of your platform, and you really drive those killer apps.

And then you need to really think through the lifecycle of your customer and make sure they're really successful. They're getting value, they're tracking that value, they're committing to that value, they're talking to others about it. And then you can build a flywheel around that. But just hiring a lot of salespeople, right, is not a go-to-market strategy. And that's what a lot of startups did, right? Hey, we just got $100 million in funding. We're going to hire tons of salespeople.

They know how to do this. Good luck. That's not going to work. The other thing I think that you have to think about in go-to-market is the product-market-people fit. Go-to-market, especially in the large enterprise, is a people business. So you need to think about which people, which profiles really fit your product. So, for example, in our case, you'll have a lot of people in go-to-market that are very technical, that understand use cases very well, because what we sell—I mean, we already achieved reasonable scale, and 60% of the people in this room didn't even know what process mining is, right?

If I asked you what CRM is, everybody would know what CRM is. So obviously our go-to-market folks have to explain what we do a lot, right? So you don't have to explain what a CRM system is, right? So in our case, our go-to-market people need to be more technical, more use-case-oriented, a bit of a different mix there in terms of commercial profiles versus technical profiles. They have to be excited about what we do because they have to likely tell the customer for the first time what process mining, process intelligence is all about.

So I think every—the other advice I could give to founders is, if you didn't sell the first million or two in revenue yourself, don't expect anybody else to do it, right? And you don't need to be a natural salesperson, right? Like, I studied math. I've never been a salesperson before, right? But we have a lot of passion for what we do, as hopefully you can tell, and obviously understand it quite well. So it's very hard to scale a go-to-market if you haven't gotten this product-market-people fit figured out.

And it's usually something where you need to do it yourself, no matter how much funding you have, no matter which big-shot executives you can recruit. Like, in the early stages, it's going to be you selling to the customers. If you're an enterprise motion, if you're a product-led motion, it might be very different. I've never done that, so I couldn't give any good advice around it.

Martin is a tough guy, though, in my defense.

Well, it's still very much a work in progress. The most important thing is that you have to have the ability to listen and really surround yourself with people that are going to be honest with you. And I think if you do that, you're going to learn a lot, right? But I think you really have to have this realization that, yes, you have a lot of advantages over other executives because you know the product very well. You have tons of passion, right?

You have seen a lot of customers from the early days, et cetera. But at the same time, you haven't run big teams before, and you're going to make mistakes. So the best way to do it is to really surround yourself with people that are going to be honest with you and listen to them, right? The other thing that I'm a huge believer in is mentorship. So you wouldn't believe what kind of people you just email: "Hey, can you get on a call with me? I'm trying to do this startup, and I want to learn something," that actually are going to mentor you and help you.

So find other founders that are maybe a little bit ahead of you. I always had mentors, to this day, that have been ahead of me that I can always call and say, "Hey, I have this problem. How do I figure this out?" It's a safe space. I can ask any question. So I think mentorship is also really important.

Well, I think we can really become a horizontal platform for process intelligence to our customers, to drive a ton of value and be one of the most strategic systems that they employ to run their businesses, which we already see with our most advanced customers. We want to kind of get every customer there, something that provides huge amounts of value to—like, we have this principle that we are extremely obsessed with customer value. So, for example, I can pull out a dashboard right now for every single customer that we have: how much value have they realized, right?

Data-driven. Most companies can't. I don't know any other company that can do that. So, providing value to customers is huge for us, but also to connect processes between companies and build an ecosystem of people that can build more value on top of our process intelligence graph is also a very big part of our ambition, which is basically platformizing our offering. And then just build a company that's a great place to work for our people, that customers are really excited about, that partners are excited about.

Obviously, going public, all those things are important, but it's really about what substance do we want to create? What mission are we on? And how do we really focus on that? And then everything else will work out wonderfully.