The European AI unicorn run by a baker’s son—he learnt the fundamentals of business watching his father make bread rolls

Bastian Nominacher, co-founder of European AI unicorn, Celonis, comes from a family of bakers from Munich, Germany. A computer-gaming enthusiast, he would help his father use new digital technologies to run the five-generation business more efficiently. 

“I mostly helped on point of sale,” Nominacher told me. “In baking, you have a lot of demand spikes. The highest is at Christmas, because everyone buys all the stuff, and you never want to be short.” 

“We collected the data, so we really knew. Because the margin is not very high. You cannot just say: ‘Hey, let’s make 10,000 rolls’. Because if you throw away 2,000, first of all it’s a pity because it’s good food. But it also ruins your margin. So, we took the data, and we were really good and accurate and we could understand the demand patterns. We had a very low scrap rate.” 

In the 1980s, Warren Buffett said that the key to running a successful business was not the moonshot or the launch of the remarkable new product. Instead, success came from doing the basics very well and then finding simple efficiencies along the way.  

For many Fortune 500 companies, the Sage of Omaha’s advice is well taken. More than 25% of Fortune 500 companies are Celonis customers—using agentic AI to find better ways to bake bread. 

Celonis is a “process mining” firm—mapping every stage of a company’s business, discovering overlapping priorities, human and technological sticking points and inefficiencies. In 2022 it was valued at $13bn after a Series D $1bn fund-raising round. The firm, founded in 2011, is often spoken of as the next European tech IPO giant. 

To explain what Celonis does, Nominacher talks about a brewing firm that couldn’t get the basics right. 

“They were struggling with delivery to the supermarkets and bars, that was very often late, which is bad because you lose revenue and erode customer trust” he says. 

Celonis analyzed the 5,000 delivery runs a day the brewery was executing, worked out how to make the runs more efficient and reduced the number of journeys by 17%. 

“We brought more revenue in and increased customer satisfaction. We brought down costs because they needed less fuel. It also brought in more than 10% emission reduction.” 

This is what Celonis calls the “top-line, bottom-line, green-line” effects of process mining: increased revenue, reduced costs, and lower environmental impact. In 2025, Celonis was named #3 on the Fortune Future 50 list. 

“Today in Europe, most of the demand is in the supply chain, because supply chains have been under massive pressure, first with COVID, then with tariffs, now with the closure of Strait of Hormuz,” he says. 

“Obviously the economy is not going well in many countries, so companies need to free up [savings] and, if you do that right, you can easily find double- or triple-digit millions. Macroeconomic pressure really helps create clarity.” 

Read more: Silicon Valley has no monopoly on AI brainpower. That’s why Demis Hassabis is very happy to stay in London

The software-as-a-service field is under pressure, with many arguing that the very agentic AI that is being sold will ultimately mean the end of the traditional SaaS commercial model. AI-leveraged coding from within businesses will destroy the value of the sector. 

“I’m not a software stock market analyst,” Nominacher says with a smile when I ask him about the $1trn selloff in SaaS-listed stocks in February. “I think some software companies will suffer if they’re not differentiated, if you can easily be replaced. But, for Celonis, it is a massive opportunity because we are this piece of infrastructure that you need to drive this, and with massive use of agentic AI, it is a growing market. [Agentic AI] allows us to be much more effective and efficient. Our biggest constraint is massive demand.” 

The upsides are obvious, he argues, across the private and public sector. Do you want a doctor freeing up his or her time to speak to patients or spending hours, as Nominacher describes it, “chasing 25 clipboards”? He agrees with those who argue that a jobs apocalypse is not imminent. 

“Should an engineer spend their time triangulating everything?” he says. “Or can you really spend the time understanding what the issue is and determining the right strategy to drive it? It doesn’t mean there will be less work. There’s actually more work, and more relevant work, because engineers can spend more time speaking with customers, doing workshops, doing the high-value work, instead of getting bogged down.” 

Any talk of a public listing is not for now, he says. 

“In the long term we see Celonis as a public company,” he says. “But it needs to be at the right point in time. Currently we’re fully focused on our customers and building the technology. We’re not pushing it, [but] there will be a point in time when we do it—when it’s right from an internal perspective and also from a market conditions perspective. But nothing in the short- to mid-term future.” Many argue that agentic AI will “change everything”. And it will. But sometimes you also have to focus on how to make your bread rolls a little more efficiently.