At 19, Leopold von Waldthausen acquired a useful accessory: Sam Altman’s Y Combinator Demo Day badge. He had finished school and was working on a startup called Weview. In California, after meeting investor Marc Samwer at a nearby café, he encountered Altman in a parking lot and asked for the badge. Altman handed it over. Waldthausen says he understood whose name he was wearing surprisingly late. Fortunately, he still had time to take a photograph.
It is an excellent opening anecdote for a founder, partly because the equipment required was so modest. No ingenious software. A question, a stranger, a piece of plastic. Nearly a decade later, when Y Combinator asked applicants to describe a system they had hacked, he had his answer ready. Zeit AI entered the accelerator’s summer 2024 batch. This time, Waldthausen could attend under his own name.
Before the enterprise tables, the shopping reviews
Weview began with a problem familiar to anyone who has tried to buy something online: the purchase is straightforward; reaching a decision can consume the evening. The company brought video reviews into shopping, connecting people who made product videos with brands and retailers. Its founding team included Waldthausen, Jakob von Egidy and Christoph Pröschel. Pröschel dates their start to 2015.
The model put the explanation close to the purchase. Instead of sending a shopper away to search for a demonstration, a retailer could put video on the product page. Creators could earn from recommendations. The business won contracts with brands including MediaMarkt, entered the 2018 Techstars SAP.io cohort in Berlin and was acquired by DemoUp in 2019.
For Waldthausen, this meant encountering the less photogenic parts of entrepreneurship early: turning an idea into something another business would adopt. A product can sound perfectly reasonable in a pitch and still need a place in somebody else’s working day. That question would follow him into a very different kind of software.
Oxford, Yale, and an unfinished degree
His next chapter included computer science at Oxford, where his public education record lists 2019 to 2020 and a distinction. He also published research in process mining and cryptography. He went to Yale for an MBA, then left it. The compact account in his founder biography contains a sale, technical research and an abandoned degree. A conventional career ladder would have required rather a lot of alterations.
At Palantir, he spent three and a half years and eventually led the established customer business across Germany, Austria and Switzerland. That brought him close to companies with complex operations and substantial collections of data. It also brought him into a sustained working relationship with Marvin Bornstein, who would become Zeit AI’s technical co-founder.
The significance of that partnership is practical. When they started their company, they had already worked through difficult assignments together. Their friendship had acquired the useful additional feature of being tested at work.
A co-founder you have already seen at work
Waldthausen’s account of leaving Palantir puts trust near the centre of the decision. He and Bornstein had spent long stretches solving problems together. They knew how the other worked, how dependable he was, and what collaboration looked like when the assignment became complicated. Starting together carried that history into the new company.
Their target was European industry’s middle tier: businesses with serious operational demands, but without the data resources of the largest enterprises. Agriculture, manufacturing and logistics make less glamorous presentation slides than consumer gadgets. They also produce questions with consequences. A procurement decision, an inventory position or a financial report can demand information scattered across several systems.
Waldthausen had watched the work involved in getting those answers. Zeit AI was founded in 2024 around making that work easier to perform. The company’s technical ambitions were substantial; the point of entry was an ordinary business question.
The railway ticket is part of the product
Consider his trip to Cologne with colleague Johannes Nawrath. They took the train to show a customer a decision-making application built on that customer’s own data, for a specific business problem. Waldthausen described an engagement in which the team meets the people who will use the software and lets them try it before a contract is signed.
This gives the job title “forward-deployed engineer” something firmer than a fashionable ring. The engineer works close to the customer, where assumptions about a process can meet the process itself. The journey is part of discovering what deserves to be built. A railway ticket can be a surprisingly effective research instrument.
His explanation also sets a demanding condition for selling: users should see a working application and handle it themselves. That makes the demonstration an encounter with their own business, rather than a tour of an immaculate fictional company whose data has never misbehaved.
- 01 Meet the people
- 02 Build with their data
- 03 Put it in their hands
What the question has to travel through
ZeitMind, the company’s product, is designed to connect business systems, prepare their data and build applications through natural-language instructions. The intended users include finance, procurement, controlling and supply-chain teams. The promise concerns the whole route from a question to something usable, including the unglamorous preparation in between.
For a person running a business function, that preparation is often the obstacle. Records can exist in several places without being ready to answer a single question together. A conversational interface has to reach beyond the conversation if it is going to help. Otherwise, it merely makes the request for work sound more pleasant.
In a September 2026 interview, Waldthausen argued for beginning AI projects with the operating problem. He also described continuing to work with IT, including its control over permissions. The distinction matters: starting with the business user still leaves technical responsibilities to handle. His ambition is to shorten the distance between identifying a problem and putting a useful application in that person’s hands.
The first website could wait
Waldthausen has offered a small, revealing detail about the company’s early presentation. Its first website was assembled in a night during Y Combinator. It stayed in service for a year before an upgrade that explained the team, the customer engagement model and delivered work more clearly.
There is something reassuringly prosaic about the admission. A startup can build software intended for important commercial decisions while leaving its own shop window looking temporary. Customers and a polished explanation do not always arrive in the order a branding consultant might prefer.
The replacement mattered because it made the company’s method legible: who would turn up, how the work would proceed, and what it had already produced. For a founder asking established businesses to try unfamiliar software, those are useful details. The website had to catch up with the work taking place behind it.
Some people belong on the roadmap
One of Waldthausen’s recruiting stories begins two years before the hire. He told early investor Taylor Oliver that Georg Berecz was on the company’s roadmap. Waldthausen and Bornstein had each listed people they most wanted to work with again. Berecz was among them.
They kept in touch through occasional calls, product updates and a visit to Munich. Berecz eventually joined as a founding engineer after seven and a half years at Palantir. Other former colleagues, including Angelika Wieck and Jonas Otten, had also joined Zeit AI. The company was giving old working relationships a new setting.
“Code got cheap. Judgment did not,” Waldthausen wrote when announcing Berecz’s arrival. The appeal of the hire lay in deciding what deserved engineering effort. His account makes recruitment sound less like filling a vacancy than waiting for the right collaborator to believe the opportunity had become concrete.
“Code got cheap.
Judgment did not.”Leopold von Waldthausen, on recruiting Georg Berecz
A phone call, a hike, a working relationship
Wieck tells a complementary story. When Waldthausen contacted her about a call, she recognised the likely subject: he and Bornstein had founded a company. Her later account connects the decision to join with an idea they shared about letting the person with a question build an analysis through language. Waldthausen described her as the first person to join.
The pattern also appears outside the office. After a hike with LinkedIn co-founder Konstantin Guericke, Waldthausen reflected on how much a company depends on its people: early customers, complementary founders, the first employees and investors who remain trustworthy when circumstances become difficult.
These are business relationships described through shared experience. A walk, a previous project, a call kept open over several months. For someone building automation, Waldthausen spends a noticeable amount of his public account on the distinctly manual business of getting people to work together.
From the borrowed name to his own obligations
By September 2026, Zeit AI had announced seed backing that included Y Combinator, Oxford Seed Fund, Sequoia Capital Scout Fund, ACE Ventures and Hasso Plattner Venture Capital. HPVC described its investment as part of a $5 million seed. The group of individual backers included Mario Götze. The company was recruiting across engineering and roles that bring its software into customers’ operations.
Waldthausen was also taking those customers into the conversation. The BIG BANG KI Festival programme paired him with representatives of logistics business Wahl and agricultural trader Allgaier, moderated by ALPHAZIRKEL’s Albert Geiger. It is a fitting setting for his present argument: the people running established businesses should have a role in explaining what the software changes.
The badge story survives because it is funny and because it ends neatly. The current undertaking offers fewer neat endings. Applications have to work, recruits have to join, customers have to find a reason to keep using the product. Years after asking to borrow somebody else’s name, Waldthausen has plenty to do under his own.