Jan-Philipp Mohr and the value of knowing where things are

A bottle of wine and a warehouse have something in common: both become easier to understand when you know what is in them. Jan-Philipp Mohr has worked on each side of that proposition. Before co-founding the business now called DARVIS, he helped build a venture that connected people’s taste preferences with the wines a retailer could actually offer. Later, his attention moved to rooms, objects and the information that physical operations leave behind.

The route is pleasingly indirect. There are software deals in Berlin, a California company with a name that sounds at home in a gaming studio, and an international team whose geography partly grew out of conversations in shared accommodation. Today, Mohr is DARVIS’s Chief Product Officer. His working territory is the gap between seeing something and understanding its place in a process. That gap can be expensive. It can also be a very good place to start asking questions.

A taste for context

Mohr’s early career included work for the Berlin investment house WCI. He handled software mergers and acquisitions and interim management responsibilities within its portfolio. There is a useful distinction between buying a software business and operating one: a spreadsheet can accommodate a plan rather more politely than customers do. His work covered both the transaction and the practical business that continued after it.

He also co-founded a wine recommendation venture, listed as That’s My Wine. Its premise brought together a customer’s preferences and the retailer’s available inventory. Machine learning helped with the matching. The recommendation was tied to the things that could actually be sold, giving the exercise a physical constraint as well as a computational one.

A wine recommendation can be wonderfully confident until someone discovers the bottle is unavailable. This is the modest comedy of all recommendation systems: the world retains a vote. Mohr’s later work would make the relationship between data and physical reality much more explicit. The earlier venture is an interesting prelude because the shelf already mattered.

He holds a business degree from the University of Kent. His career brings commercial work into the same frame as technology. That combination gives this story its particular texture. The question is never simply whether a machine can perform an interesting trick. There is also the matter of which decision the trick improves, and who has a reason to pay for it.

The world outside the headset

In 2015, Mohr co-founded Hashplay in California. Ingo Nadler and Jan Schlüter were his fellow founders. The company’s origins lay in gaming and immersive technology. Its subsequent work reached into the less theatrical business of understanding operational data. A virtual world is an inviting place to experiment with space; the physical world supplies a considerably less obedient cast of objects.

By 2018, Hashplay was describing a product called Immersive Data. It brought operational information into a spatial setting, including digital models of real environments. The aim was to help managers grasp data in context. Mohr’s formulation included the need to “visualize everything around it”. A data point had neighbours, a location and consequences.

That year, Hashplay announced EAST, a live-video analysis extension. It used camera footage to derive information such as flows of people and vehicles, and presented it within a spatial view. The audience included managers without an engineering or data-science background. There is a practical ambition in that detail: understanding the operation should not require becoming the person who built the software.

Mohr dates the company’s pivot to 2019. Hashplay became DARVIS, concentrating its computer-vision work on the visibility of physical operations. A new name marked the turn. The interest in connecting data with space continued through it.

The difference a map makes

In his 2022 account of the business, Mohr explained that DARVIS extracted data from imagery and tracked assets in warehouses. He identified the US Postal Service as a customer with deployments in multiple locations. The product could consider where assets were, how long they remained idle and their fill levels. A photograph became the starting point for operational information.

His short description was direct: “We use computer vision to translate the real world into data.” Translation is an apt word. A useful translation preserves relationships. In a room, those relationships include where things stand and how they move through the work.

Mohr emphasised mapping as a distinguishing part of the product. Recognising an item gives it a label. Locating it can help describe its status in a process. Consider a generic warehouse trolley: the same trolley beside a loading area or parked elsewhere may mean different things to the people running the building. The example is ordinary; the value lies in making that distinction available to software.

His explanation also used the familiar relationship between payment activity and geography. A location can alter how a transaction is interpreted. He applied the same interest in context to physical assets. The place is part of the information.

A company built through conversations

DARVIS’s geography contains a founder anecdote with very little varnish. Mohr said he stayed in Silicon Valley hacker houses because hotels were expensive. In one of them, he met the person who would run the company’s Pakistan operations. Islamabad entered the business’s map through a conversation in a place he had chosen partly to keep the bill down.

He described another deliberate search around Nashville, including research through SelectUSA and encouragement from an early employee from Memphis. The choice involved considering costs, talent and where it made sense to build. “First of all, Nashville is cool,” he said. Some location strategies begin with rather heavier prose.

The anecdote is worth keeping at its actual size. A shared house introduced two people. A discussion about another city helped a company consider a move. These encounters make the international structure more tangible than a collection of dots on a presentation slide. Before there is a country operation, there is someone prepared to explain why it might work.

“First of all, Nashville is cool.”

Jan-Philipp Mohr, 2022

The useful limits

Mohr has also been explicit about the boundaries of the technology. In a public explanation of DARVIS’s object-detection approach, he described concentrating on objects and avoiding identification of the person. The distinction places a limit on what the system is trying to learn from a scene. For a founder working with cameras, that is a consequential product choice.

In 2022, he said customers could train models through the platform and deploy them on their own premises. He described an interest in controlling security and the operating environment. Deployment location, then, was another part of the map. The relevant room included the machines processing the data, as well as the objects being observed.

His discussion of future opportunities was equally grounded in the physical difficulties. He pointed to distributed computing, multiple models and the limits of embedding models in chips. He also saw value in combining different sensor inputs. An algorithm arrives in a world with hardware constraints; the world declines to excuse them on the grounds that the demonstration looked good.

Those details temper the story in a useful way. The work asks for choices about inputs, processing and permissions. It also asks for the patience to fit a product into an existing operation. Mohr described spending two years developing the company’s solution on Messe Berlin’s grounds. A site offers the kind of feedback a diagram cannot: things have to happen there.

A different chair at the same table

In August 2025, DARVIS announced Michael Prokopis as its new CEO. Mohr and Michael Dietz, the former Co-CEOs, would continue as Chief Product Officer and Chief Financial Officer respectively. For Mohr, responsibility for the product became the explicit title. He remained inside the business he had helped create a decade earlier.

A founder’s career can be described through the companies he starts, but it also includes the jobs he occupies within them. Here, the company had already changed its name and focus. The leadership arrangement changed too. The public announcement sets out the new responsibilities without requiring a grand theory of what anyone privately felt about them.

Earlier that year, Mohr joined Oliver Rößling and Alois Krtil on DEEPTECH DEEPTALK to discuss European AI entrepreneurship, investment, talent and policy. His interests extend beyond a single product to the conditions under which technology businesses can be built. Finding an application and finding an environment where a company can develop it are related pieces of work.

Coffee with people who build things

There is a smaller scene in Hamburg. A MAKERs’ Breakfast invitation names Mohr and Rößling as hosts, bringing together people working on hardware, drones, robots, sensors and autonomous systems. The invitation asks what they are making, what already works, what still does not, and whom they ought to meet. Coffee and bread rolls make an appearance. Mercifully, the questions can be answered without a keynote.

That local table sits comfortably alongside the wider geography of his career. A business degree, software transactions, wine matching, immersive data and operational mapping make for an untidy sequence on paper. Follow the practical questions, though, and the route becomes easier to read. What is available? Where is it? What can someone do with that knowledge?

A shelf does not volunteer an inventory report. A trolley does not explain why it has stopped moving. People have to supply that information, or build a way to collect it. Mohr’s work has increasingly concerned the second possibility. His career gives the question a human scale: several changes of direction, collaborators met along the way, and a continuing interest in making the physical world intelligible.