At 77, a Berkeley Alum Who Started on Punch Cards Is Using AI to Rebuild a Global Real Estate Platform

In the 1960s, Karl Lingenfelder was writing code on punch cards as a student at UC Berkeley, in the earliest days of computing as a discipline. Six decades later, he's back at the keyboard — this time working alongside an AI system to rebuild viewr.com, a real estate platform he originally launched in 2004, into something closer to a technology company than a traditional property listing site. These days, he works from wherever he happens to be — the latest chapter for a founder who has long favored a global, on-the-move lifestyle over a fixed home base.

Viewr.com now covers listings in more than 110 countries, spanning roughly 200,000 properties. But the number that makes the platform unusual isn't its size — it's who's building it, and how.

Lingenfelder is viewr's sole founder and its only hands-on developer. There is no engineering team. Instead, for the technical work of rebuilding and running the platform — writing code, debugging servers, designing features — he works directly with Claude, an AI model built by Anthropic, treating it less like a search tool and more like a technical collaborator sitting across the desk.

“I'm a hacker who makes mistakes,” Lingenfelder said of the arrangement. “Claude is my coder.”

The results are concrete. Every property listing on viewr.com now gets an automatically generated video, assembled from the listing's own photos, at a scale that would typically require a production team. The site's centerpiece is a live, interactive 3D globe that lets visitors browse real estate visually, spinning the earth to explore listings by country rather than scrolling through search filters. Behind the scenes, custom import tooling ingests property data from dozens of real estate feed providers across different countries and formats — the kind of infrastructure work usually handled by a backend engineering team, not a single founder in his late seventies.

It's a notable data point in a broader, still-unfolding question: what happens when AI models become capable enough that a single, technically fluent founder can do the work that once required a funded team. Lingenfelder isn't a professional software engineer by trade — his Berkeley-era computing experience predates modern programming by decades — but he's spent the past year directing a genuine, full-stack platform rebuild: server infrastructure, database architecture, video generation pipelines, and a from-scratch 3D visualization engine, largely by describing what he wants and iterating with Claude on the implementation.

The platform itself reflects that positioning. Rather than compete as another real estate listing aggregator, viewr describes itself as “the global position on real estate” — leaning into its technology, rather than its listings volume, as the differentiator.

For Lingenfelder, the project is also personal. He's spent decades around computing, watching it evolve from punch cards to mainframes to the AI systems now capable of holding up their end of a real, technical working relationship. At 77, he's not slowing that curiosity down — he's putting it back to work.