Physical AI Is Ready, Buildings Are Not | Kateryna Portmann, ANYbotics
Kateryna Portmann of ANYbotics and Women in Robotics on physical AI commercialization, robotics scaling in the DACH region and why infrastructure, not algorithms, is the bottleneck.
Key insights: executive summary
- Physical AI has left the lab: legged inspection robots are already running commercial shifts in industrial plants across Switzerland and the DACH region.
- The bottleneck for robotics scaling is no longer autonomy software but the built environment: doors, stairs, network coverage and safety processes designed for humans only.
- Deep tech commercialization is an integration problem. Revenue arrives when the robot slots into an existing maintenance workflow, not when a demo impresses.
- Diverse engineering teams shorten the path from prototype to deployment because real deployment sites are messy, human and full of edge cases.
- Zurich's density of ETH Zurich spinoffs gives hardware startups access to talent that has already shipped robots, which compresses years off a hardware roadmap.
“Physical AI is ready. The buildings, processes and safety rules it has to work inside are the part that still needs upgrading.”
“A robot only becomes a product when someone's Monday morning gets easier because of it.”
How is Switzerland leading the physical AI revolution?
Switzerland concentrates three things physical AI needs at once: world-class robotics research at ETH Zurich, industrial customers with high-value inspection and maintenance problems, and investors comfortable with hardware timelines. ANYbotics, one of the best known ETH Zurich spinoffs, is the clearest example of that pipeline turning research into deployed machines.
The Swiss tech ecosystem's advantage is proximity. Lab, workshop, pilot customer and investor sit inside an hour of each other, so a hardware iteration loop that takes a quarter elsewhere can take weeks in Zurich.
From lab research to real-world deployment
Moving a robot from a controlled lab to an oil and gas plant or an energy site means solving dust, heat, connectivity and permissions before anyone talks about autonomy.
Why hardware innovation needs patient capital
Swiss venture capital and instruments such as Innosuisse funding let robotics teams survive the long validation phase that pure software startups never face.
What does robotics scaling in the DACH region actually require?
Scaling in DACH is a fleet problem, not a robot problem: remote operation, spare parts, service contracts, on-site training and compliance with local safety regimes. Each new customer site adds integration work that has to be productised or it eats the margin.
Why do diverse engineering teams matter in robotics?
Deployment sites are unpredictable and the people who work in them are not a monoculture. Teams with varied backgrounds catch failure modes and usability problems earlier, which is why the Women in Robotics community matters commercially and not only culturally.
Who is Kateryna Portmann?
Kateryna Portmann is a Robotics leader and Women in Robotics community builder at ANYbotics, part of the Zurich deep tech ecosystem this series documents. Topics covered in this episode: Physical AI commercialization, Robotics scaling DACH, ETH Zurich spinoffs, Women in robotics, Hardware innovation.
Visit ANYboticsEpisode show notes
When Kateryna Portmann was a kid in Ukraine, she asked her parents for a robot. They kept bringing her dolls. She works in robotics now. She is Senior Product Manager at ANYbotics and she runs the Women in Robotics Society. We drove through Zürich and talked about why robot deployments actually fail. Her answer is not the robot. Assume the robot works. The problem is the building, the workflows, and the people who were already there.
Follow the Swiss tech ecosystem, one drive at a time
New in-car interviews with founders, investors and researchers building Zurich's deep tech, AI and robotics scene.