Article · July 2, 2026

Physical AI has to fit the workflow

In July 2026, you had to step around AI on the pavement in Davos. More than fifty autonomous systems worked in shops, hotels, the station plaza and a supermarket.

This was RobotCity, part of Davos Tech Summit (opens in a new tab), held from 1 to 4 July under the banner Touching Intelligence. A cart guided a blind shopper. Delivery robots rode hotel lifts. Firefighting robots and rescue drones worked with the local fire brigade. Our team designed the summit’s scenography, from the RobotCity flags to the robot signs.

A red tracked firefighting robot spraying a jet of water across a car park in Davos.
A firefighting robot at work in Davos.

What deployment means

#

The Robot City Award was judged by a jury walking the town, not by pitches. Most Impactful Real-World Deployment went to ANYmal, from the Swiss company ANYbotics: an inspection robot whose data flows into SAP, where the business already works. Pudu’s FlashBot MAX took the Robot for Society award.

The lesson travels. A robot, like any AI system, becomes useful when its output reaches the systems people already use.

Why a robot in a supermarket is a governance question

#

A public setting makes the operating questions visible: who is accountable when the system makes an error, what data it records, who may see that data, and how staff work alongside it. Those need design and operating decisions, not just a better model.

A service robot with a screen face moving between bread shelves in a supermarket.
Between the shelves of a Davos supermarket.

It is the same argument as in software: once generation gets cheap, what gets expensive is everything around it. Physical AI just makes it visible from the pavement.

What it means for your project

#

Whether the AI moves through a supermarket or drafts a reply, plan the integration and the operating rules together with the model. Our ticket-change copilot works inside the booking system travel agents already use, through the company’s single sign-on, and records every request it sends.

Getting an AI system from a working demo into something a business can operate is most of what we do. See advisory & implementation, or Start a conversation.

Download as PDF

Have a project in mind?

Tell us what you are building or improving. We'll help you choose a practical next step.