When AI Agents and Robots Enter the Physical World | Nils Pihl

In this talk

1. 10:03

   A shared spatial layer for devices

   Pihl says Hocke Labs works on collaborative perception, mapping, positioning, and getting physical AI devices into use. Its system lets phones, glasses, cameras, and robots share a coordinate system, so information gathered by one device can guide another. He describes it as a decentralized network, with data kept on local or self-hosted systems.

2. 11:12

   Retail workers and robots working together

   Pihl describes an augmented reality task manager where store managers can leave workers location-based notes with photos and instructions. One large European retailer estimated that its 17,000 employees lose 30 to 60 minutes per person each day, often because workers do not know where to begin. A robot could identify tasks and direct people, while workers handle jobs such as restocking that robots may not do as quickly.

3. 22:52

   Open robots and near-term revenue

   Pihl favors programmable robots and expects robot capabilities to be sold like apps for at least the next decade. He points to Unitree, AGIBOT, and Booster as companies embracing developer access, and argues that closed systems make integration and deployment harder. He sees commercial potential in perception work, including mall security robots charging $5,000 a month and deploying Chinese robots already in the US with more usable English-language software.

4. 28:49

   Why China is moving quickly in robotics

   Pihl credits China's progress to government funding, dense populations and supply chains, and urban conditions that encourage computer vision and positioning research. He says grants and facilities can support robotics teams even when venture funding is limited, while local governments also back humanoid initiatives. He notes that many humanoid companies may fail, and that government support can also weaken commercial focus.

5. 41:25

   Physical AI and manufacturing's future

   Pihl argues for balancing globally connected manufacturing with the ability to build critical products domestically. He expects AI to move toward edge devices because robots and glasses need nearby computing and low latency. He predicts more economic activity will center on physical work, with AI glasses supporting workers and urban Chinese households potentially spending more on robots than cars within a decade.