Dimensional Seeks Workers to Walk Robot Dogs

The San Francisco startup is paying $20 per hour for human help to train AI agents to navigate urban environments.

Updated on Oct. 28, 2026 in Robotics

A four-legged metallic robotic dog stands on a concrete sidewalk in a San Francisco industrial area. — San Francisco startup Dimensional has begun hiring workers to train its fleet of robot dogs, using physical navigation to improve AI agent performance. AI Illustration. Upload story photo >

Live Poll

Do you believe robots will successfully replace most manual labor jobs in your lifetime?

On October 5, 2026, San Francisco-based startup Dimensional began hiring workers to walk their fleet of robot dogs in the Dogpatch neighborhood. The company is using human operators to collect environmental data for training AI to handle physical tasks autonomously.

Why it matters

The company needs humans to intervene because current AI agents struggle to operate the hardware effectively without physical supervision. The initiative aims to help robots eventually function independently in factories, data centers, and construction sites.

Dimensional purchased 100 Unitree robot dogs for approximately $2,000 each and 50 humanoid robots to develop its open-source dimOS operating system. Each robot dog weighs 33 pounds and requires manual handling by workers if the unit shuts down during a four to six-hour shift.

The players

Dimensional

This San Francisco-based startup is developing an open-source operating system called dimOS to control robotic hardware.

Unitree

This robotics manufacturer supplied the 100 robot dog units currently used by the startup.

The details

Workers steer the robots using a mobile app or handheld remote while sensors record data for the company's dimOS platform. This follows a model where a human operator based in the Philippines remotely supervises an existing robot dog patrol at an Atlanta apartment complex.

October 5, 2026: The company posted the job advertisement for dog walkers.
October 28, 2026: The current status of the startup and its operations.

The Tech Race

Dimensional is building on the legacy of specialized robotic OS platforms like dimOS to transition from manual, remote-operated machines to fully autonomous agents. This strategy mirrors efforts by larger tech rivals to replace legacy remote-monitoring systems with AI capable of handling physical infrastructure.

Local residents in the Dogpatch neighborhood may see individuals guiding these 33-pound robots as they traverse urban terrain for data collection. Participants earn $20 per hour but must be physically prepared to carry the equipment in the event of a system failure.

The takeaway

The transition to autonomous robotics requires significant human-led data collection to overcome current limitations in AI navigation. Workers participating in such programs provide the physical edge case training necessary for machines to safely interact with complex environments.