direct-video-action
(2)
Jagdeep Singh Keeps Choosing the Problem That Might Not Work
After networks, batteries, and seven companies, Jagdeep Singh has arrived at robotics with the same old appetite: find a consequential problem, accept the possibility of failure, and keep building until the laboratory meets real life.
Rhoda AI
Rhoda AI is a Palo Alto robotics company building foundation models that teach robots to act in the messy physical world by learning from internet-scale video instead of staged lab demos. Its FutureVision system - a Direct Video Action (DVA) model - watches an environment, predicts what happens next, and converts those predictions into motion in a closed loop that updates every few hundred milliseconds. Founded in 2024 and out of stealth in March 2026 with a $450M Series A at a $1.7B valuation, Rhoda targets high-variability manufacturing and logistics work.