FutureVision

FutureVision is the intelligence layer that Rhoda AI unveiled on 10 March 2026 when it left eighteen months of stealth with a 450 million dollar Series A. The company, based in Palo Alto, was founded by chief executive Jagdeep Singh, chief science officer Eric Ryan Chan and Stanford professor Gordon Wetzstein, who runs the university's Computational Imaging Lab. The round drew Capricorn Investment Group, Khosla Ventures, Leitmotif, Matter Venture Partners, Mayfield, Premji Invest, Prelude Ventures, Temasek and Xora, with John Doerr also participating. The architecture is called Direct Video Action, and it is the reason the company is worth a profile. Instead of mapping observations straight to joint commands, the model predicts what the near future should look like — as video — and then converts that prediction into actions, executes them and re-observes, closing the loop continuously. Pretraining runs on hundreds of millions of internet videos to build a working sense of motion and physics; post-training adds robot-specific data. Rhoda's claim from this is a data argument rather than a benchmark one: new behaviours often need as little as ten hours of teleoperation data. Evidence of it working outside the lab is still thin and anonymised. The company cites one high-volume manufacturing evaluation in which robots completed a component-processing workflow in under two minutes per cycle without human intervention; the customer is not named and no independent measurement has been published. Rhoda says FutureVision will drive its own robot platforms and, over time, be licensed to partners across other hardware and software stacks. This profile is kept at pilot status deliberately. As of August 2026 what exists publicly is a funding announcement, an architecture description and a single unnamed deployment — no product page, no specification sheet and no shipping platform. It will be upgraded when Rhoda names a customer or a product.

▸Videos

No videos yet.

▸Reviews