01
Mechanism
Posha uses a Tesla-like data collection strategy where customer deployments generate training data rather than expensive supervised learning
Ragav explains that instead of spending heavily on chef-supervised training like Waymo did with driving data, Posha lets customers use the robot while it collects culinary data, creating a flywheel similar to Tesla's approach.
transcript
Ragav: In the early days we had chefs alone training Porsche. Now we have our customers training Porcha. We possibly have the world's largest culinary vision data set for cooking food. And we are taking the Tesla approach. Veos spent 500,000 hours of supervised driving and $20 billion in venture capital to crack self-driving. Tesla did it a different way. Tesla said we'll release something that's useful to consumers. also use it to collect tons of data and then put robot taxis out.