Jason's two-sentence pitch for why Dyna is different: they do both the research and the product, and the best way to succeed is to build research and product at the same time. The feedback loop from product to research and research to product is what makes existing AI products so good and so sticky.
More from this episode
Jason's pitch for why Dyna differs from other robotics companies: "at dyno we do both the research and also the product, and I think the best way to succeed is to build research and product at the same time." He points to ChatGPT as a product you can use, but with a lot of research behind it to make it good — "the feedback loop from product to research and from research to product is what makes the existing AI products on the market like so good and so sticky." Some robotics companies are purely trying to take existing technologies and turn them into a product, but he thinks existing technology as it is isn't ready to get to general-purpose manipulation. At the same time, if you only focus on research, you can end up working on toy versions of real-world tasks that may not be representative of actual deployment. At Dyna they go straight to the real version of the task, so if their research is good enough to solve it, they can immediately turn it into product — "that feedback loop makes our technology and business much more solid."