{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/21d36102972640a48b2e2c9cad28a97c\" frameborder=\"0\" width=\"1720\" height=\"1290\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1290,"width":1720,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1290,"thumbnail_width":1720,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/21d36102972640a48b2e2c9cad28a97c-7761d510c728db26.gif","duration":232.427,"title":"Evolving Expired Patents into New Designs for Better Physics and Additive Manufacturing","description":"This Loom demonstrates how Exodar evolves expired patents into novel, additive-manufacturing-ready designs. It first uses the hog API to identify potential customers, for example a mechanical parts quality engineer at NASA, by scanning platforms like LinkedIn, Twitter, and Reddit, then converts the patent PDF into an image and uses text-to-CAD with CADQuery to generate geometry. Using an evolutionary algorithm paired with an LLM, it evolves the design to improve novelty, manufacturability, and physics, increasing a baseline score. The output includes an invention card, LLM reasoning, and a manufacturing brief, shown with a shock absorber example where struts and a loft improve strength and load paths for additive manufacturing."}