{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/e9e0089db98740f18cde217bd9aff0a2\" frameborder=\"0\" width=\"1920\" height=\"1440\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1440,"width":1920,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1440,"thumbnail_width":1920,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/e9e0089db98740f18cde217bd9aff0a2-c522dab1075d0a8e.gif","duration":342.401,"title":"🌸 Moms Verdict Demo","description":"I’m presenting Mom’s World Day, a system for Mom’s World that reads up to 340 product reviews so a mom with a newborn does not have to. It runs end to end in Python with 7 decoupled states, and the LLM only extracts claims and generates Arabic, while everything else like fake detection, confidence scoring, and clustering is deterministic. I demoed bottle and stroller datasets, including one with suspiciously similar 30 reviews flagged by pairwise cosine similarity, and another with low volume that returns a safe verdict needing at least five reviews. The validator merges outputs and outputs a confidence score, like 0.88 in my main demo. I did not request any specific action from viewers."}