Manifest0 AI

Setting up a new hazardous waste customer is manual. A field rep walks the site, writes down every container and photographs the labels. Back at a desk, they search for each product's Safety Data Sheet (SDS) one at a time and type everything up. That takes hours to days, and every missing detail means another call or visit. Manifest0 AI does that work during the site walk. The rep starts a visit and uploads label photos room by room. Muse Spark 1.3 reads every photo in a batch in one vision call and returns strict JSON for each product: manufacturer, product name, product code, container, signal word, DOT hazard class, UN number, hazard statements, pictograms and ingredients. It merges front and back shots and remembers what it already found in that room, so overlapping photos update a product instead of adding a duplicate. For each new product, the app first checks a shared SDS library. If the sheet is not there, Muse Spark uses Meta's web search tool to find the official SDS PDF and must say whether the document really is an SDS. Confirmed PDFs are saved and reused on later visits. The rep reviews the draft and taps Send to client, and the inventory and SDS PDFs go out by email. On our 3 sample products, 10 of 10 checkable label fields were correct, and an SDS was found for all 3 (confirmed right for 2). The app is live for approved beta testers. Muse Spark does not assign EPA waste codes; a person reviews everything before it is sent.