Skip to Main Content

Loreg

Built at Meta Global AI Developer Hackathon · Oct 26, 2026 · Remote

Loreg — Demo video

Loreg (LOW-reg) is a smart log and register for small electronics and appliance shops. It shows the owner how much money is tied up in stock that has stopped selling, and warns them when a new product launch is about to make that stock harder to move. Loreg is built on a deterministic Python core that computes every number, including days since last sale, days of stock and capital at risk. Muse Spark, through the Meta Model API, only explains those numbers and picks one option per at-risk item (discount, bundle, return or hold) from the choices code allows. Every answer cites its SKUs, figures and dated sources, and code checks it before the owner sees it; a failed answer falls back to a labelled rules answer, never a wrong number. The owner accepts, declines or undoes each option, and the capital-at-risk figure updates. All data in the demo is synthetic, Loreg Watch reads one news source today, and the owner always makes the final call.

Team