Nortex

MuseNortex lets small business owners in Nicaragua ask about sales, stock, and products by voice. Powered by Meta Muse Spark and connected to Nortex through MCP, it turns business records into clear spoken answers in Spanish, in the middle of the workday. At a pulpería, a neighborhood convenience shop in Nicaragua, the owner may be serving customers, counting change, and checking shelves at the same time. A question such as "How did we do last week?" should be easy to answer during that working day. MuseNortex is a voice assistant connected to Nortex, a business management platform for small businesses in Nicaragua. It lets an owner ask about sales, inventory, and products in everyday language and hear the answer. In our prototype, the owner asks in Spanish and the assistant answers in short spoken sentences designed to be heard behind the counter. Meta Muse Spark, accessed through the Meta Model API, provides the reasoning and tool-use layer. Nortex's Model Context Protocol server connects that conversation to business information. Nortex remains the source of the sales and inventory records; the voice experience makes those records easier to use. Our demonstration focuses on practical questions an owner actually asks: what is running low, how much stock is left, a product's price, a sales comparison between two explicit date ranges (October 25–31 vs. October 18–24, 2026), and the state of an open stock count. It uses synthetic shop records with real queries and live responses: the session in the video was recorded live on November 3, 2026, with Meta Muse Spark calling the production MCP server. The demonstration scope is read-only. We are starting with the working conditions of small shops in Nicaragua, with Spanish as the target operating language. The assistant depends on information recorded and synchronized in Nortex, and fresh voice answers require an internet connection. The project builds on an existing Nortex platform and an initial MuseNortex technical core. Those foundations predate the event and are disclosed as pre-existing work. Our next step is a small merchant pilot measuring whether owners can get accurate answers with less effort, how often they return, and what each successful interaction costs. Our ambition is to make useful business information accessible in the middle of the workday, starting with the questions owners already ask.