Amanece

Amanece is a WhatsApp agent for every small business in Latin America. In Colombia, Mexico, Peru and most of the region, small businesses sell through WhatsApp: a bakery, a coffee roaster, a clinic, a courier. The owner answers every message by hand, at night and on weekends, and loses the sale when they can't. Most of them can't hire a team to answer, and they will never configure a chatbot builder. With Amanece, the owner just tells it what they do. Paste your website or describe your business, and the owner's agent reads the page, finds your products and prices, asks only what it still needs (how do your customers pay you?), and opens your business with a wa.me link to share. From that moment, a customer agent answers your customers on WhatsApp from your catalog, takes and confirms orders, books appointments, and opens a Case for you when something needs your decision. The owner never leaves the conversation. "If someone asks for shipping outside the city, don't promise a date and open a Case." "Show me confirmed orders per day." The owner's agent turns that into a rule or an Indicator, tests it against the business's own data, and shows the result in a preview. Nothing reaches a customer until the owner taps Apply and then Publish. Every change can be reverted. What exists today, built entirely during the hackathon (first commit October 26, 274 commits): the public landing in Spanish and English, sign-up, the conversational setup, the WhatsApp customer agent on a real number (text Café Ladera, our sample business, from app.amanece.co), the Panel with Home, Cases, Conversations, Orders, Appointments, Customers and Board, rules and knowledge, Indicators and Sections written as bounded SQL, Cases with a drafted reply the owner sends with one tap, taking over a conversation, reading photos and PDFs, and Publish with preview. What we measured on Muse Spark: a change costs USD 0.03 to 0.04 at standard pricing and takes about 21 seconds; zero data leaks between businesses across 35 changes; bounded SQL gets 88% of Indicators right against 56% for generated JavaScript. Thirty-three test suites run against a real Neon database and the real Meta Model API. Amanece runs in Spanish. The demo video is narrated in English. Café Ladera and its customers are fictional.