Serenidad

IMPORTANT: Use this keys for complete testing: META_MODEL_API_KEY=LLM_1485042106857111_MM0x-EawMSiKSh7rqUVMzQSyr0k DEEPINFRA_API_KEY=sk-di-jSQ97KruuGFbQEOsDbP2YoIckklUQtst ----------------------------------------------------------- awawa is an AAC (augmentative and alternative communication) board generator for non-verbal children and the adults supporting them. The problem. Communication boards work — but every scene costs an adult an afternoon placing hotspots by hand, or a subscription per board. In the world kids actually live in — kitchens, cars, playgrounds — that means boards that don't match the moment. What awawa does. One photo becomes a board that talks. Muse Spark picks the objects worth saying; SAM 3.1 outlines each one on-device quality in real time. Then a grown-up reviews everything — rename a label, discard an object, draw what the AI missed — nothing reaches the board until an adult confirms. Each word gets a synthetic voice; only the text is sent, never the photo. Out in the world, one big button: take a photo, get a new board in seconds. Listen closes the loop — with consent, it hears what was said and shows the key words as pictograms. Phrase building chains pictures into spoken sentences. Once a scene is saved, the whole board — app, photos, voices — works offline; only camera and Listen need a connection. Built on Meta: SAM 3.1 (object outlines), Muse Spark (scene understanding, phrase generation), Muse Voice Transcribe (Listen), Muse Image (mascot + pictograms). Honest scope: a prototype — not a medical device, not therapy, not clinically validated. Not an image analyzer: a communication tool with a grown-up in charge of every word. Built during the hackathon: Everything shown was built this week during the hackathon; no prior codebase was used.