Skip to Main Content

Nzangi

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

Nzangi — Demo video

DukaBrain AI – See Your Stock. Know Your Business. DukaBrain AI is an AI-powered visual inventory management solution designed to help small retailers make stocktaking faster, easier, and more reliable. Many small businesses still depend on manual stock counts and handwritten or outdated records, making it difficult to maintain accurate inventory, identify discrepancies, and know when products need restocking. DukaBrain AI addresses this challenge by allowing shopkeepers to upload photographs of their store shelves and use Meta's SAM 3.1 to detect and segment visible product instances. The system then provides an interactive review process where users can verify, correct, or add missed detections before confirming the physical stock quantities. These verified counts are compared against existing inventory records to highlight stock differences and potential shortages. Powered by Meta Muse Spark Open, DukaBrain AI also transforms verified stocktake results into clear explanations, useful insights, and practical recommendations to support better inventory decisions. The platform maintains stocktake records and reports, helping retailers monitor inventory changes over time. The solution combines computer vision, human verification, deterministic inventory reconciliation, and AI-powered analysis in one accessible platform. Rather than replacing the shopkeeper's judgment, DukaBrain AI gives small retailers better tools to understand their stock, reduce the burden of manual counting, and make more informed business decisions. Our vision is to make intelligent inventory management more accessible to small businesses, especially retailers who cannot afford complex enterprise inventory systems.

Team