SnapSell

SnapSell helps individual resellers turn a video of their goods into a practical selling workflow. Upload an English-narrated inventory video, review AI-extracted product drafts and photos, research comparable asking prices, and prepare listings for eBay, Facebook Marketplace and Craigslist. AMD Token Factory powers visual inventory extraction, product identity research, research planning, buyer-intent classification and evidence summaries. Local Whisper transcribes narration. Task-specific routing, bounded concurrency and intent reuse reduce repeated inference, while an execution dashboard exposes measured usage. The application uses Next.js, a Python backend and persistent SQLite storage. Pricing is computed from retrieved merchant evidence rather than invented by a language model: matching-source requirements, merchant-balanced recency weights, outlier filtering and weighted quartiles produce asking-price positions constrained by recorded costs, fees and seller minimums. Asking prices are clearly distinguished from completed sales, and forecasts remain unavailable without sufficient observed outcomes. Sellers review listing kits and buyer reply drafts before using them. Facebook Marketplace and Craigslist posting is manual; eBay API publication requires configured developer credentials and seller authorization. Pickup scheduling, Google Calendar integration and recorded sale contribution connect preparation to measurable outcomes. Contribution is calculated from actual recorded sale price, fees and acquisition cost; revenue uplift has not yet been established. Built on the MIT-licensed CollinQ/sbhacks2025 foundation, SnapSell extends video-to-listing automation into an evidence-based commerce workspace for Track 3: Reinvent Commerce.