SmartSort AI
Semantic file organization without relying on file extensions.
Extension-based rules can't tell an invoice from a poster saved as the same PDF, and near-identical images quietly eat storage for years.
Metadata and embeddings feed a ChromaDB store for semantic search over local files; content and perceptual hashing surface true and near duplicates, with storage analytics recommending cleanups.
Perceptual hashing beat a classifier here — the useful signal for near-duplicate images was structural, not semantic, and a model would've been the expensive wrong answer. The first version shipped as a dry run only: it shows you the plan before it touches a single file.
A sorted file tree plus a concrete cleanup plan, built without a single manual folder rule.