Mortise: Private On-Device AI

Mortise runs open language models entirely on your own device. There is no server to trust, because there is no server at all — no account, no sign-in, no cloud inference, no telemetry, and nothing metered. A MODEL CATALOG THAT FITS YOUR HARDWARE A curated list of open-weight models, most of them licensed Apache 2.0 or MIT. Every entry carries an on-device memory estimate, so you can see what your Mac, iPad, or iPhone will actually run before you download anything. The weights land on your own disk and run for as long as you like. CHAT WITH YOUR DOCUMENTS Attach a PDF, Word document, or PowerPoint deck and ask questions about it. Mortise indexes it locally and picks the right retrieval strategy — whole document, chunked, or summarized — then answers citing the page or slide it drew from, shown beside the answer. The file never leaves the machine. BUILD A SEARCHABLE LIBRARY Group related…