Find
Locate a filename, relative path, or indexed passage without generation.
Local AI for documents
FileNyani searches, compares, summarizes, and extracts from the files you already have. Models run locally, citations reopen the underlying evidence, and your folder stays an ordinary folder.
No account. No FileNyani cloud. Model preparation is explicit, and answers remain bounded by the evidence retrieved.
The agreement requires 60 days' written notice before renewal. S1
Either party may provide written notice at least 60 days before the renewal date.
One folder-native workspace
Browse the hierarchy and use deterministic Files and Content search first. Prepare local AI only when you want a grounded answer, comparison, overview, or extraction.
Locate a filename, relative path, or indexed passage without generation.
Ask focused questions and inspect citations against the current source revision.
Select files or folders and ask a focused question with evidence from each side.
Pull named fields across documents, review uncertain cells, and export CSV or JSON.
Select one file, several files, a folder, or mixed subtrees independently of navigation.
Return from an answer or extraction cell to the exact page, paragraph, row, or structured entry.
Trust boundary
FileNyani re-reads current source evidence before generation, checks source revisions, validates citation IDs, and keeps uncertain extraction visibly uncertain.
Small local models can omit, misread, or overstate evidence. Retrieval is bounded and does not prove that an answer, summary, or comparison is exhaustive.
No collection import, cloud knowledge base, or document reorganization is required. Workspace-owned state stays below .filenyani; model files remain in browser or application-owned storage.