FileNest
Index, search, and chat with the files already on your computer using local AI. In local mode, content, indexes, notes, and conversations stay on this device.
Version 0.2.4 · 15.1 MB · macOS 13 or later
Private index
Grounded answers, local vectors
One library for the things scattered across your desktop
See it in motion
A short product film with the current FileNest UI and a fictional, privacy-safe workspace.
Five workflows. One private workspace.
Explore a guided product simulation built entirely with fictional Northstar Studio files. Nothing below comes from a real user or device.
The deck from our spring offsite about Aurora pricing
Find the research that changed our onboarding plan.
Searching notes, documents, and meeting summariesFileNestThese two files connect the research findings to the revised onboarding flow.
What are the launch risks, and who owns each one?
FileNestGrounded answerThe brief identifies three launch risks:
Showing Indexing
Smart Search turns everyday language into keywords, dates, file types, and semantic intent. Hybrid retrieval brings the strongest signals together, then shows why a result matched.
FileNest waits until downloads and project folders are stable, indexes before moving, and follows your ordered rules. Unmatched files can be classified by local AI, while installers and temporary files stay untouched.
FileNest manages local Ollama models and keeps chat, embeddings, and OCR independently configurable. Cloud providers remain optional and visible.
Local-first is not a setting hidden in preferences.
File metadata, extracted content, embeddings, notes, and conversations are stored on your device by default. Local models can handle search, OCR, and answers without uploading document content.
One product model, shaped for the conventions of each operating system.
A SwiftUI app with menu bar presence, Finder integration, native previews, system appearance, and global quick search. macOS 13 and later.
Available now for macOSA Windows 11 implementation with tray workflows, native shell actions, secure credential storage, and matching search, chat, indexing, and rules.
Public preview available for Windows
Open source, in the open
FileNest is released under the MIT License. The macOS and Windows source, automated tests, issue tracker, and contribution guide are all public on GitHub.
Release installers are published as GitHub Release assets and are not committed to the source repository.
Clear answers, before you trust an app with your files.
Not in local mode. File contents, indexes, notes, and conversations stay on your computer. If you explicitly configure a cloud model, only the requests for that provider are sent to its API.
It can extract text and structure from PDFs, Office documents, images, archives, project folders, and—when enabled—audio and video transcripts. OCR and parser fallbacks depend on your setup.
FileNest checks that files and directories are stable before processing them. It also ignores locks, temporary downloads, installers, and unsupported items by default.
Yes. You can run compatible local models with Ollama or configure OpenAI-compatible and Anthropic cloud providers separately for chat, embeddings, and OCR.
Available for macOS and Windows
Download the notarized macOS app or get the Windows 11 installer from the public GitHub release.