LocalRAG!

Supports 25 different file formats, including documents (PDF, DOCX, EPUB, TXT, etc.), images (JPG, PNG, HEIC), audio (MP3, WAV, FLAC), and video (MP4, MOV), allowing users to create a single searchable collection.

All original files, text extraction, indexing, and search operations are performed directly on the user's device, ensuring that sensitive data never leaves the device or is uploaded to the cloud.

When Local AI is selected and the necessary models are downloaded, the tool can generate answers to questions about documents entirely on-device and without an internet connection.

Provides natural language answers with inline citations (e.g., [1][2]) that users can tap to jump directly to the exact cited page or timestamp within the source document, with the passage highlighted.

Users can choose between using on-device AI models for a fully local and offline workflow, or selecting cloud-based Claude models for potentially higher-quality answer generation, with privacy safeguards for cloud interactions.

Utilizes Okapi BM25 search combined with an on-device embedding model (E5 Multilingual) to enable semantic search across different languages, allowing users to ask questions in one language and find answers in documents written in another.

Allows importing photos of whiteboards, receipts, or scanned documents, performing on-device Optical Character Recognition (OCR) to extract text, and providing an editor to refine the extracted text before indexing.