FileFinder

AI-powered semantic file search and chat for your local files. Find any document by meaning, find any photo by describing what's in it, and chat with your files — all on your own machine.

---

✨ Features

• 🔍 Semantic search — search documents (DOCX, PDF, TXT, XLSX, PPTX, MD) by meaning, not just keywords (BGE-m3 embeddings + hybrid keyword score)
• 🖼️ Image content search — describe what's in a photo ("red car", "id card") and find it — CLIP vision-language model, multilingual (English / 中文 / O'zbek)
• 📝 OCR built-in — text from images and scanned PDFs (Tesseract chi_sim+eng), auto-installed on first run with poppler
• 💬 Chat with your files — RAG chat with streaming responses and collapsible thinking blocks; attach files with the + button (DeepSeek Cloud built-in, Ollama, or Custom)
• 👁️ Automatic indexing — standard user folders + all fixed drives indexed automatically on launch; live file watching keeps everything in sync
• 🌐 3 languages — 中文 / O'zbek / English UI
• 🌙 Dark & Light themes, system tray with global hotkey (Alt+Q)
• 🔄 Auto-update — new versions are delivered via GitHub Releases
• 🔒 Private by design — all indexing and embeddings run locally; nothing leaves your machine except the LLM chat API call you choose

📸 Screenshots

Chat with files (RAG)	Index management

Settings & Pro license

⬇️ Download

Get the latest release from the Releases page:

Package	When to use
FileFinder-Setup-x.y.z.exe	Standard installer for most Windows PCs
FileFinder-x.y.z.zip	PCs with Smart App Control enabled (signed-Python setup, no unsigned executables)
FileFinder-vX.Y.Z-windows.zip	Used by the in-app auto-updater

First launch is fully automatic: embedding models (BGE + CLIP) download themselves, Tesseract OCR and poppler install themselves, your standard folders are indexed in the background. No manual setup.

💎 Free vs Pro

	Free	💎 Pro (¥19.9/month)
Semantic file search	✅ Unlimited	✅ Unlimited
Image content search	✅ Unlimited	✅ Unlimited
OCR	✅ Unlimited	✅ Unlimited
File indexing	✅ Unlimited	✅ Unlimited
AI Chat (DeepSeek Cloud built-in)	20 queries	✅ Unlimited
RAG chat + file attachments	—	✅
Export results (CSV/Excel)	—	✅
Watched folders	1	✅ Up to 20

Activate Pro: open Settings → upgrade → pay with WeChat Pay or Alipay (QR code in the app), then enter your activation key (FF3-XXXX-XXXX-XXXX).

🚀 Run from source

git clone https://github.com/BotirBakhtiyarov/Filefinder.git
cd Filefinder
python -m venv .venv && .venv\Scripts\pip install -r requirements.txt
.venv\Scripts\python -m filefinder.main

Run the tests:

.venv\Scripts\python -m pytest tests/ -q

🛠 Tech stack

Area	Technology
UI	Flet (Flutter for Python)
Text embeddings	BAAI/bge-m3 (ONNX Runtime, ~2.3 GB, auto-downloaded)
Image embeddings	CLIP ViT-B/32 multilingual (ONNX, auto-downloaded)
OCR	Tesseract (chi_sim+eng) + poppler — auto-installed
Vector store	ChromaDB (desktop) / sqlite-vec (Android, planned)
LLM	DeepSeek (built-in cloud), Ollama, custom OpenAI-compatible
Packaging	flet build + Inno Setup / SAC-safe zip

🔐 Privacy

FileFinder indexes your files locally. OCR results, embeddings and the search index never leave your computer. The only outbound calls are: model downloads (HuggingFace), update checks (GitHub), and the chat API you configure (DeepSeek/Ollama/custom).

👤 Author

Botir Bakhtiyarov — GitHub · WeChat: nick_2620