Dot App
Dot is an AI tool developed for the privacy-conscious users leveraging the power of language models. Engineered to run locally on the user's device, Dot facilit...
Last verified:
What is Dot App?
Dot is a standalone, open-source AI application that enables users to interact with their documents using powerful local large language models (LLMs) and Retrieval Augmented Generation (RAG), all without sending data to external servers. It supports loading multiple files like PDFs, DOCX, PPTX, XLSX, and Markdown from local directories, allowing natural language chats to extract insights, summarize content, or answer questions based on the documents. The tool runs entirely offline on your device, powered by models such as Mistral 7B or Phi-3.5, ensuring complete data privacy and security for sensitive information.
Key features include a user-friendly chat interface for document-specific queries and 'Big Dot' mode for general conversations unrelated to uploaded files, mimicking experiences like ChatGPT but locally. Built with Electron JS, it leverages libraries like FAISS for vector stores, Langchain, llama.cpp, and Huggingface for efficient conversation chains and document processing. Users can download pre-built apps for Apple Silicon and Windows, with Linux support upcoming.
Dot is ideal for professionals handling confidential documents, such as lawyers, researchers, doctors, and businesses needing secure AI assistance without cloud risks. It caters to non-technical users seeking simplicity, privacy-focused individuals avoiding data leaks, and developers interested in local AI experimentation. With recommended specs of 6GB storage and 16GB RAM, it democratizes advanced AI for personal computing.
Dot App pricing
Pricing model: Free
Dot is completely free with no paid plans or tiers. Download apps directly from the website for Apple Silicon, Windows (Linux coming soon). Open-source under GPL-3.0, no subscriptions, usage limits, or hidden costs.
Dot App pros
- Runs entirely offline without internet
- Zero data sent to external servers
- Supports PDF, DOCX, PPTX, XLSX, Markdown
- Powered by Mistral 7B and Phi-3.5 LLMs
- Uses RAG for accurate document responses
- Big Dot mode for general AI chats
- Open-source with GitHub access
- User-friendly Electron JS interface
- FAISS for fast local vector search
- Langchain and llama.cpp integration
- Load multiple docs from directories
- GPU acceleration with Nvidia CUDA 12.2
- No subscription or usage fees
- Privacy guaranteed by local processing
- Inspired by Chat with RTX but simpler
- Future Linux support planned
- Choice of LLMs available
- Enhanced security for local LLMs
Dot App cons
- Requires 16GB RAM recommended
- 6GB storage minimum needed
- Windows needs C++ Redistributable
- Nvidia GPU for acceleration only
- Linux support not yet available
- Windows Defender false positives
- No code signing due to costs
- Apple Silicon only for Macs
- No mobile app support
- Dev setup needs Node.js and npm
Frequently asked questions about Dot App
What documents does Dot support?
Dot supports loading PDFs, DOCX, PPTX, XLSX, Markdown, and more from any local directory. Users select folders, and the app indexes them for chat interactions using RAG.
Is Dot completely offline?
Yes, Dot runs 100% locally on your device with no internet required. All processing uses local LLMs like Mistral 7B or Phi-3.5, ensuring no data leaves your computer.
What are the system requirements?
Recommended: 6GB storage and 16GB RAM. Windows needs Visual Studio C++ Redistributable; Nvidia GPU with CUDA 12.2 for acceleration. Apple Silicon and Windows downloads available.
How does Dot ensure privacy?
By operating fully locally, Dot never sends documents or data to servers. It uses open-source libraries like FAISS, Langchain, and llama.cpp for secure, on-device AI processing.
What is Big Dot mode?
Big Dot allows general conversations unrelated to documents, like asking for recipes, code examples, or rap battles, powered by the latest local Mistral model, similar to ChatGPT.
Is Dot open-source?
Yes, Dot is open-source under GPL-3.0. Check the source code on GitHub at pixiumdigital/DotApp or original alexpinel/Dot to verify and contribute.
Can I use GPU acceleration?
Yes, on Windows with Nvidia GPU and CUDA toolkit 12.2 installed. This speeds up LLM inference; CPU fallback available but slower.
Why does Windows Defender flag it?
As an open-source project without expensive code signing, Defender may flag it. Users should check GitHub source code; it's safe and community-verified.
How to install for developers?
Clone the repo from GitHub, install Node.js, run npm install (use --force if needed), cd to aadotllm, npm install again, then build with Electron.
What future features are planned?
Upcoming: Full Linux support, image file handling, individual file selection, more document types, better file database management, and enhanced LLM security.