Dot is a standalone, open-source desktop application enabling local interaction with documents via LLMs and RAG, inspired by Nvidia's Chat with RTX. It targets users without programming backgrounds, offering a simple interface to query PDFs, DOCX, PPTX, XLSX, and other files using the default Phi-3.5 LLM, with an additional "Big Dot" feature for general inquiries.

How It Works

Built with Electron JS, Dot leverages FAISS for local vector stores and Langchain, llama.cpp, and Huggingface for conversational AI chains. This stack facilitates efficient document indexing and retrieval within a fully local environment, avoiding external API dependencies and ensuring data privacy.

Quick Start & Requirements

• Download: Pre-built applications are available for Apple Silicon and Windows from the Dot website.
• Developer Install: Clone the repository, install Node.js, then run npm install (or npm install --force if issues arise) in the project root, followed by npm install in the aadotllm subdirectory.
• Resources: Requires Node.js. Specific hardware requirements for LLM inference are not detailed but are implied by the local LLM usage.

Highlighted Details

• Supports multiple document types (PDF, DOCX, PPTX, XLSX, and more).
• Fully local operation for privacy and accessibility.
• Includes a "Big Dot" feature for general LLM interaction.
• Default use of Phi-3.5 LLM for out-of-the-box functionality.

Maintenance & Community

The project is student-managed and welcomes contributions. Links to community channels or roadmaps are not provided in the README.

Licensing & Compatibility

The repository is licensed under the MIT License, permitting commercial use and integration with closed-source projects.

Limitations & Caveats

Linux support is listed as a future feature, indicating it is not currently available. The project is student-managed, which may impact the pace of development and long-term maintenance.