Dot is an open-source desktop application on GitHub that brings large language models, document chat and text-to-speech together in one fully local, private package. The project has gathered about 1,900 stars and is designed for people who want AI assistance without sending their files or conversations to the cloud.
Everything in Dot runs on your own machine. It loads local LLMs through llama.cpp, including compact models such as Phi-3, and lets you chat with your documents using retrieval-augmented generation: files are embedded locally with FAISS vector search, so answers are grounded in your own material. Speech features are local too, with text-to-speech output and Whisper-based transcription handled by whisper.cpp. Because the app is standalone and self-contained, it works offline and keeps sensitive documents, notes and recordings entirely under your control. Dot presents these capabilities in a single approachable app rather than a pile of scripts, which makes local AI accessible to non-developers as well.
Dot began as a personal project by developer Alex Pinel and grew through community interest in capable offline AI, showing that document chat, voice interaction and large language models can feel like one polished product even without any cloud backend.
Dot is free and open source. You can download it from the project website or build it from the GitHub repository at no cost; there are no subscriptions, API keys or usage charges, since all processing happens on your own hardware.
Dot is ideal for privacy-conscious professionals, researchers, writers and anyone handling sensitive documents who still wants LLM chat, document Q&A and voice features. It is also a great fit for offline environments where cloud AI services are unavailable or not allowed.