Eclaire

Eclaire is a self-hosted AI assistant for people who want to ask questions about their saved information and act on it in one place. It brings documents, notes, photos, bookmarks and tasks into a shared library.

• Answers with sources: Ask questions or request summaries across your content, then inspect the material behind an answer.
• Document and image analysis: Extract text with OCR, search document contents, and use AI to classify and tag documents and photos.
• Notes and saved pages: Create linked notes in rich text or Markdown, and save web pages as readable copies or PDFs for offline reading.
• Tasks and automations: Track due dates and recurring tasks, ask the assistant to create notes or resolve tasks, and schedule requests such as a weekly task summary.
• Connected access: Reach your instance through the web interface, an installable PWA or Telegram, and connect Apple Shortcuts, Android automation tools or custom applications through the REST API.

Models and data storage run locally by default. Eclaire supports llama.cpp, Ollama, LM Studio, vLLM and MLX through OpenAI-compatible APIs, with text and vision models including Qwen, Gemma, DeepSeek and Kimi. Connecting a cloud inference API sends model requests to that provider.

Eclaire runs through Docker on macOS, Linux and Windows. The website specifies Apple Silicon Macs or Linux/Windows machines with GPUs; memory needs depend on the selected models. It uses the MIT license and is open source. There isn't a hosted edition, though managed deployment services are available for organizations.

The project isn't hardened for direct public internet exposure. Remote access needs an additional security layer, such as a private network or a proxy with authentication.