I've created a project called Eclaire.

The motivation was simple: my personal data was growing exponentially—bookmarks, photos, documents, notes scattered everywhere. Eclaire brings it all together and lets AI manage it.

Once you feed your data into the system, it starts processing: fetching bookmark content, tagging, categorizing, analyzing images, and extracting text (including OCR). After that, the AI can use this material to search, answer questions, and even generate new content. You can also schedule recurring tasks for your AI.

By default, it uses Qwen2-14B via llama.cpp as the assistant backend and Gemma3-4B for multimodal processing. The models can be easily swapped out.


The project is open-source under the MIT license. The demo and code are linked below—feel free to try it out or contribute.

* Demo: https://eclaire.co/#demo

* Code: https://github.com/eclaire-labs/eclaire