Wilde Prototype

source

The prototype of my second brain system, which I’m calling Wilde is now ready! Everything is open source and available on GitHub. This tool lets you create a 3D visualization of your physical bookshelf using embeddings. It helps with every stage of the process, including manual data entry, scanning in books from images, gathering metadata from the Google Books API, and finally, generating embeddings. Because it makes heavy use of multi-modal LLMs under the hood, it’s able to handle a great deal of ambiguity in the data. The result of all this is an interactive 3D point cloud visualization of the semantic geography of your bookshelf. You can pan, zoom, and orbit through the various clusters of books that form. Everything is automatically grouped by semantic similarity with its neighbors, resulting in surprisingly subtle arrangements.

At the moment, it’s more of an interactive-digital-bookshelf-explorer than a fully-fledged second brain tool, hence “prototype”. In the future I plan to elaborate on the book exploration side of things, with support for lists, advanced filtering, and recommendations. But more importantly, I’d like to expand the system to handle many other kinds of physical artifacts like sticky-notes and todo-lists. The idea is to collect all the little markers of our thoughts, plans, and dreams that accumulate throughout the day in one place and hopefully help us make sense of them.