Comtus AI is a study workspace built around the material a person brings to it. Sources, conversation, and practice sit together in a notebook, so the path from reading a document to testing what you remember stays visible.
I designed and built the React and TypeScript application, its document tools, and the surrounding product website. The interface borrows from a working study desk: paper, ink, a brass accent, generous reading type, and clear divisions between work areas. The visual language gives long reading sessions room to breathe while keeping the next useful action close.
From material to practice
A notebook can hold source documents, notes, and saved study work. Users choose which sources to include in a conversation, ask questions, and generate flashcards, quizzes, summaries, or guides. Source search preserves those inclusion choices. Saved work stays easy to find and revisit.
Mathematical notation carries through questions, choices, worked answers, and review. Supported coordinate graphs include exact data tables. Export tools preserve rendered notation so a study artifact remains useful outside the browser.
The workspace includes an optional notebook companion, Compad. Its compact chat shares the same conversation and draft as the full pane. Moving between the two views should feel like changing the size of a tool, rather than starting the task again.
A desk that adapts
Sources and study tools use resizable, keyboard-operable panes. Narrow screens keep the reader and composer usable. Theme, reading size, reduced motion, dialog focus, and recovery after deleting saved work are part of the same product experience.
Comtus AI is available on the web. A Tauri desktop client is also part of the codebase. The latest local design work refines the notebook catalogue, source reader, and practice hierarchy; that work has its own release step. The public product continues to center one simple progression: gather material, connect ideas, then recall them.