POC 2023litus is a compact, thread-style notetaking application at the edge of your collection.
Composition
litus provides an alternate approach to logging notes and data, acknowledging the fragmented nature of modern media consumption.
litus encourages Cornell-style notetaking with user thoughts, notes, & prompts compared against an LLM response, Powered by OpenAI.
Thread Search
litus provides a chat and search interface to your personal collection. litus is not suggested for use on personal or proprietary data.
Content Generation
Easily generate meaningful responses from your notes, or extract key features from longer text inputs.
Guide
litus incorporates icons from the Carbon Design System.
Retrospective (2025)
- Short-form notes retrieved with related notetaking session context, based on the user's prompt, to build LLM query context within the <20k token windows of GPT-3.5 Turbo
- Prompted early models for structured outputs (JSON) with client-side parsing to populate UI components
- Design experiment steering users away from non-conversational interactions with LLMs in an edtech context