WDYM
"WDYM" - Agentic memory plus augmentation that lets AI understand what you really want.
Core trick
Stores preferences extracted from past chats and uses them to rewrite each new prompt before it reaches the model.
Description
"WDYM" - Agentic memory plus augmentation that lets AI understand what you really want. WDYM remembers your personalized preferences based on previous chats, and augments your prompts to say "what you really mean" - so you can get a personalized experience with faster and better results across every AI Interaction.
README excerpt
# hack092725 - agent rules but for people - Coding agents like Claude and Codex have guidance files, e.g. `Claude.MD` - We need the same for humans so we don’t ask stupid questions and actually learn from using AI-assisted tasks like coding - Prototype: Learnings for both Agents and Humans