These items are just me playing with AI to build, improve and understand model capabilities.
A Chrome extension that flashes Claude's responses one chunk at a time (rapid serial visual presentation), so long completions scan in seconds instead of minutes. Start comfortable at 300 WPM and dial it up to 900 as your eye adjusts.
My first AI prototype, built in 2024. A chat interface that lets an LLM call real search functions (filter by price, location, beds, amenities) instead of guessing from free text, so a conversational query turns into an actual filtered listing search.