Rajay McPherson

Fragrance Coach is a web app that picks a fragrance for the day from a real seasonal, occasion-based wardrobe with layerings. Picking is hard because the right pick depends on season, occasion, mood, and live weather, and most pickers give one generic name and ignore layering. Fragrance Coach combines season, occasion, mood, and live weather into one pick with a reason, layering with spray counts, and Notes (top / heart / base). Built during this hackathon, on a pre-existing Oct 5 base of 6 fragrance cards, occasion/mood inputs, and a rules-based pick: AI coach with Meta Muse Spark (muse-spark-1.3) via the Meta Model API, standalone Ask Muse Spark with weather and last 3 turns of history, geolocation weather via Open-Meteo, LocalStorage favorites, dark mode, GitHub Pages deployment, and Sillage/Longevity bars with season numbers on every card. Nov 4 additions: prompt polish to a general coach, Ask removed from Get my pick, structured JSON outputs (reasoning, mainPick, layering, notes) parsed and rendered as sections, Ask history, and readability fixes. Muse Image is not used — the text AI alone is the Meta API proof, shown locally in the demo video because Pages shows "AI key not set.