ShortsAI
Plan what to wear before the weather changes.
Live forecast timing, activity load, and comfort memory become a few safe clothing choices.
Calm signalLoading live weather
NowLive
+1hForecast
+3hForecast
Plan the activity, then choose your comfort level.
ShortsAI creates safe lighter, standard, and warmer options for the full weather window.
Built around the moments where outfit plans fail.
The recommendation is assembled from a small set of signals that matter before, during, and after the activity.
Forecast window
ShortsAI compares start, finish, and return-home conditions instead of using one static weather snapshot.
Temperature, feels-like, wind, rain, humidity, UVActivity load
Running, walking, and commute subtypes are treated differently, so body heat and outdoor exposure do not distort every recommendation.
Mode, intensity, commute type, exposure, return timeComfort memory
Post-activity feedback updates only the matching run, walk, or commute context.
Context offset, actual wear, changes, problem areasSafety policy
Cold, rain, wind, heat, and visibility are checked separately before any candidate is ranked.
Required items stay in every safe variantEvery model has a narrow job.
Safety stays rule-based. A sufficiently trained first-party ranker may order safe candidates, while language AI only classifies a follow-up.
Rules create safe choices first
The app creates lighter, standard, and warmer candidates, then applies required safety items before ranking.
AI classifies, rules recalculate
OpenRouter returns a structured intent only. ShortsAI handles any permitted adjustment and writes the explanation.
Fallback keeps it usable
If the API, model artifact, or language service fails, the same local safety rules still return a recommendation.