This application is an on-line Proof Of Concept for Generative AI that comments live, during a web request, the data obtained from a weather API, during same web request (data stream analyser). The resulting web page contains AWS Gen AI recommendations for cloth and activities according to weather forecast for next 12 hours. Other use-cases for AI data stream analysis are comments and live game analysis in sports.
If page refreshed, Gen AI recommendation are expressed differently for same set of weather forecast. The AI prompt for Titan Text LLM is built dynamically with a Velocity template that includes transformed weather data for each of the next 12 hours. Recommendations can be according to user location. In the recommendation, Titan use literary styles like: formal, casual, funny, cosy, romantic.
More details about the fields from the page are presented in the table after the page.
| Field | Details |
|---|---|
| Weather Planner | Title: application plans clothing and activities according to weather forecast |
| Start time | Request time in time zone specified below |
| Location | The default location is Bucharest |
| Timezone | The default timezone is Bucharest |
| Style | Literary style randomly used by Titan Text LLM to give recommendations. Possible values: formal, casual, funny, cosy, romantic, meteorological, news, philosophical. |
| Vertical left section | Weather forecast from weather API for next 12 hours after Start Time |
| Vertical section row | Forecast hour and weather icon. On icon mouse-over, some weather details appear. Row colours shows day/night hours. |
| Central section | Amazon Titan Text Premier Generative AI Large Language Model response: styled title and recommendations |