VERY CLEAR WEATHERGRAPHICAL USER INTERFACES

VERY CLEAR WEATHER
Conversational Weather.

Most weather apps are dashboards, humidity percentages, UV indices, precipitation probabilities, wind speeds. A lot of numbers that answer questions nobody actually asked. The information is all there, but it takes effort to extract the one thing you actually needed to know before walking out the door.

Course: Graphical User Interfaces
Mentors: Benjamin Zweig, Julian Francis Kelly
Timeline: ~ 1 week




What if the app just told you what you needed to know?
Very Clear Weather skips the dashboard. Instead of presenting data to interpret, it surfaces two questions based on current conditions and answers them. If rain is likely, it asks "Do I need an umbrella?" and answers - yes, and a jacket. No charts. Numbers only when they matter. And if you need more detail, you can ask for it.



The app uses AI to answer any follow-up weather question in natural language, the same way you'd ask a person.



This concept was prototyped in Figma & Origami Studio, exploring transitions, the loading state while the AI generates a response, and how new Q&A pairs stack into the conversation. 


I built a working version on Android Studio with the help of Claude & Figma MCP. Live weather data, real AI responses, functional location switching, making the concept fully testable on-device. This is a constantly developing project - something I keep working and experimenting on. Next steps would be giving it additional context from my calendar - so it can give me smarter responses and also add my commutes in it’s context while generating responses.
OpenAI GPT4o Mini 
System Prompt: You are the AI behind "Very Clear Weather", a conversational weather app. Answer weather questions in a direct, warm, conversational tone. 1-3 short sentences. No data dumps or charts. Numbers only when they help. {weatherCtx} Today is {date}. If the question isn't weather-related, gently redirect.


OTHER PROJECTS
UPDATED APR,2026