*This is a submission for Weekend Challenge: Earth Day Edition Agentic Acres is an intelligent, multimodal permaculture design assistant built to accelerate regenerative agriculture and make ecological landscaping accessible to everyone. My primary goal was to take the overwhelming complexity of designing sustainable ecosystems, knowing which plants support each other, fix nitrogen, or accumulate nutrients, and boil it down to a single photograph. Users simply snap a picture of their yard or planting site (or upload one from their gallery) and provide their location. The application instantly assesses the site and engineers a complete, climate-specific Plant Guild. Instead of endless research, users are presented with a gorgeous, dynamic dashboard that outlines their primary canopy tree, nitrogen fixers, dynamic accumulators, and ground covers. All are precisely tailored to their local environment and the exact physical constraints shown in the photo. Try the web app: https://agentic-acres.web.app See a demo video on YouTube: https://github.com/RealWorldApplications/agentic-acres I built Agentic Acres as a fully responsive Web Application using Flutter, and deployed it using Firebase Hosting. I built the entire app in Google Antigravity and used Gemini 3.1 as my AI assistant. The core intelligence of the application is driven by *gemini-3.1-flash-lite-preview using the google_generative_ai SDK. I used Gemini's multimodal vision capabilities to achieve something unique: the app passes the user-uploaded image alongside their geographic coordinates directly into the model. I implemented strict prompt engineering so that Gemini would output a highly constrained JSON schema rather than conversational text. This allows the frontend to confidently parse the AI's ecological assessment and dynamically construct the localized "Bento Box" UI. To make the app feel alive and premium, I incorporated several other key technologies: Geolocation & Geocoding: I integrated the geolocator package to allow users to pull their exact GPS coordinates with a single click. Dynamic Media Pipelines: Because Gemini returns common plant names rather than hardcoded URLs, I integrated the Wikipedia (Wikimedia) Action API. As the dashboard renders, the app asynchronously executes fuzzy searches against Wikipedia's backend, ripping the main 400px thumbnail of the matched plant and fading it into the glassmorphic UI cards in real-time. Modern Glassmorphism Design: I styled the frontend using Flutter widget compositions to build a dark-mode, frosted-glass interface that feels modern, premium, and instantly trustworthy. Best Use of Google Gemini