I built GeniusPlants, an AI-powered gardening assistant that uses LLMs to solve practical gardening challenges. The app provides personalized plant recommendations based on your garden conditions, generates optimal garden layouts, and creates customized care plans.

Technical stack:

- Next.js (App Router) with TypeScript and Tailwind CSS - OpenAI GPT-4o-mini for plant recommendations and care plans - Prisma ORM with PostgreSQL for data management - Zustand for state management - Auth.js for authentication

Instead of maintaining a traditional plant database, I implemented a pure AI approach with structured prompts and response parsing. The system includes a block-based garden space visualization that calculates optimal plant placement and uses a natural language interface for adding plants (e.g., "I want to grow 3 tomatoes, some basil, and carrots").

The app handles interesting problems like:

- Garden space management using a block-based system (1 block = 1 sq ft) - AI-driven plant positioning within garden layouts - Intelligent care plan generation that adapts to automated watering systems - Plant age tracking for growth-stage-appropriate care instructions

I'd love feedback from the HN community, especially on the AI implementations, UX design choices, and handling garden space constraints.