Tendril: A Garden Assistant You Barely Have to Open

Developer Archana has introduced Tendril, an open-source garden assistant designed to minimize screen time while managing plant care. Built for the Hacktoberfest Open-Source AI Challenge, Tendril uses a local Qwen 2.5 (3B) model via Ollama and faster-whisper for voice processing, ensuring all data remains private on the user's machine. The application integrates weather data from Open-Meteo to generate concise, actionable daily tasks, which are rendered as simple images optimized for phone lock screens. By prioritizing local execution and minimal interaction, the project avoids the pitfalls of traditional habit-tracking apps. The stack includes Python, FastAPI, SQLite, and Docker, with Pydantic ensuring the AI output remains structured and reliable. Tendril demonstrates how small, open-weight models can effectively automate household chores without requiring cloud dependencies or constant user engagement, offering a practical, privacy-focused solution for hobbyist gardeners.