Touch Grass AI

Touch Grass AI is an open-source garden planner that helps users spend less time researching gardening and more time actually growing plants. The application uses locally running, open-weight language models to provide planting recommendations, weekly gardening calendars, frost-date information, and answers to common gardening questions. By running inference on your own machine, you can use the core planner without needing an internet connection once the required model and data are downloaded.

Key features of Touch Grass AI include:

- Local frost-date calculations to plan around the growing season

- Plant recommendations based on your growing zone, season, and garden conditions

- A weekly planting calendar to help organize sowing, transplanting, harvesting, and avoiding tasks

- Offline AI gardening assistance for queries about planting times, pests, companion planting, and more

- Privacy-first design that requires manual location entry without GPS access

- Model flexibility to experiment with compatible GGUF models like Llama, Mistral, Phi, and Gemma

- Offline operation, allowing gardening information to be available even when connectivity is unavailable

The project, available on GitHub, is organized into modular Python components for easy experimentation with models, planting data, frost-date logic, and prompts. Developers are welcome to contribute, provide feedback, and suggest improvements.

To get started, install the dependencies using "pip install -r requirements.txt", download a compatible model with "python download_model.py --model phi-3-mini-4k-instruct-q4", and then run the garden planner with "python garden_planner.py --zip 90210". This initial setup requires downloading the model and any necessary external data. Once assets are available locally, the planner operates offline to the extent supported by its implementation.

The planner provides practical gardening advice, such as suggesting cover crops, compatible companion plants, and addressing frost-sensitive plants. Users can ask the AI any gardening-related questions, and it will use the available local dataset to provide explanations, comparisons, and next-step suggestions, though exact recommendations should be verified against local conditions.