A case study in applying AI and GIS to sustainable small-scale farming practices.  This study focuses on the use of satellite imagery and machine learning to detect agroforestry practices in the complex Alas Mertajati region in Bali. Historically, small-scale food production hasn't been a priority for AI-supported analysis of satellite imagery due to limited image resolution and the challenge of articulating the needs of small-scale farmers. However, this study demonstrates the potential of applying satellite assets and machine learning to identify agroforestry, a common small-scale farming practice in Southeast Asia.  Agroforestry involves compact spatial units with various tree and plant species. These small plots are manually tended and provide a continuous source of food. They also help reduce landslides, making them resilient to climate change. However, detecting agroforestry in satellite imagery with statistical approaches is challenging due to plot size and plant diversity.  The study uses the latest Planet Labs satellite imagery, offering spectral information to detect agroforestry practices in Alas Mertajati. Machine learning algorithms were employed to create classifiers, producing the first-ever maps of agroforestry in Bali. Local communities provided valuable ground truth data, improving classification accuracy and map readability. Additionally, the study highlights the COCKTAIL software repository, simplifying GIS land cover classification and data management, especially in resource-constrained environments. This research not only advances agroforestry detection but also emphasizes the significance of ground truth data and effective science communication in remote sensing projects.