Deep learning has achieved widespread success, yet its broader adoption is limited by two key challenges: the need for large labeled datasets and the lack of model transparency. These issues are especially relevant for small and medium-sized enterprises, which often face a lack of annotated data, limited adaptability of existing solutions, and insufficient expertise in artificial intelligence (AI). This work-in-progress paper presents TrustAI, an open-source platform for interactively trainable and adaptable machine learning (ML) models, designed to address these limitations. By combining explainable AI and interactive ML, the platform enables users to iteratively train ML models by providing feedback on model predictions and model explanations. User feedback is integrated into model training to improve transparency, detect bias, and align model behavior with human reasoning. We outline the platform’s design, architecture, and ethical considerations. The TrustAI platform offers a transparent and human-centered alternative to traditional ML systems.