Jaxon is a rapid prototyping platform that aims to accelerate the development of custom, fully-trained AI models. It provides productivity tools for data scientists to automate data labeling, simulate prototype models, explore design tradeoffs, generate synthetic training data, and correlate signals across documents. Jaxon is designed to improve model accuracy even with limited pre-labeled training data by filtering based on a confidence and quantity tradeoff. Its unique features include AI-assisted labeling, specialized AutoML for sparse data models, an interactive design studio, generative AI for synthetic data augmentation, and extracting knowledge graphs from various documents. Jaxon is suitable for data science teams seeking to iterate quickly on AI model development, optimize training data, and uncover actionable insights from unstructured data.