This paper introduces Jill Watson, a conversational Virtual Teaching Assistant (VTA) powered by ChatGPT. Jill Watson is designed to answer student queries based on course materials such as slides, notes, and syllabi. Unlike previous versions, Jill Watson uses a skill-based architecture inspired by XiaoIce, allowing for modular design and integration of new APIs. It is also well-suited for intelligent textbooks as it can process and converse using multiple large documents. The system is built using publicly available resources to ensure reproducibility and extensibility. Comparative analysis shows that Jill Watson outperforms the legacy Jill Watson and the OpenAI Assistants service in terms of response quality and safety. The system includes safety measures to reduce hallucinations and toxicity. Jill Watson is also designed to answer questions based on multiple large documents, making it suitable for intelligent textbooks. The paper presents the architecture of Jill Watson, including coreference resolution, skill classification, and contextual answering. It also discusses the results of experiments comparing Jill Watson with other systems, showing its effectiveness in answering student queries and minimizing harmful responses. The paper concludes that Jill Watson is a reliable and safe virtual teaching assistant that can enhance education by providing accurate and helpful responses.This paper introduces Jill Watson, a conversational Virtual Teaching Assistant (VTA) powered by ChatGPT. Jill Watson is designed to answer student queries based on course materials such as slides, notes, and syllabi. Unlike previous versions, Jill Watson uses a skill-based architecture inspired by XiaoIce, allowing for modular design and integration of new APIs. It is also well-suited for intelligent textbooks as it can process and converse using multiple large documents. The system is built using publicly available resources to ensure reproducibility and extensibility. Comparative analysis shows that Jill Watson outperforms the legacy Jill Watson and the OpenAI Assistants service in terms of response quality and safety. The system includes safety measures to reduce hallucinations and toxicity. Jill Watson is also designed to answer questions based on multiple large documents, making it suitable for intelligent textbooks. The paper presents the architecture of Jill Watson, including coreference resolution, skill classification, and contextual answering. It also discusses the results of experiments comparing Jill Watson with other systems, showing its effectiveness in answering student queries and minimizing harmful responses. The paper concludes that Jill Watson is a reliable and safe virtual teaching assistant that can enhance education by providing accurate and helpful responses.