The 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, without requiring extensive training datasets. The system uses a modular design inspired by Xiaoice, allowing for easy integration of new APIs. Key features include coreference resolution, skill classification, dense passage retrieval (DPR), and safety measures to prevent hallucinations and toxicity. Comparative analysis shows that Jill Watson outperforms legacy Jill Watson and the OpenAI Assistants service in terms of response quality and safety. The paper also includes real-world examples from a classroom setting, demonstrating Jill Watson's effectiveness in handling various student queries and maintaining educational integrity.The 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, without requiring extensive training datasets. The system uses a modular design inspired by Xiaoice, allowing for easy integration of new APIs. Key features include coreference resolution, skill classification, dense passage retrieval (DPR), and safety measures to prevent hallucinations and toxicity. Comparative analysis shows that Jill Watson outperforms legacy Jill Watson and the OpenAI Assistants service in terms of response quality and safety. The paper also includes real-world examples from a classroom setting, demonstrating Jill Watson's effectiveness in handling various student queries and maintaining educational integrity.