Trustworthiness by Design: Structuring Trustworthy RAG-Augmented LLM Assistants via Enterprise Architecture

The integration of Explainable Artificial Intelligence (XAI) into proactive chatbot systems powered by Retrieval Augmented Generation (RAG) (a enhances languagemodels by retrieving relevant external documents to generate more informed and accurate responses) has the potential to transform educational environments by improving transparency, trustworthiness, and usability. This study reviews the architectural components, design frameworks, and evaluation methods required to embed trustworthiness into chatbot systems. Through a systematic exploration of XAI frameworks and mechanisms like Retrieval-Augmented Generation Assessment Suite (RAGAS) (evaluates RAG pipelines by scoring their outputs based on relevance, faithfulness, and answer quality) and Retrieval-Augmented Generation Evaluation (RAGE) (a scoring system that evaluates trustworthiness in RAG systems by measuring input and output faithfulness, and source attribution), the following research identifies key patterns and trade-offs involved in designing scalable, compliant, and user-personalized solutions RAG-powered Large LanguageModel (LLM) assistants. In addition, the study tests different Enterprise Architecture (EA) frameworks and system architectures to assess their influence on the performance and integration of Trustworthy Artificial Intelligence (TAI) in chatbot systems. Thus, this study’s main research questions focus on analyzing the impacts of different EA approaches on the integration of TAI, scoped to RAG and LLMs, in Artificial Intelligence (AI)-powered assistants. It explores how TAI can enhance the interpretability of RAG-based outputs, identifying the most effective methods for evaluating chatbot transparency, and examining the role of TAI in fostering trustworthiness and adoptionwithin educational systems. Additionally,the following article investigates the challenges of aligning XAI models with Information Technology (IT) management frameworks and governance standards, with a particular focus on compliance.