Design of an LLM-Powered AI Assistant Chatbot for Nonprofit Trust Management

Authors- S. Ian Steve Waugh, R. Priya Professor

Abstract--This study presents the design and development of a human-centred AI assistant chatbot powered by Large Language Models (LLMs), tailored for nonprofit trust management systems. Method: Leveraging Lang Chain and OpenAI ’s GPT-4 API, the chatbot system integrates with a Fast API backend, React Native frontend, and MongoDB database. It is structured to deliver modularity, real-time interaction, and data-driven responses in a transparent and scalable framework. Human-centred principles were prioritized during design, inspired by Shneiderman ’s vision of application focused AI. Results: The chatbot demonstrated a 92% query resolution accuracy in test environments, with an average backend response time of under 150 milliseconds. Feedback from usability testing confirmed ease of navigation and improved donor engagement. Conclusion: The proposed system confirms the effectiveness of LLM-driven AI assistants in nonprofit platforms and showcases how modern AI frameworks like Lang Chain and OpenAI ’s GPT-4 can simplify donation workflows, enhance trust, and improve transparency. Impact: This chatbot model offers a blueprint for NGOs and nonprofit organizations to integrate intelligent support systems that reduce manual workload, increase donor trust, and scale communication outreach.

DOI: /10.61463/ijset.vol.13.issue3.152