E.C. Zhu Fainman
Texas State University,
United States
Keywords: large language models, AI for education, institution-specific AI
Summary:
PATHOS AI (Personalized Academic Toolkit for Higher-Order Success) is a secure, institution-specific AI advising platform that provides students with intelligent, personalized academic guidance by leveraging proprietary university data. Unlike generic AI chatbots or static degree planning tools, PATHOS AI operates entirely within the university’s infrastructure, ensuring FERPA compliance and robust cybersecurity. Built on fine-tuned large language models (LLMs) and real-time decision logic, PATHOS AI integrates multiple internal data sources—including degree roadmaps, historical syllabi, grade distributions by instructor and section, instructor evaluations, enrollment trends, and upcoming course schedules. This rich dataset enables the platform to dynamically generate tailored academic roadmaps and course recommendations, while also supporting nuanced student queries such as: • “Can I graduate on time if I switch majors?” • “Which course sections have the highest success rates?” • “What electives align best with my career goals and schedule?” Students interact with PATHOS AI through a conversational, ChatGPT-style interface designed to simulate the responsiveness of a human advisor while delivering institutionally grounded information. When questions fall outside the AI’s scope—due to complexity or ambiguity—the system generates a comprehensive summary of the student’s background, queries, and attempted solutions, and routes the case to a human advisor for follow-up. This triage feature enhances advising capacity without replacing human expertise. The platform was developed through Texas State University’s FIAP and I-Corps programs and has received strong validation from students, advisors, and faculty. Interviews revealed high demand for features such as instructor-specific grade distributions, transparency around course assessments, and early visibility into schedule conflicts. Advisors also expressed support for PATHOS AI’s ability to reduce repetitive queries and improve student preparedness for advising appointments. Although PATHOS AI’s first deployment target is higher education, its architecture is extensible to any institution that needs to provide secure, data-driven, conversational support based on internal, proprietary information. Potential applications include healthcare systems (patient triage), government services (benefits navigation), and financial services (personalized client guidance). By unifying conversational AI with internal academic data in a privacy-compliant environment, PATHOS AI represents a significant step forward in student support technology and opens new commercial opportunities in institutional AI applications beyond education.