Project Details
[Return to Previous Page]Penn State AI Advisor: Phase II
Company: PSU Department of Computer Science and Engineering
Major(s):
Primary: CMPSC
Non-Disclosure Agreement: NO
Intellectual Property: NO
This project continues the Penn State EECS AI Advisor piloted in Spring 2026 (you can read about it here: https://sites.psu.edu/lfshowcasesp26/2026/04/29/eecs-ai-advisor/) and aligns with Penn State’s strategic priorities in student success, digital transformation, and responsible AI integration. Its design continues to be guided by feedback from Computer Science and Engineering academic advisers, pre-major advisers, and the study abroad office. Background: What Phase I Delivered (Spring 2026) In Spring 2026, a senior capstone team built and validated the EECS AI Advisor (V1), a document-grounded advising assistant configured in Microsoft Copilot Studio on top of a Penn State SharePoint knowledge base of advisor-approved PowerPoint modules. The agent reliably answers common questions across EECS, pre-major, and study abroad advising, including course substitution, grade forgiveness, AI violations, course repeats, major and minor pairings, and late adds. Guardrails restrict responses to approved content and redirect students to a human adviser when a question falls outside the knowledge base. Advisers can update the SharePoint modules directly, with changes automatically reflected in the agent, and by final testing the agent answered scenario-based questions consistently with minimal hallucinations. The system is currently in an adviser review cycle prior to student release. Main Objective To move the AI Advisor from a validated prototype to a deployed, measurable student service: complete adviser review, release the agent to students through Microsoft Teams, expand its knowledge coverage, and build the evaluation and analytics layer needed to demonstrate impact on adviser workload and student outcomes. Specific Aims for Fall 2026 • Deployment: Close out the adviser review cycle, resolve remaining inaccuracies, and publish the agent to Microsoft Teams and other Penn State Microsoft 365 channels students already use, with clear onboarding materials. • Knowledge expansion: Extend the advisor-validated knowledge base beyond EECS, pre-major, and study abroad to additional engineering departments, following the same stakeholder review workflow established in Phase I. • Evaluation and analytics: Instrument the agent with usage tracking, accuracy auditing, and student and adviser feedback collection to measure deflected routine inquiries, response quality, and after-hours usage. • Responsible enhancement: Prototype multilingual support and explore FERPA-aligned personalization using enrollment data, with human-in-the-loop oversight and governance review before any student data integration. What Students on This Project Will Do and Learn The Fall 2026 capstone team will work directly with academic advisers and the faculty sponsor in weekly working sessions, gaining hands-on experience in AI agent configuration and grounding (Microsoft Copilot Studio), knowledge engineering and content governance (SharePoint), prompt and guardrail design, deployment within an enterprise Microsoft 365 environment, and evaluation of a live AI system with real users. This is a rare opportunity to ship an AI product to thousands of fellow students while learning how responsible, policy-compliant AI is actually built inside a large institution.
