Spartan One-Stop Bot 

Spartan One-Stop, Enrollment Services, and MSU IT Enterprise Services

Primary contact: Lucy Chen
 

The Spartan One-Stop Bot is a conversational AI assistant designed to provide students and parents with accurate, consistent, and timely answers to common questions related to student services at Michigan State University. The bot supports topics such as enrollment, financial aid, billing, registration, student records, and general One-Stop processes.

The bot serves as a first-line, self-service support channel available 24/7, directing users to authoritative MSU information sources. The bot is currently deployed at onestop.msu.edu and will be continuously improved through monitored usage patterns, feedback, and governance oversight.

Students and parents frequently need timely, accurate answers to high volume, repetitive questions outside of normal business hours. Existing support channels (email, phone, and in-person services) can experience peak demand during key academic periods, leading to delays, inconsistent responses, and increased staff workload.

This use case addresses the opportunity to:

• Improve access to student services information 24/7
• Reduce response time for common inquiries
• Improve consistency and accuracy of messaging
• Allow staff to focus on complex or sensitive cases rather than repetitive questions

This matters because timely and clear information directly impacts student experience, enrollment confidence, and institutional trust. 

• Students and families/caregivers

The Spartan One-Stop Bot is implemented using a retrieval augmented generation (RAG) architecture built primarily on Google Cloud Platform (GCP) services.

Curated public MSU websites and internally approved frequently asked question content from One-Stop are first collected and reviewed by stakeholders to ensure accuracy and policy compliance. These approved knowledge sources are ingested into Vertex AI Search and Conversation, where the content is embedded and stored in a managed vector database to support semantic search and retrieval.

On top of this retrieval layer, a RAG-based AI agent is developed using Vertex AI Search and Conversation. The agent is designed with carefully structured conversation flows, system prompts, and guardrails to:

• Ground responses in retrieved MSU-approved content
• Limit responses to defined One-Stop topics
• Avoid hallucinations or speculative answers
• Direct users to official MSU resources or human support when appropriate

Before production deployment, the system undergoes a series of structured testing and validation cycles with One-Stop and partner stakeholders. These tests are used to verify answer quality, identify gaps, and correct outdated or incomplete knowledge sources before public release.

Once validated, the AI agent is deployed as a conversational chatbot and integrated into the MSU One-Stop website via Sitecore as an embedded chatbot widget. The system is continuously monitored post-deployment, and the knowledge base and conversational logic are iteratively updated as policies, deadlines, or institutional information change. 

Success will be measured using both qualitative and quantitative metrics, including:

• Reduced average response time for common questions
• Increased self-service resolution rate (questions answered without staff intervention)
• Usage and adoption metrics (unique users, sessions, repeat usage)
• User satisfaction and feedback trends
• Improved consistency of answers compared to manual, multi-channel support

These metrics help ensure the bot delivers meaningful value without compromising accuracy or trust.  

• Incorrect or outdated information: Mitigated through curated data sources, regular content reviews, and clear disclaimers
• Bias or misleading responses: Addressed through controlled prompts, content review, and scope limitations
• Privacy concerns: The bot is designed to avoid collecting or storing sensitive personal data and does not provide personalized or record specific responses
• Over-reliance on automation: Users are clearly guided to human support when issues require staff involvement 

It’s accessible to the public at onestop.msu.edu.