CampusWatch X: Autonomous Multi-Agent Campus Management System

Abstract

Campus services often span separate processes for room reservations, faculty consultations, incident reporting, and institutional information access. This paper presents CampusWatch X, an AI-assisted campus service framework that brings these tasks behind a common conversational interface. A Supervisor agent routes requests to specialist agents for resource booking, faculty appointment scheduling, incident management, and college-information assistance. The design combines language-model interpretation with API-mediated tools and persistent application records; the College Guide is intended to ground answers in content collected from the official institutional website. This paper describes the architecture, workflow safeguards, implementation approach, and an evaluation plan. Rather than claiming unmeasured improvements, it identifies testable measures for routing correctness, scheduling conflict prevention, incident confirmation, and answer grounding. The contribution is an integration-oriented design for campus workflows that retains explicit validation and authorization boundaries.

Introduction

CampusWatch X is an AI-powered campus assistance system designed to simplify administrative services in higher-education institutions. Traditional campus services often rely on separate forms, emails, and manual procedures, making it difficult for students and staff to access information and complete requests efficiently. The proposed system provides a unified conversational interface for resource and seminar-hall booking, faculty appointment scheduling, campus incident reporting, and college-information assistance.

The system uses a Supervisor agent and specialized agents to understand user requests and direct them to the appropriate service. The Supervisor routes requests to the booking, appointment, incident-reporting, or College Guide agent. These agents collect necessary information, interact with structured records or retrieved website content, and return relevant results.

The system follows a hybrid approach that combines language-model reasoning with deterministic application logic. AI helps interpret requests, ask follow-up questions, and suggest structured information, while backend code handles availability checks, data validation, authorization, and database updates. Confirmation steps are included before important actions to reduce errors and ensure that users understand what will happen.

The main features include:

• Resource and seminar-hall booking: Checks availability, detects scheduling conflicts, and manages booking requests and approval status.

• Faculty appointment scheduling: Collects appointment details and checks faculty availability before processing requests.

• Campus incident reporting: Classifies incident descriptions, suggests titles and priorities, identifies potentially similar reports, and requests user confirmation before submission.

• College Guide: Answers questions using information retrieved from the official JNNCE website, with the goal of providing reliable, institution-specific information.

The system uses a layered architecture consisting of a frontend, a Node.js and Express application API, a Python FastAPI agent service, specialized agents, and a MongoDB database. Google ADK and Gemini 3.1 Flash Lite support agent orchestration, while Firecrawl collects institutional website content for the College Guide.

Security and reliability are important aspects of the design. Protected operations should use verified user authentication and authorization, while booking and appointment workflows must validate availability using authoritative records. Incident information generated by AI should remain reviewable by the user. Logging, error handling, and request tracking can also help identify failures and maintain an audit trail.

The literature review highlights the importance of retrieval-augmented generation (RAG), scheduling constraints, incident-management workflows, and multi-agent systems. It also identifies challenges such as incorrect agent routing, outdated website information, conflicting booking requests, unsupported AI-generated answers, and insufficient authorization controls.

The evaluation approach focuses on observable results, including service connectivity, specialist workflow correctness, Supervisor routing, information grounding, and authorization. Initial development checks reportedly confirmed successful communication between services and passing specialist tests. However, comprehensive benchmark results, measured task-success rates, latency statistics, and a controlled evaluation dataset have not yet been established.

Conclusion

CampusWatch X presents a multi-agent approach to unifying resource booking, faculty appointment scheduling, incident reporting, and institutional information access. A Supervisor routes requests to specialist agents, while API services and structured persistence support operational workflows. The College Guide is designed to ground answers in official institutional web content. The design emphasizes that model-generated text must not replace deterministic validation, trusted identity, user confirmation, or auditable state changes. Future work includes validating college-managed authentication, enforcing role-based permissions across services, making booking and appointment writes concurrency-safe, refreshing website content with traceable metadata, and adding source-linked answers. A formal evaluation should measure routing accuracy, task completion, conflict prevention, grounded-answer rate, latency, and authorization correctness. Only results obtained from that evaluation should support quantitative claims.

References

[1] T. Guo et al., “Large Language Model Based Multi-agents: A Survey of Progress and Challenges,” Proc. 33rd Int. Joint Conf. Artificial Intelligence (IJCAI), 2024, pp. 8048–8057, doi: 10.24963/ijcai.2024/890. [2] Y. Wang et al., “A Survey on Large Language Model Based Autonomous Agents,” Frontiers of Computer Science, vol. 18, 2024, doi: 10.1007/s11704-024-40231-1. [3] P. Lewis et al., “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks,” Advances in Neural Information Processing Systems, vol. 33, pp. 9459–9474, 2020. [4] J. Swacha and M. Gracel, “Retrieval-Augmented Generation (RAG) Chatbots for Education: A Survey of Applications,” Applied Sciences, vol. 15, no. 8, art. 4234, 2025. [5] M. Chen, F. Werner, and M. Shokouhifar, “Mathematical Modeling and Exact Optimizing of University Course Scheduling Considering Preferences of Professors,” Axioms, vol. 12, no. 5, art. 498, 2023, doi: 10.3390/axioms12050498. [6] B. Anthony Jnr., G. P. Maureen, and M. A. Majid, “A Web Deployed Multi-Agent Based Approach for Student-Lecturer Appointment Scheduling in Institutions of Higher Learning,” Journal of Physics: Conference Series, vol. 1830, 012007, 2021, doi: 10.1088/1742-6596/1830/1/012007. [7] “Automation Tool for Grievances Using Artificial Intelligence,” Procedia Computer Science, vol. 283, pp. 2770–2779, 2026, doi: 10.1016/j.procs.2026.06.346.

Copyright

Copyright © 2026 Ashwini S P, Prajwal P J, Prajwal Veerappa Ujjagol, Pramodh N P, Prateek K. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.