StreamHub - Multi - Agent Framework for Movie Recommendation System Using Langchain

Abstract

The rapid growth of digital entertainment platforms has made it difficult for users to discover suitable movies and music from a large collection of content. Conventional platforms commonly depend on separate search interfaces and recommendation features, which may not provide a unified or context-aware experience. This paper presents StreamHub, an OTT and Music Multi-Agent Recommendation System that combines movie discovery, music discovery, personalized recommendations, user interaction, and premium-content access in a single web application. A Supervisor Agent coordinates user requests and directs them to specialized agent functions. The Movie/Content Agent integrates the TMDB API to retrieve available movie and television information, while the Music Agent connects to an external royalty-free music API to obtain available tracks and related metadata. User preferences, watch history, ratings, and watchlists can be maintained through a MySQL database to support more relevant content suggestions. The system uses React.js for the frontend and FastAPI, SQLAlchemy, and Pydantic for backend services. JWT-based authentication supports protected access, while the modular multi-agent design separates responsibilities and supports future extension. The platform demonstrates how Al-assisted interaction, external APIs, persistent user information, and coordinated agents can simplify entertainment discovery. The system is intended to improve usability and personalization by bringing movie and music exploration together through one interface.

Introduction

The proposed StreamHub system is an AI-powered, web-based entertainment platform designed to simplify movie and music discovery through a unified interface. It addresses the limitations of traditional entertainment applications, where users often need to switch between multiple platforms and manually search for content that matches their interests.

StreamHub uses a multi-agent architecture in which a Supervisor Agent interprets user requests and directs them to specialized functions. The Movie/Content Agent retrieves movie and television information through the TMDB API, while the Music Agent obtains available tracks from an external royalty-free music API. Profile and recommendation functions use user preferences, watch history, ratings, and watchlists to support personalized suggestions.

The platform is built using a React frontend, a FastAPI backend, and a MySQL database for persistent data storage. JWT-based authentication protects supported user accounts and restricted application features. Separate modules manage content discovery, recommendations, notifications, and payment-related workflows for premium access.

Conclusion

The StreamHub — OTT and Music Multi-Agent Recommendation System demonstrates how multi-agent coordination and modern web technologies can be used to bring movie and music discovery into a unified platform. The system provides a common interface for users to explore entertainment content and access supported recommendation and activity-management features. The Supervisor Agent coordinates requests and directs them to the appropriate specialized functions. The Movie/Content Agent retrieves available movie information through TMDB, while the Music Agent retrieves available tracks through the configured external music API. User preferences, watchlists, ratings, and activity records can be maintained through MySQL to support continuity and personalization. The application combines React.js, FastAPI, SQLAlchemy, Pydantic, MySQL, and JWT-based authentication. The separation between frontend, backend, agent coordination, external APIs, and database operations creates a modular structure that can be maintained and extended. The actual content displayed depends on the external services and features configured in the application. Overall, StreamHub demonstrates the potential of a multi-agent architecture for simplifying entertainment discovery and coordinating different services in one web application.

References

[1] Jianghao Lin et al., \"How Can Recommender Systems Benefit from Large Language Models: A Survey,\" arXiv, 2023. [2] Qi Wang et al., \"Towards Next-Generation LLM-based Recommender Systems: A Survey and Beyond,\" 2024. [3] H. Li et al., \"Recommender Systems in the Era of Large Language Models (LLMs),\" IEEE Transactions on Knowledge and Data Engineering, vol. 36, no. I l, 2024. [4] \"CD-LLMCARS: Cross Domain Fine-Tuned Large Language Model for Context-Aware Recommender Systems,\" IEEE Open Journal of the Computer Society, vol. 6, 2025. [5] \"Machine Learning-Based Music Recommendation System Based on User Interest,\" 2024 2nd World Conference on Communication & Computing (WCONF). [6] Q. Peng et al., \"A Survey on LLM-powered Agents for Recommender Systems,\" Findings of EMNLP 2025, pp. 11574— 11583, 2025. doi: 10.18653/v1/2025.findings-emn1p.620. [7] L. Wu et al., \"A Survey on Large Language Models for Recommendation,\" World Wide Web, vol. 27, article 60, 2024. doi: [8] 10.1007/s11280-024-01291-2. [9] S. Jayalakshmi, N. Ganesh, and R. Cep, \"Movie Recommender Systems: Concepts, Methods, Challenges, and Future Directions,\" Sensors, vol. 22, no. 13, article 4904, 2022. [10] A. Yenkikar et al., \"Machine Learning-Based Music Recommendation System Based on User Interest,\" Proceedings of WCONF, 2024. doi: 10.1109/WCONF61366.2024.10692020. [11] F. M. Harper and J. A. Konstan, \"The MovieLens Datasets: History and Context,\" ACM Transactions on Interactive Intelligent Systems, vol. 5, no. 4, article 19, 2015. doi: 10.1145/2827872. [12] [l l] S. Rendle, \"Factorization Machines,\" Proceedings of the 2010 IEEE International Conference on Data Mining, pp. 995 1000, 2010. doi: 10.1109/1CDM.2010.127.

Copyright

Copyright © 2026 Vanitha Mani K, Vidhyavarshini K. K. H., Vaishnavi N. G.. 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.