A Multi-Agent Engineering Team for Prompt-Based Application Generation
A Multi-Agent Engineering Team for Prompt-Based Application Generation
Prof. Ashvini Bamanikar
Assistant Professor
PDEA’s College of Engineering Manjari(Bk), Pune, India
Amruta Pawar
Department of Computer Engineering PDEA’s College of Engineering Manjari(Bk), Pune, India
Soham Mulik
Department of Computer Engineering PDEA’s College of Engineering Manjari(Bk), Pune, India
Gayatri Patil
Department of Computer Engineering PDEA’s College of Engineering Manjari(Bk), Pune, India gayatripa34@gmail.com
Shubham Kamble
Department of Computer Engineering PDEA’s College of Engineering Manjari(Bk), Pune, India shubhamkamble2504@gmail.com
Abstract—The emergence of Agentic Artificial Intelligence (AI) represents a paradigm shift from traditional reactive AI models to autonomous, goal-driven systems capable of percep- tion, reasoning, and self-improvement. This paper presents a novel Multi-Model Collaboration Framework for Agentic AI, designed to overcome the limitations of monolithic Large Lan- guage Models (LLMs) such as hallucination, lack of long-term memory, and limited reasoning depth. The proposed system orchestrates multiple specialized AI agents—each equipped with domain-specific expertise—through an API-driven architecture that enables verifiable, transparent, and scalable collaboration. By introducing task decomposition, inter-agent validation, and a persistent memory layer, our framework enhances reliability and adaptability across multiple domains, including financial fraud detection, healthcare diagnostics, and dynamic content generation. Experimental evaluations demonstrate improved ac-curacy, efficiency, and robustness, establishing this system as a foundational blueprint for the next generation of autonomous, collaborative AI systems.
Index Terms—Agentic AI, Multi-Agent Systems, LLM Collab-oration, Orchestration Framework, Autonomous Intelligence, AI Reliability, Multi-Model Architecture