Requirement Gathering Using Agentic AI

The process of creating a Software Requirements Specification (SRS) can be very laborious, repetitive and contain a great deal of human error during the development of software. This initiative introduces the Auto SRS generator, which is an agent based AI assisted system, enabling the automatic creation of the SRS documents from start to finish. The agent-based solution based upon intelligent agents that not only understand written descriptions of projects, but also interact with structured prompting, have the ability to analyse user input and produce accurate and extensive requirement specifications according to standard IEEE SRS guidelines. The system accepts user provided basic project information and produces a complete SRS Document by expanding it into numerous well-organized sections of Functional Requirements, Non-Functional Requirements, System Models, Use Case Descriptions and Constraints. The Auto SRS Generator reduces the effort to create a manual document, improves the quality of the document and helps software developers create a clearer, more consistent format. Additionally, the Auto SRS Generator provides information in Real-time, meaning less human involvement and consequently the generation of higher quality documentation. The experimental evaluation demonstrated that the auto-generated SRS Document was well-structured, clear, and could produce a SRS Document faster than typically done. In addition, this work has demonstrated various components of the System Architecture and Workflow and described how future work will be aimed toward facilitating the adaptation of this Tool for Academic, Industrial and Large-Scale Software Engineering Projects.