SEER: S ustainability E nhanced E ngineering of S oftware R equirements

The rapid expansion of software development has significant environmental, technical, social, and economic impacts. The global effort to achieve the United Nations (UN) Sustainable Development Goals (SDGs) by 2030 necessitates the adoption of sustainable practices by developers. Existing sustainable software development methods mostly offer high-level guidelines that are time-consuming to implement and rely on team adaptability. Moreover, they focus on design or implementation, while sustainability assessment should start at the early requirements engineering phase. In this paper, we introduce SEER , a framework that addresses sustainability concerns in the early software development phase. The framework operates in three stages: (i) it identifies sustainability requirements (SRs) relevant to a specific software product from a general taxonomy; (ii) it evaluates how sustainable system requirements are based on the identified SRs; and (iii) it optimizes system requirements that fail to satisfy any SR. The framework is implemented using the reasoning capabilities of large language models and the agentic RAG (Retrieval Augmented Generation) approach. SEER has been experimented with four software case studies in different domains, combining with the knowledge contained in the UN SDG document, as well as the Sustainability Standard References of the Canadian Government. The results generated using the Gemini 2.5 reasoning model demonstrate the effectiveness of the proposed approach in accurately identifying a wide range of sustainability concerns in diverse domains.