Requirements AssIstant data

Requirements Engineering (RE) plays a critical role in software development, yet traditional elicitation and specification practices remain time-consuming, error-prone, and heavily dependent on human interpretation and documentation. These challenges are intensified in dynamic environments where requirements evolve rapidly, and communication gaps between stakeholders and developers can lead to inconsistencies, rework, and project delays. To help address these issues, we have explored AI technologies and developed RAI (Requirements AssIstant), a multi-agent tool that has an architecture based on Large Language Models (LLMs) and automates key RE activities. We conducted a multiple case study with four organizations from different sectors to assess the effectiveness and feasibility of our proposal. The results suggest that the tool reduced the time and effort for requirements elicitation and documentation — transforming a process that typically takes several days into a workflow completed within minutes — while producing results with high levels of accuracy, relevance, and coverage. These promising results highlight the potential of LLM-based agents to support agile, high-quality, and cost-effective RE.