Matthias Priebe — Solution ArchitectThe Importance of Early-Stage Requirements Engineering
The earlier high-quality Requirements Engineering is applied, the faster a coordinated and prioritized target image develops from the initial rough concept. This optimizes both the technical product design and the associated development costs. The key term here is “Shift-Left,” which refers to starting quality and cost control early in the development process.
The Use of Artificial Intelligence in Requirements Engineering
The precise and efficient capturing of requirements can be enhanced through the use of Artificial Intelligence (AI) tools like ChatGPT. Since requirements are often expressed in natural language, AI serves as an ideal partner for analyzing them. With its extensive knowledge base, AI can identify missing requirements and supplement them in line with the overall vision.
Training and Education for the Optimal Use of AI
Effective use of AI in Requirements Engineering requires reproducible processes and in-depth knowledge of Requirements Engineering methods. During the “KI-REG - Künstliche Intelligenz im Requirements Engineering” training offered by SOPHIST GmbH, we explored these topics in depth. Here are some of the key insights from the training: Understanding the Basics: A solid understanding of Requirements Engineering is essential to evaluate the quality and relevance of the requirements. Generating Prompts: AI can assist in generating prompts to facilitate easy return to a specific state or point in the process. Quality of Requirements: The quality of the requirements largely depends on effective prompting. A step-by-step approach and clear delineation of context improve outcomes. Prompting Frameworks: Familiarity with prompting frameworks further optimizes results. Dialogue Management: It is important to know when to start a new dialogue with the AI. Lingering too long in a dialogue or starting new ones too frequently can negatively impact results. When starting a new dialogue, setting the context correctly is crucial. Interdisciplinary Perspectives: AI can behave like an interdisciplinary team, examining requirements from multiple perspectives. Retrieval-Augmented Generation: Specialized tasks can be better addressed through Retrieval-Augmented Generation. This approach allows the creation of GPT experts tailored to specific tasks, such as formulating requirements in a particular template.
Conclusion and Outlook
These insights will help us manage requirements more quickly and efficiently in the future and accelerate iterative idea generation. Although working with AI in the context of Requirements Engineering still requires a solid foundation in the field due to the inherent randomness of AI outputs, AI support can rapidly generate results that facilitate and enhance the thought process.
Special thanks to SOPHIST GmbH for the excellent training and the many new ideas for our daily work. We look forward to applying what we’ve learned in practice and continuously improving our processes.
