The AI SDLC Manifesto: Rethinking AI native Software Development

Most organizations adopt AI in software engineering by providing developers with coding assistants and chatbots, but a more transformative approach is to redesign the entire software development lifecycle around AI. The author proposes a Specification-Driven Development model that leverages machine-consumable knowledge, warning that AI transformation could fail like Agile did if organizations repeat the same mistakes.

Member-only story The AI SDLC Manifesto: Rethinking AI native Software Development How Specification-Driven Development and machine-consumable knowledge can reshape software delivery Most organisations begin their adoption of AI in software engineering in roughly the same way. They provide developers with an AI coding assistant.They provide chatbots over their company knowledge-base. These tools can certainly improve individual productivity, but they do not fundamentally change how software is built. A more interesting possibility is to redesign the SDLC itself around AI. What if we apply AI transformation with respect to Agile principles. What would remain the same, what could change ? Why Most Transformations Fail The last major transformation in the industry was the change towards Agile software development methodology. However many organisations that adopted Agile still find it impossible to release even monthly. The best many could do was cut a yearly release cycle to half-yearly. This is worth revisiting, because the AI transformation could fail in exactly the same way if we repeat the mistake. Agile attempted to place enough decision-making ability inside one team so that the team could move without repeatedly waiting for…