What happens when a well-defined user story becomes working, tested, and documented code with a pull request open in under an hour? This session explores that question through real-world experience at Atlassian, working across large-scale codebases, monorepos, and distributed teams. It examines how the traditional software development lifecycle, designed for a world where humans were the primary producers of code, is being reshaped. The talk introduces five structural shifts that define an AI-native SDLC, including moving from writing code to specifying intent, from sequential phases to parallel workstreams, from reactive bug fixing to predictive code health, from documentation as an afterthought to living knowledge, and from human gatekeeper to human governor.

Beyond these shifts, the session addresses challenges encountered at scale, including AI systems that lack awareness of architectural context, governance models that do not scale across tools, and hallucinations that pass tests but fail in production over time. It shares internal engineering tooling developed to address these gaps, such as semantic codebase search, a pull request knowledge graph, a code governance engine, hallucination detection against live service catalogues, and an AI migration platform for large-scale code changes. The session concludes with a grounded view of what works today, what is improving, and what remains uncertain, along with practical actions that teams can take in the near term.

What You Will Learn

• The structural shifts that define an AI-native software development lifecycle
• Challenges of applying AI at scale, including governance and hallucination risks
• Tooling approaches to improve code understanding, governance, and large-scale change

Who Should Attend

• Developers
• Software Engineers
• Architects
• Technical Leads
• Engineering Managers
• Engineering Leaders