The adoption of Generative AI (GenAI) in software development is currently dominated by the discussion of “vibe coding,” in which developers describe the desired functionality in natural language and let large language models generate the corresponding code with minimal review. This describes a fundamental shift in how software is developed and in the competencies necessary for a software developer. Vibe coding reduces the need for manual coding, leaving the software developer to focus on high-level design and code evaluation. While effective for prototypes, this approach is insufficient for complex software systems, where development spans analysis, architecture, design, testing, and deployment, and where every phase requires human understanding and evaluation. In this position paper, we argue that AI-assisted development of complex systems requires methodology encoding: the systematic translation of a project’s adopted methodology (architecture, patterns, workflows, standards, and roles) into AI tool configuration. This creates a configured agentic development environment that enables consistent AI assistance across the entire software development lifecycle. We define a new professional role, the Agentic Architect, who is responsible for designing this environment. We illustrate the concept using Claude Code and discuss implications for developer competencies and organizational roles.