AI-powered development tools are evolving at an extraordinary pace. Currently, many frontrunners are guarded about their detailed experiences, withholding valuable knowledge to the community and slowing down progression within the area. This session breaks that pattern, drawing from hands-on experience inside a large enterprise. It will share the experiences of moving a traditional development model toward an agentic AI-augmented team, including how to handle enterprise limitations, align and upskill developers, designing AI-centric workflows, and confronting lessons learned from implementation reality, from the trenches.

Modern AI coding tools are no longer limited to autocomplete. They now can operate across the entire software delivery lifecycle, assisting with specification clarification, architectural scaffolding, structured code generation, automated test design, defect reproduction, pull request pre-screening, documentation synthesis, systematic refactoring, and much more. Some are evolving toward agentic patterns, capable of planning and executing multi-step tasks under human direction. When deliberately integrated, they reshape how individuals work, how teams collaborate, and how organizations think about throughput and engineering leverage.

Becoming an agentic AI development team is not about installing a tool. It's a mental shift and a journey of learning a new way of working. It requires rethinking how value is created in software teams. It requires structured experimentation with emerging capabilities, staying current without chasing every trend, defining shared vocabularies that guide generation, building prompt libraries aligned with architecture principles, and establishing governance patterns that address industry-specific requirements without slowing delivery. We will examine how these elements intersect in practice and what fundamentally changes when engineers move from primarily writing code to designing, directing, and constraining AI systems.