As AI powered development tools mature, software teams face a critical inflection point in how to integrate agentic systems into existing workflows without abandoning the context embedded in your codebase, documentation, and project management tooling.

This talk examines how Spec Driven Development reframes the collaboration model in brownfield environments, where agents must reason over legacy repositories, existing tickets, and accumulated architectural decisions rather than starting from a clean slate. By treating specifications as first class artifacts, teams establish a shared contract that both humans and agents can interpret, reducing ambiguity and drift across iterations.

A key focus is the role of wrappers and manifest layer tooling, specifically how teams can shape the context presented to underlying language models to improve translation fidelity between product intent and generated output. Structuring these wrappers to surface relevant repository context, enforce team conventions, and constrain agent behavior to acceptable solution spaces is covered in depth.

The talk maps out what an agentic team topology looks like in practice, covering how Guardians and Airmen operating at the intersection of human intuition and machine intelligence can divide responsibility with AI agents and know when to delegate versus when human judgment is essential. Custom tailored training curriculum and real world use cases demonstrate how existing personnel can upskill incrementally without requiring wholesale transformation of their current role. Attendees will leave with a concrete framework for building an AI fluent workforce while preserving the institutional knowledge they have already built.