AI coding agents are powerful, but they often lose the architectural and product context behind the code they change. This session introduces Knowledge-Driven Development (KDD) as a practical way to keep that knowledge alive and usable by both humans and AI agents. Attendees will learn how to structure a living knowledge layer, turn repository signals into focused context, connect decisions and work items to code, expose curated knowledge through MCP, and detect knowledge drift as the system evolves. The session is aimed at developers, platform engineers, architects, and teams adopting AI-assisted development.