Context Management for Everyday Software Development - Ragunath Jawahar

This talk addresses the challenges posed by AI coding agents and how their chat-style interfaces can lead to context contamination and degradation of code quality. The speaker presents practical strategies for creating and managing context when working with these agents, introducing methodical techniques that are not tied to specific tools. By understanding context as an information architecture issue rather than relying on chat-based interaction patterns, teams can develop proactive designs that align with the processing methods of large language models. This session equips attendees with the knowledge to maintain high-quality context and improve the reliability of AI-assisted code outputs.