AI in the Agentic and Agentâ¦Age
This talk provides a critical review of the intersection between open source and AI in the emerging “agentic” era, moving beyond hype-driven experimentation toward a pragmatic framework for production-ready systems. We’ll talk through aligning technical definitions for generative AI (genAI) and its problem-solving scope, then introduce a rigorous set of five fundamental questions to anchor the development lifecycle against known best practices for policy, licensing, and risk management. We’ll spend time clarifying the three distinct, often conflated, definitions of an “AI Agent”: the architectural approach (chained LLM calls), the labor unit (AI employees), and the UI-operating program (software acting on your behalf). Finally, we explore the genuinely novel shifts in this wave-specifically the evolution of machine-to-machine interactions and the unprecedented capacity for global deployment at scale-to identify actionable mitigations for contributors and maintainers navigating this socio-technical landscape.
Bio: Amanda Casari is a staff developer relations engineer in Google’s Open Source Programs Office, where she leads the Contributor Experience and Open Source + AI teams. She is an engineer and researcher who has worked in many technical disciplines for over 20 years, including developer relations, product management, data science, complexity science, and robotics. Amanda was named an external faculty member of the Vermont Complex Systems Center in 2021 and co-authored Feature Engineering for Machine Learning Principles and Techniques for Data Scientists for O’Reilly. Her community work includes volunteering and serving as President on the Board of Directors for Green Mountain Roller Derby. Amanda is persistently fascinated by complexity, the differences between the systems we aim to create and the ones that emerge, roller derby, and pie.