This slide deck provides an overview of how agentic AI systems can support the real-world work of Gene Ontology (GO) and related biomedical knowledge base development. While AI is often applied to narrow tasks such as term extraction or text summarization, most ontology work consists of more complex activities: reviewing extensive literature, synthesizing multiple lines of evidence, evolving design patterns, refactoring large ontology branches, performing database-wide quality control, and coordinating consensus among diverse stakeholders. These activities are poorly matched to traditional “LLM-as-oracle” usage patterns. The material introduces agentic AI as a practical framework for addressing these needs through tool-augmented, stateful agents operating over open-ended workflows. Examples are drawn from GO and Mondo, illustrating how agents can be integrated into GitHub-based processes to research biology, draft ontology changes, run deterministic validation and reasoning checks, manage pull requests, summarize project progress, and assist with issue triage and maintenance tasks. Emphasis is placed on interactive use, human oversight, and deterministic safeguards to reduce hallucinations and maintain curation quality. The slides also outline challenges in evaluating agentic systems for ontology development, including the presence of multiple valid modeling solutions and the consensus-driven nature of knowledge representation. Overall, the resource documents current practices, architectural patterns, and limitations encountered when using agentic AI as a collaborative assistant in large, community-governed biological knowledge bases, rather than as a replacement for expert curation