Ex-Apple dev reveals his Agentic Engineering Workflow

In this technical discussion, David Ondrej interviews an experienced ex-Apple security developer about his sophisticated 'agentic' workflow. The conversation covers the development of a 'centralized memory system' (Cortex) and an action-oriented 'harness' (Bunker) that allow AI agents to operate autonomously, based on a deep, structured understanding of his personal and professional goals, beliefs, and operational procedures (the 'LifOS' philosophy).

The guest emphasizes the crucial need for individuals and companies to build 'context' for their AI agents. He argues that the primary failure of AI adoption is not the models themselves but the inability of users to effectively describe how their work, tasks, and business processes actually function. He demonstrates his own system, which uses these structured 'LifOS' documents as a single source of truth for all agent activities, effectively acting as an interface between human intent and machine execution.

Throughout the video, the pair reflects on the changing relationship between humans and AI, touching on the idea that standard ways of working (like using Jira or manual task management) will become obsolete. The guest outlines a future where humans focus on articulating high-level desires while agents handle the tactical 'how' of execution. The discussion concludes by exploring the potential impact of these systems on modern productivity and the necessity of maintaining a connection to reality, rather than fully surrendering to machine-driven processes.