Descriptions:
Greg Isenberg hosts entrepreneur and builder Remy Gasill for a beginner-focused course on AI agents, covering the conceptual shift from chat interfaces to autonomous goal-execution systems. Gasill frames the core distinction simply: chat is “question to answer,” agents are “goal to result” — and walks through the internal agent loop of observe, think, and act that drives multi-step autonomous execution.
The course centers on practical setup using platforms like Claude Code, Codex, OpenClaw, and Manis. A significant portion covers claude.md files — the mechanism for “onboarding” agents with personal and business context, structured across role-specific folders (executive assistant, head of marketing, CFO). Gasill shows how these files define an agent’s persona, responsibilities, and knowledge base so it operates consistently across tasks.
Perhaps the most actionable segment addresses memory persistence — the common frustration where agents forget preferences between sessions. Gasill demonstrates adding a self-update instruction to the claude.md file, creating a loop where agents write new preferences and learnings back to their own context files. The course is aimed squarely at non-technical operators who want to use agents to run business departments without writing code, using tools already widely available in 2026.
📺 Source: Greg Isenberg · Published March 17, 2026
🏷️ Format: Course Lesson
