Descriptions:
Nate Herk explains the conceptual and practical shift from traditional node-based n8n automation to agentic AI workflows, then builds a live working example inside Claude Code. The video opens by contrasting the manual, step-by-step nature of n8n (drag nodes, configure each connection, debug errors manually) with the outcome-driven model of agentic systems, where you describe a goal in plain language and the agent determines its own execution path.
Herk outlines four core properties that define agentic workflows: self-healing (the agent detects, diagnoses, and fixes its own runtime errors without user intervention), true natural language control (the agent conducts a structured interview before writing any code, asking clarifying questions users often wouldn’t think to raise until weeks later), dynamic tool selection (it identifies the right APIs and functions based on context), and persistent memory across sessions. These aren’t framed as future capabilities — Herk demonstrates them live.
The hands-on build tasks Claude Code with generating dentist leads in Chicago for his AI agency UpAI: scraping business data via the Google Places API, enriching profiles, writing personalized outreach messages, and exporting everything to Google Sheets. The agent creates a full implementation plan with clarifying questions, builds three modular tools, and executes them sequentially — surfacing the real-world gap between scripted workflow automation and adaptive agent behavior in a single contained demo.
📺 Source: Nate Herk · Published January 25, 2026
🏷️ Format: Hands On Build
