Descriptions:
Nate Herk demonstrates how to turn Claude Code into a fully functional executive assistant that handles daily planning, team pulse checks, content creation, and data visualization — all running as parallel sub-agents simultaneously. The video is structured as a complete build guide, walking from initial onboarding through the final folder architecture that powers the assistant.
The system works by giving Claude Code persistent context through a set of structured markdown files: ME (personal profile and preferences), WORK (active projects and goals), TEAM (collaborator details), and CURRENT PRIORITIES. These are populated via an interactive onboarding conversation and connected to live tools including Google Calendar and project management platforms like ClickUp, Asana, or Notion. Skills — plain-text markdown recipes — tell the agent exactly how to handle recurring tasks like the “morning coffee” planning session or the team pulse check, ensuring consistent output every time they’re invoked.
The live demo shows four agents running concurrently: one pulling together a calendar-based day plan, one spinning up a research sub-agent to write a LinkedIn post and Twitter carousel about Claude Code’s voice feature, one auditing team progress across quarterly goals, and one generating a custom visualization. Herk covers the git-backed memory structure, how to keep context files current, and how to migrate workflows away from ChatGPT Projects into this self-contained setup — making the case that depth of context, not model choice, is the primary lever for assistant quality.
📺 Source: Nate Herk | AI Automation · Published March 05, 2026
🏷️ Format: Hands On Build
