NemoClaw Knowledge Wiki

Building an AI Marketing Team with Claude Code Agents Skills

4 min read

Building an AI Marketing Team with Claude Code Agents & Skills

Clip title: Claude Code: Build Your Full AI Marketing Team (Agents + Claude Skills) Author / channel: Grace Leung URL: https://www.youtube.com/watch?v=yLXLHnD4fco

Summary

This video provides a detailed, step-by-step guide on how to build and manage an AI marketing team using Claude Code, Anthropic’s integrated development environment. The core concept revolves around transforming Claude into a comprehensive AI marketing team capable of research, writing, analysis, and design, all working collaboratively. The presenter illustrates this by creating a marketing team for a fictional travel brand, GoTravel, involving five specialized AI agents and twelve marketing-specific skills. The video aims to equip viewers, even non-technical ones, with the knowledge to set up such a system from scratch.

The setup begins with establishing a structured project environment in VS Code, including specific folders for brand context (_context), standard operating procedures (_sop), reusable templates (_templates), and various marketing outputs. Crucially, a CLAUDE.md file is created, serving as the central instruction document that Claude reads to understand the project’s structure, brand guidelines, and operational rules for agents and skills. Following this, the video demonstrates creating custom marketing skills. Two methods are highlighted: a “reference-based method” where Claude analyzes an existing branded presentation template to create a new “branded-deck” skill that adheres to specific design rules, and integrating with external tools by setting up an .mcp.json file to connect Claude to an image generation model (Nano Banana) for a “social creative designer” skill, enabling it to produce on-brand social visuals.

The next phase focuses on building specialized AI agents to manage different aspects of marketing. The presenter explains that while individual skills are useful, combining too many into one conversation can dilute Claude’s focus, much like a single person trying to do too many different roles. To overcome this, agents are created, each with a defined role, specific tools, and a set of relevant skills. Examples include a “Data Analyst” agent, equipped with data visualization and campaign reporting skills to produce interactive performance dashboards, and a “Content Creator” agent, utilizing blog writing, social copy, lead magnet, and landing page builder skills to generate integrated content. These agents are designed to be brand-agnostic, pulling brand context at runtime, ensuring reusability across various brands.

Finally, the video demonstrates how to orchestrate these agents and integrate the AI team into a real-world workflow. The CLAUDE.md file is updated to include explicit agent routing rules, dictating when Claude should delegate tasks to specific agents based on the nature and complexity of the request. For collaborative task management, the system is integrated with Notion, allowing the AI agents to scan a shared task board, prioritize tasks, execute them (researching, designing, writing, analyzing), and update their status with output file paths. The presenter also showcases a remote control feature, enabling users to interact with their local Claude Code session and assign tasks to their AI team directly from a mobile device, effectively creating a 24/7 AI marketing team.

Graph View

Backlinks