This AI Agent Runs My $100K Media Company (free n8n template)

In today’s fast-paced digital landscape, managing content creation and marketing workflows can quickly become overwhelming. Imagine having an AI-powered assistant that handles your daily newsletter, creates custom images, repurposes content for social media, and even generates short-form video scripts and videos—all without the need for a large team. This is exactly what we've built: an AI marketing agent that runs the core work-streams of a $100K media company, saving time and costs.

If you're interested in automating marketing tasks and want to learn how to build such an AI agent yourself, consider joining our AI Automation Community. There, you can access the exact n8n templates, integrations, and prompts powering this agent, along with support from fellow AI builders.

How the AI Marketing Agent Streamlines Content Creation

Our AI agent acts as a senior marketing employee, seamlessly integrating into our daily workflow. It handles multiple tasks including:

Writing a daily AI industry newsletter
Generating custom images aligned with brand guidelines
Repurposing newsletter content into engaging Twitter threads
Creating viral-style short-form video scripts for platforms like TikTok and Instagram Reels
Producing avatar-style talking head videos for content promotion
Conducting deep content research and emailing detailed reports

By automating these tasks, the agent effectively replaces a full content team at a fraction of the cost.

Voice-Activated Workflow with 11 Labs and n8n

The AI agent's workflow begins with an ElevenLabs voice agent, which serves as the entry point for user requests. Through a conversational interface, team members can speak commands such as “write today’s newsletter” or “generate Twitter thread.” When the voice agent receives a request it cannot fulfill on its own, it forwards the task to an internal n8n AI agent via an HTTP webhook.

This two-agent system allows for smooth handling of complex marketing tasks. The voice agent manages real-time conversational interactions, while the n8n agent processes the work asynchronously, leveraging various custom and built-in tools to complete the requests.

Setting Up the ElevenLabs Voice Agent

Creating the voice agent involves configuring a new agent within ElevenLabs, starting from a blank template. Key steps include:

Selecting a voice persona—for example, a Jarvis clone from Iron Man for a witty, data-driven tone.
Adjusting speech speed and other voice settings to suit team preferences.
Crafting a detailed system prompt that defines the agent’s personality, environment, goals, and guardrails.
Adding a custom tool called Forward Marketing Request, configured as a webhook to send complex tasks to the n8n agent.

The system prompt is crucial because it guides the voice agent’s behavior. We defined a personality that is proactive, witty, slightly sarcastic, growth-obsessed, and concise.

Guardrails help prevent the agent from going off track, such as avoiding inappropriate responses or tasks outside its scope.

Building the n8n AI Agent

The n8n AI agent is the brain behind the scenes. It receives user messages forwarded from ElevenLabs, analyzes the request, and decides which tools to use to fulfill the task. This modular approach allows the agent to handle a wide variety of marketing workflows.

Key components of the n8n agent include:

Input nodes: Accept chat messages or webhook triggers.
AI agent node: Processes the prompt and selects tools.
Memory node: Stores interactions and outputs to maintain context throughout the day.
Tool nodes: Custom workflows for newsletter writing, image generation, social media repurposing, video script creation, and more.
No-operation node: Manages asynchronous task completion without immediate response.

Configuring Input and Memory

For testing, the agent can receive prompts via a chat trigger node. In production, it listens to webhook triggers from the voice agent. To handle both input methods smoothly, the agent uses a JavaScript expression that prefers chat input but falls back to webhook data.

Memory is configured to store all interactions for the current day, keyed by the date. This allows the agent to recall previous outputs like newsletter content, which is essential for repurposing tasks such as creating Twitter threads or video scripts.

Enhancing Reasoning with the Think Tool

We incorporated a think tool that acts as an internal scratchpad, allowing the agent to reason through multi-step or complex tasks. This tool helps the agent plan and execute workflows more effectively by appending thoughts to a log without altering data.

Crafting the n8n Agent System Prompt

The system prompt defines the AI agent’s identity, purpose, capabilities, tools, and environmental context. We designed it to:

Position the agent as a marketing team assistant integrated into daily workflows.
Emphasize its mission to empower marketing members to complete tasks more effectively.
Detail the core capabilities such as content creation, strategic repurposing, and multi-channel marketing asset generation.
List the available tools and specify their roles, e.g., newsletter writing, image generation, Twitter thread creation, video script writing, and avatar video production.
Explain memory and context management to avoid duplication and ensure continuity.
Include dynamic variables like the current date and any relevant guidelines or constraints.

These detailed instructions enable the AI agent to make better decisions and produce consistent, high-quality marketing outputs.

Adding Custom Tools to the Agent

Each marketing function is implemented as a custom tool within n8n, connected to the AI agent through workflow triggers. Examples include:

Write Newsletter Tool: Generates the daily newsletter by aggregating AI news and Twitter content, selecting top stories, crafting subject lines, and formatting the email.
Generate Image Tool: Creates custom images for newsletter sections using AI image generation APIs, applying brand style guidelines such as watercolor painting effects.
Repurpose to Twitter Tool: Transforms newsletter content into engaging Twitter threads, using example threads for style reference.
Write Short Form Video Script Tool: Produces viral-style scripts for TikTok and Instagram Reels based on newsletter stories.
Generate Talking Avatar Video Tool: Creates avatar-style talking head videos by sending scripts to HeyGen’s API, providing ready-to-edit promotional videos.
Email Research Report Tool: Sends deep research reports generated by Perplexity’s research model directly to inboxes via Gmail integration.

This workflow starts by pulling markdown news content and Twitter data for the day, cleansing it, and then using a large language model (Gemini 2.5 Pro) to select top stories. A human-in-the-loop approval step ensures quality before crafting the subject line and individual story sections. The final newsletter is formatted and shared internally for publishing.

Generate Image Tool Workflow

After the newsletter is drafted, the agent generates images for each top story. Brand guidelines are embedded into the prompt to ensure style consistency. The AI image generation API returns images as base64 strings, which are converted to files and shared via Slack for easy access.

Repurpose to Twitter Workflow

Once the newsletter exists in memory, the agent creates Twitter threads by studying example tweets and applying a crafted prompt to maintain voice and engagement. This process ensures that content is repurposed effectively without duplication.

Write Short Form Video Script Workflow

Using scraped news and tweets, the agent identifies compelling stories and generates multiple short-form video script options designed to capture audience attention on TikTok and Instagram Reels.

Generate Talking Avatar Video Workflow

The AI agent selects a script from memory and sends it to HeyGen’s API to create a talking head avatar video. The workflow manages video generation status and posts the final video for review and editing.

Content Research and Email Reporting Tools

For deep content research, the agent uses Perplexity’s research model to gather detailed information on topics like the model context protocol. Once the research is complete, the agent formats the findings into an email and sends it to team members for review.

Final Thoughts and Getting Started

This AI marketing agent demonstrates how to combine voice interfaces, AI reasoning, and automation workflows to handle complex marketing tasks with minimal human intervention. The key is designing modular tools and crafting detailed prompts and system instructions that guide the AI to act as a reliable and strategic team member.

Keep in mind that this system is not a plug-and-play solution. Each business or client may require customized tools, prompts, and workflows tailored to their unique marketing style and goals.

If you want to build your own AI marketing agent or automate other business workflows, join our AI Automation Mastery community. You’ll get free access to the n8n template powering this agent, plus expert guidance and a supportive network of AI builders.