AI Content Generation Agent via Telegram - n8n Workflow

Marketing managers seeking to automate content generation using voice commands.
Technical users looking to deploy advanced, persistent conversational AI agents using n8n.
Developers needing robust n8n templates for integrating LLMs (via OpenRouter/GPT-4) with complex, chained actions.
Agencies that require rapid drafting of social media posts, blog content, and marketing images.

This comprehensive n8n workflow transforms a Telegram chat into a command center for an AI Marketing Team. The primary challenge this n8n automation solves is the speed and variety required in modern content creation. Instead of manually interacting with several different tools, users send a single message (text or voice) to Telegram, and the central AI agent, powered by GPT-4, decides which specialized sub-workflow (tool) to use—whether it's generating a video, drafting a LinkedIn post, or creating a new image. This specific n8n template showcases the power of the n8n agent node to manage complex decision logic and task delegation, providing an instant, integrated content generation pipeline.

The n8n workflow begins with the Telegram Trigger n8n node, listening for incoming messages.

1. Input Detection: A Switch n8n node checks if the message contains text or a voice note.


2. Voice Processing: If a voice note is detected, the Download Voice File n8n node retrieves the audio, which is then sent to the Transcribe Audio n8n node (using OpenAI credentials) to convert it into text.


3. Agent Orchestration: The resulting text command feeds into the Marketing Team Agent n8n node. This agent is configured with Simple Memory (for retaining conversation history) and uses the GPT 4.1 n8n node (via OpenRouter) as its brain.


4. Tool Selection: Based on the user's intent (e.g., “write a blog post about X,” “create an image of Y”), the Agent intelligently selects and calls one of the pre-configured workflow tools using a ToolWorkflow n8n node. Tools include specific n8n templates for Blog Post, LinkedIn Post, Video, Create Image, Edit Image, and Search Images.


5. Execution and Response: Once the selected tool workflow completes its task, the output is returned to the Agent, which formats the final response. The final step of this n8n workflow sends the output back to the user via the Telegram n8n node, completing the automation cycle.

To deploy this powerful n8n workflow, follow these steps:

1. Import Core Workflows: Before importing this primary JSON n8n template, you must import the six supporting workflows referenced as tools: Video, LinkedIn Post, Blog Post, Create Image, Edit Image, and Search Images. These sub-workflows contain the actual logic for content generation and image manipulation.


2. Import This n8n Template: Import the provided JSON into your n8n instance.


3. Configure Credentials: Set up the required credentials in n8n:

1. Link Tools: In the Marketing Team Agent n8n node, ensure that the embedded ToolWorkflow nodes are correctly linked to the corresponding workflows imported in Step 1. The sticky note provides a detailed setup guide on necessary credentials for the sub-workflows (e.g., PiAPI, Runway, ElevenLabs, Tavily).


2. Activate: Save and activate the n8n trigger to start listening for Telegram commands.

Telegram Trigger: The starting point and primary n8n trigger for the entire system. It listens for message updates from the configured Telegram bot, initiating the n8n workflow.
Switch (Voice or Text): A critical flow control n8n node that determines if the incoming message contains text or a voice file, branching the execution path accordingly.
Transcribe Audio (OpenAI): If a voice message is detected, this n8n node uses the OpenAI Whisper model to accurately convert the speech into text input for the AI agent.
Simple Memory (LangChain): This memory n8n node stores the conversation history using the chat ID, allowing the Marketing Team Agent to remember context across multiple interactions.
GPT 4.1 (OpenRouter): The large language model n8n node, configured to use openai/gpt-4.1 via OpenRouter, serving as the core intelligence for the agent’s decision-making and content generation.
Marketing Team Agent (LangChain): The central orchestration n8n node. It receives the user's request, uses the LLM and memory, and intelligently decides which specialized ToolWorkflow n8n node to execute based on the user's intent, acting as the project manager for the whole n8n workflow.

• ToolWorkflows (Video, Blog Post, Create Image, etc.): These specialized n8n node types call external, linked n8n templates (sub-workflows) to perform specific, complex actions like creating a video or generating a LinkedIn post. Each tool is described to the Agent with a precise function description.