Multilingual E-commerce Copywriting with Gemini Vision AI - n8n Workflow

E-commerce Managers and Shopify store owners who sell globally.
Content teams needing rapid, accurate translations and copywriting at scale.
Users looking for advanced examples of prompt engineering and structured AI output using the n8n agent node.
Developers needing reusable n8n templates for vision-based AI generation.

Generating high-quality, localized product descriptions across multiple international storefronts is a time-consuming challenge for e-commerce businesses. This comprehensive n8n workflow solves this by integrating Shopify data directly with advanced Google Gemini Vision AI capabilities.

First, the n8n workflow extracts product details and the main image URL from Shopify. It uses the Gemini model to objectively analyze the image, generating a 'vision description' of visible attributes (materials, colors, features). This description, along with original product data, is fed into a highly specialized n8n agent node with a strict system prompt.

This system prompt instructs the AI to act as a professional copywriter, generating product names, HTML descriptions, and SEO-friendly URL handles in six languages (ES, DE, EN, FR, IT, PT). The entire process is managed within a single n8n execution, ensuring consistency and structured output, which is then neatly organized and saved to a Google Sheet.

The process begins with the Manual Trigger n8n node, acting as the starting n8n trigger for execution.

1. Shopify Data Retrieval: The Shopify n8n node fetches a batch of products, including details like title, vendor, and the primary image URL.


2. Vision Analysis: The Google Gemini n8n node (models/gemini-2.5-flash-lite) analyzes the product image URL. The prompt demands an objective, detailed description of visible features, serving as verifiable input for the subsequent copywriting stage.


3. Multilingual Copywriting Agent: The core logic resides in the AI Agent n8n node. This node combines the visual analysis, existing Shopify product data, and a powerful system prompt defining strict linguistic rules (footwear terminology, handle generation) and the mandatory JSON output structure. It utilizes the dedicated Gemini Chat Model n8n node for the heavy lifting.


4. Structured Parsing: The Structured Output Parser n8n node ensures that the AI response adheres strictly to the defined multilingual JSON schema, preventing hallucinations or formatting errors.


5. Data Expansion and Sanitization: The Code n8n node, named 'Expand Languages & Sanitize', iterates over the six language keys (ES, DE, EN, FR, IT, PT) in the structured output. It transforms the single JSON object into six separate n8n items, linking the translated copy and handles back to the original Shopify product_id.


6. Google Sheets Output: Finally, the Google Sheets n8n node appends each language-specific item as a row into the designated spreadsheet for review and eventual bulk import back into Shopify.

To utilize this complex n8n workflow, follow these steps:

1. Import the n8n templates: Copy the provided JSON and import it into your n8n instance.


2. Shopify Credential Setup: Configure the Shopify n8n node (Get many products) using an API Access Token. Ensure the token has read access to products.


3. Google Gemini Credential Setup: Set up your Google Gemini (formerly PaLM) API account credentials and link them to the Analyze image and Google Gemini Chat Model n8n nodes.


4. Google Sheets Credential Setup: Configure the Google Sheets n8n node (Append row in sheet) using a Service Account or standard authentication. You must specify the Document ID and the Sheet Name where the multilingual output will be stored.


5. Review and Customize: Examine the parameters in the Get many products n8n node (currently set to limit: 2) and the detailed system prompt within the AI Agent n8n node to tailor the copywriting tone or add more languages if needed. This reusable n8n workflow is ready to execute once credentials are set.

When clicking ‘Execute workflow’ (Manual Trigger): Serves as the initial n8n trigger to manually start the generation process.
Get many products (Shopify n8n node): Fetches product data from the Shopify store. Key Configuration: Resource: product, Operation: getAll, Limit: 2 (for testing).
Analyze image (Google Gemini n8n node): Utilizes the Gemini 2.5 Vision model to generate an objective physical description of the product based on its primary image URL (images[0].src). Key Configuration: Model: models/gemini-2.5-flash-lite.
AI Agent (Langchain Agent n8n node): Orchestrates the multilingual generation using complex input and rules. It uses data from Shopify and the vision analysis to formulate a powerful prompt.
Structured Output Parser (Langchain n8n node): Essential for maintaining the integrity of the n8n workflow, forcing the AI output into a guaranteed JSON structure containing ES, DE, EN, FR, IT, PT keys.
Expand Languages & Sanitize (Code n8n node): A critical custom n8n node that transforms the single, multi-language JSON result into six individual items, making it iterable and suitable for writing to Google Sheets.

• Append row in sheet (Google Sheets n8n node): The final step, writing the translated product name, description, and handle for each language into a designated spreadsheet.