dolumturk.com AI E-commerce Chatbot with API ,whatsapp ,telegram , instagram , tiktok and other social media Integration with n8n platform
Budget: $250 – $750 USD
Project Title: Advanced AI E-commerce Assistant with n8n & Google Drive Integration
Project Overview:
We are seeking an expert-level n8n developer to build a sophisticated AI sales and support assistant for our e-commerce website, dolumturk.com. The core challenge of this project is to create a reliable product search functionality by processing a local data file, as direct API calls for searching have proven unreliable.
The final chatbot will be integrated into our website and WhatsApp, serving as an intelligent assistant that can handle product inquiries, order status, and other customer service tasks with high accuracy and speed.
Phase 1: Data Preparation & Secure Setup
This foundational phase must be understood by the freelancer, but will be completed by us.
Task 1.1: Product Data Export: We will export our entire product catalog (approx. 10,000 items) from our Ideasoft e-commerce platform into a single chatbot.xml file. This file will contain all necessary fields: id, name, stockCode, price1, stockAmount, details, picture1Path, etc.
Task 1.2: Cloud Storage: We will upload this chatbot.xml file to our company's Google Drive.
Task 1.3: Secure API Access (Service Account): We will create a Google Cloud Service Account, enable the Google Drive API, and generate a JSON key file. We will then share the chatbot.xml file on Google Drive with this service account's email address to grant read-only access.
Task 1.4: n8n Credential Setup: We will add the generated JSON key to our n8n instance as a "Google Drive API" credential using the "Service Account" method.
Your work begins in Phase 2, assuming the credential from this phase is ready for use.
Phase 2: n8n Workflow Development - The "Custom Search Engine"
This is the core of the project. You will build the n8n workflow that powers the chatbot.
Step 2.1: Fetch and Parse the Database
Nodes: Google Drive -> Google Drive -> XML
Process:
The first Google Drive node will use the "Search" operation to find the chatbot.xml file by its name.
The second Google Drive node will take the File ID from the first node and use the "Download" operation to fetch the raw file content.
The XML node will take the raw XML data and parse it into a structured JSON array of product objects, making it usable for the next steps.
Step 2.2: The Custom Filter Engine
Node: Code
Process: This is the most critical node in the workflow. It will act as our custom search engine.
Input 1: The full JSON array of all 10,000 products from the XML node.
Input 2: The user's raw message from the chat trigger (e.g., chatInput).
Logic: You will write JavaScript code inside this node to:
Loop through the entire product array.
For each product, check if the user's search terms are present in the product's name, SKU (stockCode), or other relevant fields. The search should be flexible (e.g., case-insensitive).
Create a new, filtered array containing only the product objects that match the user's query.
Step 2.3: The AI Interpreter
Node: AI Agent (connected to a Google Gemini Chat Model)
Process: This node will receive the filtered list of matching products from the Code node.
Task (System Prompt): The agent's new role is not to "search", but to "interpret and present". Its instructions will be:
"You will be given a pre-filtered list of one or more products that match the user's query. Your task is to analyze the data for these products (name, price, stock, details) and formulate a professional, clear, and helpful response in natural language. If the list is empty, state that the product could not be found. If there are multiple products, present them as clear options."
This architecture ensures the process is fast, reliable, and not dependent on a faulty external API search function.
Project Overview:
We are seeking an expert-level n8n developer to build a sophisticated AI sales and support assistant for our e-commerce website, dolumturk.com. The core challenge of this project is to create a reliable product search functionality by processing a local data file, as direct API calls for searching have proven unreliable.
The final chatbot will be integrated into our website and WhatsApp, serving as an intelligent assistant that can handle product inquiries, order status, and other customer service tasks with high accuracy and speed.
Phase 1: Data Preparation & Secure Setup
This foundational phase must be understood by the freelancer, but will be completed by us.
Task 1.1: Product Data Export: We will export our entire product catalog (approx. 10,000 items) from our Ideasoft e-commerce platform into a single chatbot.xml file. This file will contain all necessary fields: id, name, stockCode, price1, stockAmount, details, picture1Path, etc.
Task 1.2: Cloud Storage: We will upload this chatbot.xml file to our company's Google Drive.
Task 1.3: Secure API Access (Service Account): We will create a Google Cloud Service Account, enable the Google Drive API, and generate a JSON key file. We will then share the chatbot.xml file on Google Drive with this service account's email address to grant read-only access.
Task 1.4: n8n Credential Setup: We will add the generated JSON key to our n8n instance as a "Google Drive API" credential using the "Service Account" method.
Your work begins in Phase 2, assuming the credential from this phase is ready for use.
Phase 2: n8n Workflow Development - The "Custom Search Engine"
This is the core of the project. You will build the n8n workflow that powers the chatbot.
Step 2.1: Fetch and Parse the Database
Nodes: Google Drive -> Google Drive -> XML
Process:
The first Google Drive node will use the "Search" operation to find the chatbot.xml file by its name.
The second Google Drive node will take the File ID from the first node and use the "Download" operation to fetch the raw file content.
The XML node will take the raw XML data and parse it into a structured JSON array of product objects, making it usable for the next steps.
Step 2.2: The Custom Filter Engine
Node: Code
Process: This is the most critical node in the workflow. It will act as our custom search engine.
Input 1: The full JSON array of all 10,000 products from the XML node.
Input 2: The user's raw message from the chat trigger (e.g., chatInput).
Logic: You will write JavaScript code inside this node to:
Loop through the entire product array.
For each product, check if the user's search terms are present in the product's name, SKU (stockCode), or other relevant fields. The search should be flexible (e.g., case-insensitive).
Create a new, filtered array containing only the product objects that match the user's query.
Step 2.3: The AI Interpreter
Node: AI Agent (connected to a Google Gemini Chat Model)
Process: This node will receive the filtered list of matching products from the Code node.
Task (System Prompt): The agent's new role is not to "search", but to "interpret and present". Its instructions will be:
"You will be given a pre-filtered list of one or more products that match the user's query. Your task is to analyze the data for these products (name, price, stock, details) and formulate a professional, clear, and helpful response in natural language. If the list is empty, state that the product could not be found. If there are multiple products, present them as clear options."
This architecture ensures the process is fast, reliable, and not dependent on a faulty external API search function.