Build a Company Document RAG Chatbot with Google Drive and Gemini

Input: Google Drive folder with company documents; Processing: n8n auto-indexes into Pinecone, Chat Trigger receives questions, AI Agent retrieves and generates answers; Output: Context-aware document Q&A chat

Who

Teams needing a quick internal document-based Q&A assistant

You get

Employees upload or update documents in Google Drive and the system auto-indexes them; answers are generated from retrieved document content with conversation context

You need

n8n · google-drive · pinecone · gemini

Steps

1. Prepare external accounts and credentials

   Complete three account setups outside n8n, then add the corresponding credentials in n8n. 1. Google Cloud: Create a project and enable the Vertex AI API. 2. Google AI Studio: Obtain an API Key for Gemini model calls. 3. Pinecone: Sign up, get your API Key, and create an index named company-files (choose dimensions and metric compatible with your embedding model). 4. Google Drive: Create a dedicated folder for company documents. 5. n8n Credentials: Add three credentials: - Google Drive OAuth2 - Google Gemini (PaLM) API (enter your Google AI API Key) - Pinecone API (enter your Pinecone API Key)

2. Import the workflow and configure triggers and vector store

   Import this workflow from the n8n template library (ID 2753) or via workflow.json. Then configure two node groups: - Both Google Drive Trigger nodes (File Created / Updated): set folderToWatch to your dedicated Google Drive folder. - Both Pinecone Vector Store nodes (insert and retrieval): set pineconeIndex to company-files. Leave other node parameters at their template defaults to run.

3. Configure the AI Agent system message

   Open the AI Agent node and enter the following prompt in Options > System Message. This prompt controls how the bot retrieves documents and answers employee questions.

   Execution prompt · Source text。AI Agent node Options > System Message field

   Copy prompt

   You are a helpful HR assistant designed to answer employee questions based on company policies.
   
   Retrieve relevant information from the provided internal documents and provide a concise, accurate, and informative answer to the employee's question.
   
   Use the tool called "company_documents_tool" to retrieve any information from the company's documents.
   
   If the answer cannot be found in the provided documents, respond with "I cannot find the answer in the available resources."

4. Verify auto-indexing and Q&A

   Activate the workflow and run two checks: 1. Auto-indexing: Upload a new document (or update an existing one) to the target Google Drive folder. Wait about 1 minute (poll interval), then check the Pinecone console to confirm the company-files index contains new text chunks. 2. Q&A test: Open the Chat Trigger chat interface and ask a document-related question (e.g., "What is our annual leave policy?"). Expect the AI Agent to call company_documents_tool to query Pinecone, then the Google Gemini Chat Model to generate an answer grounded in the documents. Window Buffer Memory retains recent conversation context for multi-turn follow-ups.

You are assisting someone to set up this n8n RAG chatbot. Confirm they have completed: 1) Google Cloud / Google AI / Pinecone accounts and credentials; 2) workflow import and node configuration; 3) AI Agent systemMessage filled in. If stuck on credentials, guide them to check OAuth callback URLs and API Key permission scopes.