Natural Language to Google Sheets SQL Converter (GPT-4o + n8n)

Chat to Google Sheets → AI generates Sheets query URL → HTTP fetches data → Formatted table output

Who

Anyone who wants to build a natural-language query interface for Google Sheets in n8n, with basic familiarity with API keys and OAuth2 setup.

You get

Type questions in plain language and get back queried results from Google Sheets via auto-generated SQL-like URLs; supports multi-turn conversation context.

You need

n8n · google-sheets

Steps

1. Prepare environment and import workflow

   1. Ensure you have an n8n instance, an OpenAI account with payment method, and a Google account. 2. Copy the workflow JSON provided by the template. 3. In n8n, go to Workflows → Import from JSON, paste the JSON, and import. 4. Save the workflow with a descriptive name such as "Google Sheets SQL Query Generator".

2. Configure OpenAI credentials

   1. Go to OpenAI Platform, create an API Key, and ensure your account has billing enabled. 2. In the workflow, open the "OpenAI Chat Model1 (GPT-4o)" node, click Create New Credential, select OpenAI API, paste your key, and save. 3. Apply the same credential to the "OpenAI Chat Model2 (GPT-4.1 Mini)" node for result formatting.

3. Configure Google Sheets OAuth2

   1. Go to Google Cloud Console, create or select a project, and enable the Google Sheets API. 2. Navigate to Credentials → Create Credentials → OAuth 2.0 Client IDs, set application type to Web Application. 3. Add the authorized redirect URI required by your n8n credential setup, then copy the Client ID and Client Secret. 4. In n8n, click the "Get Column Info2" node, choose Create New Credential → Google Sheets OAuth2 API, enter the ID and Secret, and complete the OAuth flow.

4. Prepare sheet and set references

   Option A (sample data): Open the author's Sample Marketing Data sheet, make a copy to your Google Drive, and set sharing to "Anyone with the link can view". Option B (custom sheet): Ensure your own sheet has the same link-sharing permission enabled. Then configure nodes: - Get Column Info2: set Document ID to your sheet ID; set Sheet Name to your column metadata sheet (e.g., "Columns") to provide AI with column-letter mappings. - HTTP Request: no manual URL needed; the AI Agent generates it dynamically. Just ensure the sheet sharing permission is correct.

5. Update AI Agent system prompt (required for custom sheets)

   If using your own sheet, update the system message in the "AI Agent3" node: 1. Replace the Google Sheet URL in the prompt with your own sheet address. 2. Update the GID to match your data sheet's gid. 3. Keep the query structure format unchanged. Below is the original system message from the node (source: workflow.json) for reference when replacing:

   Execution prompt · Source text。AI Agent3 node systemMessage field; replace URL and GID when using a custom sheet

   Copy prompt

   # MVP System Prompt — Google Sheets Ask-YOUR_OPENAI_KEY_HERE
   
   Rule, you must always use the column tool for every question and followup question to know the column names and column letters. 
   
   write a sql query to send to google sheets based on the location of the google column on the google sheet. The output should look something like this. 
   
   must include google sheets entire url before the query. do not truncate. "https://docs.google.com/spreadsheets/d/19aUQYZq02qHsCelO4eeV4sx_MTJJupC5qe0gDLQBtRA/edit?gid=365710158#gid=365710158" 
   
   example: 
   https://docs.google.com/spreadsheets/d/19aUQYZq02qHsCelO4eeV4sx_MTJJupC5qe0gDLQBtRA/gviz/tq?tqx=out:csv&gid=365710158&tq=SELECT E, SUM(H), SUM(F) GROUP BY E
   
   You can only use column letters, no names
   ouptut data like this. 
   
   {
   	"query": "https://docs.google.com/spreadsheets/d/19aUQYZq02qHsCelO4eeV4sx_MTJJupC5qe0gDLQBtRA/gviz/tq?tqx=out:csv&gid=365710158&tq=SELECT E, SUM(H), SUM(F) GROUP BY E"
   }

6. Trigger chat and view formatted results

   1. After activating the workflow, send a natural-language question via the Chat Trigger (webhook), e.g., "What was the total sales by channel last month?" 2. AI Agent3 calls Get Column Info2 to retrieve column names and letter mappings, then generates a Google Sheets query URL. 3. The HTTP Request node executes that URL to fetch CSV data. 4. Write into Table Output1 uses GPT-4.1 Mini to format the data into a single table with dimensions first, then metrics.

   Execution prompt · Source text。Write into Table Output1 node systemMessage field

   Copy prompt

   write this into one table. output as dimensions, then metrics

You are reproducing n8n template 7451: natural language to Google Sheets SQL. First confirm you have an n8n instance, an OpenAI API key, and Google OAuth2 credentials. Then import the workflow, configure credentials, prepare the sheet, and replace the URL and GID in the AI Agent system prompt if using a custom sheet.