AI-Generated LinkedIn Posts with Human Approval and Auto-Publish

Input: daily topics from Google Sheets → Process: GPT-4.1 generates caption → GoToHuman human review → Output: Blotauto publishes to LinkedIn

Who

Creators, marketers, and agencies who want to automate LinkedIn content publishing while maintaining human quality control.

You get

Automatically pull daily topics from Google Sheets, generate emoji-rich LinkedIn captions with AI, approve via human review, and auto-publish.

You need

n8n · google-sheets · gotohuman · blotato

Steps

1. Get Today’s Topic from Google Sheets

   Configure Google Sheets OAuth2 credentials in n8n and authorize your Google account. Copy the template spreadsheet (https://docs.google.com/spreadsheets/d/1oC3dbRkGF_bSIKuFKtSJsYS-ZypT2uk7jPsAni8y154/copy) and fill in topics by date. The workflow triggers daily via Schedule Trigger (or Manual Trigger for testing). A Code node generates today’s ISO date (e.g., 2025-07-30). A Merge node filters rows where the Date column matches today, outputting the idea field for the next step.

2. Generate Caption with OpenAI

   Add your OpenAI API credential in n8n, selecting model gpt-4.1 or similar. Pass the idea from the previous step into an AI Agent node, which uses a system prompt requesting a short, emoji-rich LinkedIn caption of 1–2 sentences. The AI Agent connects to a Structured Output Parser (JSON Schema example: [{"Caption":""}]) to extract the caption. After generation, the caption is written back to the Google Sheets caption column, and the complete column is updated to "Yes".

   Execution prompt · Source text。AI Agent node systemMessage field

   Copy prompt

   =your job is to expand on the idea and turn it into a social media post that will capture users attention. keep it to two sentances using emoji's
   
   Outupt like this. 
   
   [
     {
       "Caption": ""
     }
   ]

3. Human Review via GoToHuman

   Sign up at gotohuman.com and create a Review Template for content approvals. Add your GoToHuman credential in n8n and enter the reviewTemplateId in the workflow node. The generated caption is sent to the "Ask Human for approval" node, where the reviewer can Approve or Reject. A Filter node checks whether the response field equals "approved"; only approved captions proceed.

4. Publish to LinkedIn via Blotato

   Register at blotato.com to obtain your API Key and Account ID. In the n8n HTTP Request node, configure: URL https://backend.blotato.com/v2/posts, method POST, Header blotato-api-key, and JSON Body containing account info and caption content. The caption references the prior node via expression: {{ $('Ask Human for approval').item.json.responseValues.text.value }}. Once configured, approved captions are automatically published to LinkedIn. Note: the actual JSON Body in workflow.json is wrapped in an outer "post" object and uses the placeholder "your account id", while sticky-notes.md shows "YOUR_ACCOUNT_ID" — the two source files differ; the copy block below is verbatim from workflow.json. Replace it with your Blotato Account ID before use.

   Execution prompt · Source text。HTTP Request node JSON Body field

   Copy prompt

   ={
     "post": {
       "accountId": "your account id",
       "target": {
         "targetType": "linkedin"
       },
       "content": {
         "text": "{{ $('Ask Human for approval').item.json.responseValues.text.value }}",
         "platform": "linkedin",
         "mediaUrls": []
       }
     }
   }

You are an n8n workflow setup assistant. The user wants to replicate a template that generates LinkedIn captions with AI, sends them for human approval, and auto-publishes. First confirm the user has: 1) n8n instance access; 2) Google Sheets OAuth2 credential and a copy of the template sheet; 3) OpenAI API key; 4) GoToHuman account and reviewTemplateId; 5) Blotato API key and Account ID. Then guide them to import the workflow, configure credentials, test with Manual Trigger, verify node outputs, and finally activate the Schedule Trigger. Important: the AI Agent systemMessage and HTTP Request JSON Body must match the source exactly; do not alter variable names.