n8n Automation Engineer Task – Analytos

This repository contains workflows for lead qualification automation using n8n.

🎥 Demo Video

Click the thumbnail above to watch the full demo video on YouTube.

---

Workflows

1. workflow.json

   • End-to-end workflow for lead qualification:
     • Trigger: Gmail new email
     • Categorization: OpenAI (High/Medium/Low) via reusable subflow
     • High Intent → Google Sheet (High Intent tab)
     • Medium/Low → Gmail reply + Google Sheet (Low/Medium Intent tab)
     • Discord notifications for all leads
     • Retry logic enabled on all API nodes (OpenAI, Gmail, Google Sheets, Discord) with up to 3 retries and 1s wait between retries

2. lead-categorization-subflow.json

   • Reusable subflow that takes a lead message as input and returns:

     {
       "category": "High",
       "confidence": 0.92
     }

   • Can be called from any workflow using a Subflow node
   • Confidence score represents the AI's certainty (0 = low, 1 = high)
   • Retry logic applied to OpenAI node

3. subflow-wrapper.json

   • Minimal workflow to test the subflow independently
   • Trigger: Webhook
   • Calls the subflow and returns category + confidence

---

Setup Instructions

1. Connect your Google account in n8n to enable Gmail and Google Sheets operations.

2. Create a Google Sheet document to store lead data.

3. Create two tabs (sheets) inside the document:

   • High Intent → for leads categorized as High Intent
   • Low/Medium Intent → for leads categorized as Medium or Low Intent

4. Replace placeholders in workflows with your credentials:

   • YOUR_OPENAI_API_KEY → for OpenAI connection
   • YOUR_DISCORD_WEBHOOK_URL → for sending notifications to Discord

5. Import workflows into n8n and test:

   • Trigger via Gmail new email or via webhook
   • Confirm leads are categorized correctly
   • Verify Google Sheet updates, Gmail replies, and Discord notifications

---

Notes

• Retry logic ensures that temporary network/API failures don’t break the workflow.
• The subflow can be reused in multiple workflows or tested independently via webhook.
• The AI categorization returns a confidence score between 0 and 1 to indicate certainty.