<aside>

Goal

Automate personalized email responses by combining contact data lookup with AI-generated replies, eliminating manual email handling while maintaining authentic, context-aware communication.

Solution

• Gmail trigger monitors incoming emails every minute
• Contact lookup retrieves customer history and notes from n8n Data Tables
• Claude AI generates personalized HTML email responses based on contact context
• Python parser extracts structured subject and body from AI output
• Automated Gmail reply sends professional formatted response
• Smart filtering only responds to known contacts, preventing spam auto-replies

Features

• Context-aware responses: Uses contact notes and history for personalization
• HTML email formatting: Professional, clean email structure
• Selective automation: Only responds to contacts in database
• Claude Sonnet 4.5 integration: Latest AI model for intelligent responses
• Structured output parsing: Reliable extraction of subject and body
• Real-time monitoring: 1-minute polling interval

Steps

1. Email Monitoring

Node: Gmail Trigger

Purpose: Monitors Gmail inbox for new incoming emails.

Configuration:

• Poll Time: Every minute
• Event: New email received
• Output: Email data including sender, subject, body

2. Contact Lookup

Node: Get row(s) (Data Table)

Purpose: Searches for sender email in Contacts database.

Configuration:

• Data Table: Contacts
• Operation: Get
• Condition: Email = sender's email address
• Must Match: All Conditions

Output: Contact record with Name, Email, Phone, Notes (if exists)

Design Decision: If contact not found, workflow stops here. Only known contacts receive automated responses to prevent spam-like behavior.

3. AI Response Generation

Node: Message a model (Claude)

Purpose: Generates personalized email response using contact context.

Configuration:

• Model: Claude Sonnet 4.5 (claude-sonnet-4-5-20250929)
• Input: Contact data + received email

User Prompt:

You are an AI assistant helping to write professional email responses.

Create a well-formatted HTML email response based on the customer information and received email.

Customer Data:
1-Name: {{ $json.Name }}

2-Notes: {{ $json.Notes }}

Received Email:
1-Email Subject: {{ $('Gmail Trigger').item.json.headers.subject }}

2-Email Body: {{ $('Gmail Trigger').item.json.text }}

CRITICAL: Your response MUST follow this EXACT format:

Subject: [Your subject line]

HTML_BODY_START
<p>Hi [Name],</p>

<p>[First paragraph with warm acknowledgment]</p>

<p>[Main content addressing their needs]</p>

<p>[Next steps or questions]</p>

<p>Best regards,<br>
Alvaro</p>
HTML_BODY_END

Rules:
- Use <p> tags for paragraphs
- Use <br> for line breaks
- Keep it clean and professional
- The HTML_BODY_START and HTML_BODY_END markers are REQUIRED

4. Response Parsing

Node: Code in Python (Beta)

Purpose: Extracts structured subject and HTML body from Claude's response.

Logic:

• Uses regex to extract Subject line
• Finds HTML content between HTML_BODY_START and HTML_BODY_END markers
• Fallback: If no HTML found, converts plain text with <p> and `` tags
• Returns JSON with subject, body, and body_type

Why needed: Gmail node requires separate subject and body fields

5. Email Delivery

Node: Send a message (Gmail)

Purpose: Sends formatted HTML email response to original sender.

Configuration:

• To: Original sender's email
• Subject: Parsed subject from Claude
• Body: Parsed HTML body
• Options: Attribution disabled for clean signature

Tools Used

• n8n: Workflow automation and orchestration
• Claude Sonnet 4.5: AI model for response generation (via Anthropic API)
• Gmail API: Email monitoring and sending
• Data Tables: Contact database storage
• Python: Response parsing and text processing

Impact

• Zero manual email responses for known contacts
• Context-aware personalization using contact history and notes
• Professional HTML formatting in all responses
• Smart filtering prevents spam auto-replies to unknown senders
• Sub-2-minute response time from email receipt to reply
• Scalable architecture can handle multiple contacts simultaneously

Learnings

• Prompt engineering for structured output: Using markers like HTML_BODY_START ensures reliable parsing
• Edge case handling: Workflow intentionally stops for unknown contacts
• Claude integration best practices: System prompts with clear format requirements
• Python parsing in n8n: Regex patterns for extracting structured data from text
• Production-ready automation: Thoughtful design prevents unwanted auto-replies

Database Schema

Contacts Table:

• Name (string)
• Email (string) - used for lookup
• Phone_Number (string)
• Notes (string) - context for AI personalization

Sales Data Analysis Agent

This workflow also includes a conversational AI agent for analyzing sales data using natural language queries.

Architecture

Trigger: Chat interface (When chat message received)

AI Agent: Sales Data Agent with Claude Sonnet 4.5 (via OpenRouter)

Tools Available:

1. All Rows - Returns complete sales dataset
2. Product Name Query - Filters by product (Wireless Headphones, Bluetooth Speaker, Phone Case)
3. Date Query - Filters by date (YYYY-MM-DD format)
4. Product ID Query - Filters by ID (WH001, BS002, PC003) </aside>

📂 Download & Explore the Workflow:

Smart Gmail Auto-Reply with Claude.json

View GitHub Repository:

Smart-Gmail-Auto-Reply-with-Claude-n8n/Smart Gmail Auto-Reply with Claude.json at main · AlvLeoAI/Smart-Gmail-Auto-Reply-with-Claude-n8n