Automatically Crawl Sites with Dumpling AI, Summarize with GPT-4o, and Send Beautiful Newsletters via Gmail

Introduction

Imagine turning your favorite industry blogs, competitor updates, or curated websites into a polished HTML newsletter, fully automated, styled, and summarized, all delivered to your inbox or subscribers.

This n8n workflow does exactly that.

With this system, you can:

Pull website URLs from Google Sheets
Crawl those pages using Dumpling AI
Clean, extract, and format article content
Summarize and repackage the info using GPT-4o into a newsletter
Send a fully formatted email through Gmail — no writing, no copy-paste

Perfect for:

Weekly internal team updates
Curated content for subscribers
Competitive intelligence digests

Let’s break it down, step by step.

Step 1: Start the Workflow Manually

Node: Start Workflow Manually

This lets you trigger the automation by hand, ideal for testing or running on demand. You can later swap this out with a scheduler or webhook if needed.

Step 2: Fetch Website URLs from Google Sheets

Node: Get Website URLs from Google Sheet

This node reads a list of URLs from a Google Sheet where you’ve curated your sources.

What to configure:

Google Sheets OAuth2 credentials
Document ID and Sheet Name

Expected Output:
A list of URLs under a column named something like websites.

Pro Tip:
Use this Google Sheet as your source control panel. You can update links, remove low-quality sources, or prioritize domains by row order.

Step 3: Crawl Pages Using Dumpling AI

Node: Crawl and Extract Site Content with Dumpling AI

This step uses Dumpling AI’s crawl API to extract readable content from each website.

Request Payload:

{

“url”: “{{ $json.websites }}”,

“limit”: “5”,

“depth”: “2”,

“format”: “text”

}

What it returns:
An array of article-like entries, each including:

metadata.title
metadata.original_url
content (cleaned text)

Why this is powerful:
Dumpling AI handles complex scraping logic, content cleaning, and structure parsing ,so you don’t have to.

Step 4: Split Results into Individual Articles

Node: Split Extracted Results into Individual Items

The previous Dumpling response is an array. This step separates each item so it can be processed one-by-one.

Field to split: results

Why this matters:
Working with individual items is cleaner and lets us reformat them easily in the next steps.

Step 5: Extract and Map Article Fields

Node: Map Title, URL, and Page Text

This Set node captures only the fields we care about:

Title of the article
Page text
Original URL

This trims excess metadata and gets everything ready for formatting.

Mapped Fields:

metadata.title
metadata.original_url
Content

Step 6: Combine All Articles Into a Single Prompt

Node: Combine Articles into Single Prompt Format

A simple JavaScript Code node formats all items into one single long-form string, structured for GPT-4o to read easily.

JS Logic:

Numbers each article
Lists title, URL, and content
Adds spacing for readability

Output Example:

1. How to Build with AI

https://example.com

Long form content…

2. Why Automation Matters

https://another.com

Step 7: Summarize and Generate HTML Newsletter with GPT-4o

Node: Generate HTML Newsletter with Subject Using GPT-4o

This is the heart of automation.

Prompt Logic:

GPT-4o reads all articles
It creates a catchy subject line
Then for each article, it:
- Adds an <h3> with the article title
- Writes a 2–3 sentence summary in a <p>
- Adds a clickable “read more” link
- Separates entries with <br/>

Required Output Format:

{

“subject”: “Catchy newsletter title”,

“body”: “<html content with summaries>”

}

Why this step is powerful:
You’re not sending raw article dumps — you’re sending clean summaries, with links, in a structured, branded format.

Step 8: Send the Newsletter via Gmail

Node: Send Newsletter via Gmail

Finally, the newsletter is sent using your Gmail account.

Mapped Fields:

subject: From GPT
message: HTML content created by GPT
recipient: Can be your own inbox for review or a distribution list

Pro Tip:
You can duplicate this node to send to multiple team members or integrate it with Mailchimp or SendGrid later.

Workflow Summary

Here’s the full automation in action:

Start manually
Pull URLs from Google Sheets
Crawl each page using Dumpling AI
Extract article title, content, and link
Format all into a combined prompt
GPT-4o summarizes each item and generates HTML
Send it as an email newsletter

Conclusion: The Newsletter You No Longer Have to Write

With this automation in place, your newsletter becomes:

Consistent (runs on your schedule)
Well-structured (HTML, links, formatting handled)
Smartly summarized (GPT-4o captures key insights)
Beautifully delivered (straight to inbox)

Whether you’re reporting company trends, content marketing recaps, or competitor coverage ,this system saves hours per week while keeping your brand informative and relevant.