Automate Blog Writing with Make.com & Firecrawl Web Scraper

Overview

This tutorial explores how to leverage Firecrawl’s data collection capabilities with Make.com automation workflows and large language models to achieve fully automated blog article writing and SEO.

This solution efficiently scrapes website content, intelligently transforms and generates illustrated Markdown articles:

1. Full Site Scan - Use Firecrawl’s map function to get all sub-links
2. Content Scraping - scrape endpoint extracts clean Markdown content
3. AI Processing - AI model translates, removes promotions, formats
4. Auto-Storage - Save to Notion knowledge base
5. Scheduled Publishing - Daily fixed quantity batch publishing

Complete workflow: Firecrawl scraping → AI processing → Notion storage

---

Core Decision Factors

When choosing AI content automation solutions, focus on:

• Data Collection Capability - Supports single page, full site, and deep sub-link scraping
• Content Processing Transformation - Can perform translation, refinement, formatting, and image-text mixing
• Automation Level - Supports batch processing and scheduled publishing
• Cost & Deployment - Free quota, open-source options, and local deployment capability

---

Technical Specifications

Specification	Value	Notes
Firecrawl Online Free Quota	500 credits/month	~500 pages, usually sufficient for personal needs
Jina Reader API Free Quota	1 million requests	Reference comparison
Map Function Scrape Example	3,553 articles	Zapier blog, extremely fast
HTTP Request Default Timeout	40 seconds	May not be enough
HTTP Request Recommended Timeout	300 seconds	For time-consuming operations
LLM Processing Blog Token	4,096	For longer articles
LLM Processing Social Token	1,000	For short content
API Rate Limit Error Code	429	Need to control request frequency

---

Prerequisites

Before starting, ensure you have:

• Make.com account (free registration)
• Firecrawl API key (firecrawl.dev)
• OpenAI API key
• Notion account and database

---

Firecrawl Core Functions

Firecrawl is a web scraping tool designed for AI, providing two core endpoints:

Map Endpoint - Full Site Scan

Quickly get all sub-links under a website without scraping page by page.

3,553 Zapier blog article links, instantly obtained

Features:

• Extremely fast, completed in seconds
• Returns all sub-link lists
• Supports URL pattern filtering

Scrape Endpoint - Content Extraction

Converts web pages to clean Markdown format, including image links.

Output Content:

• Pure Markdown text
• Preserves image URLs
• Removes ads and navigation elements

---

Workflow Architecture

Complete automation workflow logic structure

Core Modules

1. Basic Trigger - Input target URL
2. HTTP (Map) - Call Firecrawl to get all sub-links
3. Iterator - Traverse link list
4. HTTP (Scrape) - Scrape each page content
5. OpenAI - Translate, remove promotions, format
6. Notion - Save generated articles

---

Step 1: Configure Firecrawl Map Request

Add HTTP module in Make, call Firecrawl’s map endpoint:

Request Configuration:

• URL: https://api.firecrawl.dev/v0/map
• Method: POST
• Headers: Authorization: Bearer {{your_api_key}}
• Body:

{
  "url": "{{target_website_url}}"
}

---

Step 2: Configure Content Scraping

Use Iterator to traverse link list, call scrape endpoint for each:

Request Configuration:

• URL: https://api.firecrawl.dev/v0/scrape
• Method: POST
• Body:

{
  "url": "{{current_link}}",
  "formats": ["markdown"]
}

Note: Set timeout to 300 seconds to avoid complex page scraping timeouts.

---

Step 3: Configure AI Model Processing

Use GPT-4o to process scraped content:

Prompt Design:

Please translate the following English content to Chinese and process as follows:
1. Remove all promotional and traffic-driving content
2. Maintain Markdown format
3. Preserve all image links
4. Generate article suitable for blog publishing

Original content:
{{scraped_markdown}}

Key Parameters:

• Model: gpt-4o
• Max Tokens: 4096

---

Step 4: Smart Deduplication & Storage

Configure Notion search module to avoid re-scraping:

1. Search Notion database for existing URL
2. Use Router to judge: skip if exists, save if not
3. Save processed article to Notion

Note: When saving URLs in Notion, recommend using “Text” type field to avoid errors from non-standard URLs.

---

Step 5: Scheduled Batch Publishing

Implement daily fixed quantity publishing via date formula:

Implementation:

• Add “Publish Date” formula field in Notion
• Set daily publish quantity (e.g., 10 articles)
• Make scheduled task filters articles for that day

This allows:

• Avoid system overload
• Maintain continuous content updates
• Control publishing pace

---

Final Effect

Firecrawl + AI model generated illustrated blog article

After workflow runs, you’ll get:

• Illustrated Markdown format articles
• Auto-translated and de-promoted processing
• All content auto-saved to Notion
• Supports batch and scheduled publishing

---

Important Notes

Things to note during setup:

1. Content Post-Processing - Must clearly instruct AI model to remove original promotional content

2. Markdown Format Fine-tuning - May have extra format marks (like two asterisks) when pasting to editor, need manual deletion

3. Anti-Scraping Mechanisms - Some sites have strict anti-scraping, Firecrawl may fail

4. JSON Structure Sensitive - Extra spaces or unexpected characters may cause HTTP request parsing failures

5. Large-Scale Processing Pressure - When batch processing thousands of articles, recommend batch processing to avoid system overload

6. Prompt Precision - AI model module configuration needs precision, especially JSON output format instructions

Large language model module error example, note prompt configuration

---

Use Cases

Recommended For

• Content Creators - Want to significantly improve creation efficiency, maintain update frequency
• Digital Marketers - SEO article traffic generation through high-quality content
• Industry Researchers - Batch scrape specific domain websites for structured archiving
• Businesses Reducing Labor Costs - Use AI to replace some manual content production

May Not Suit

• Users completely unfamiliar with API configuration
• Users with extremely high originality requirements
• Users needing simple text scraping only
• Users pursuing absolute zero-error zero-intervention

---

FAQ

Is Firecrawl’s free quota enough?

Online version offers 500 credits/month (~500 pages), usually sufficient for personal non-large-scale needs. Also supports open-source self-deployment with no quota limits.

Can it scrape all websites?

Most websites yes, but sites with strict anti-scraping may fail. Recommend small-scale testing first.

What’s the generated article quality?

After AI translation and polishing, articles are illustrated and well-formatted, but core content comes from scraping - note copyright issues.

How to avoid re-scraping same URLs?

Workflow uses Notion search module for smart deduplication, automatically skipping already-collected URLs.

---

Next Steps

After learning the basic workflow, you can try:

• Add more content source websites
• Integrate social media post generation module
• Add custom promotional content insertion
• Set multilingual translation output

Feel free to leave comments if you have questions!