Automate Content Creation with Make.com, Exa AI & ChatGPT

Overview

This tutorial explores an automated content creation workflow built with Make.com, Exa AI, ChatGPT, and Pexels.

The workflow automatically generates illustrated original articles based on trending topics or custom themes:

1. Topic Acquisition - RSS trending headlines or Notion custom topics
2. Material Retrieval - Exa AI real-time searches web information
3. Content Generation - ChatGPT integrates materials to generate articles
4. Image Matching - Pexels gets free copyright-free images
5. Auto-Save - Complete articles stored in Notion

Workflow final output: Illustrated Markdown format article

---

Core Decision Factors

When choosing AI content creation solutions, focus on:

• Automation Level - Automation from topic selection to publishing
• Content Timeliness - Can it get latest, most relevant information
• Retrieval Quality - Are information sources clean and high-quality
• Platform Integration - Connection capability with mainstream content platforms

---

Technical Specifications

Specification	Value	Notes
Article Generation	Can generate hundreds	High-quality original illustrated articles
Article Length	1,000-2,000 words	Varies by keyword
Content Source	RSS trending/Notion topics	Flexible selection
Exa AI Data Volume	Up to 1 million	Can collect massive data
Exa AI Update Frequency	Per minute	Real-time scraping of latest info
Exa AI Pricing	$5/1000 searches	Each returns 3-5 results
Image Source	Pexels	Free copyright-free library
Make.com Module Count	16	Complete end-to-end workflow
Article Format	Markdown	Easy multi-platform publishing

---

Prerequisites

Before starting, ensure you have:

• Make.com account (free registration)
• Exa AI API key (exa.ai)
• OpenAI API key
• Pexels API key (pexels.com, free)
• Notion account and database

---

Exa AI Core Advantages

Exa AI: Search engine designed for AI

Exa AI is the workflow’s core, solving ChatGPT’s knowledge timeliness limitations.

Difference from Traditional Search Engines

Exa AI vs Google/Baidu technical differences

Feature	Exa AI	Traditional Search Engines
Design Purpose	Designed for AI	Designed for humans
Search Method	Semantic Neural Search	Keyword matching
Data Cleaning	Auto-filter SEO spam	May include low-quality content
Update Frequency	Per minute	Irregular updates
Return Format	Clean text, AI-friendly	HTML pages

Core Technical Features

• RAG Technology - Extends knowledge sources to entire web for retrieval integration
• Semantic Search - Transformer-based architecture, understands query intent
• Data Cleaning - Auto-filters SEO-generated spam content
• Multiple Filters - Domain, date, data category precise filtering

---

Pexels Library Introduction

Pexels provides massive free copyright-free material library:

• Massive Materials - Images and videos covering all industries
• Free Access - No payment needed, quick API integration
• Keyword Search - Get related visual materials via keywords

Note: For highly specific keywords (like “H100 model”), Pexels may only match broader terms (like “GPU”, “innovation”).

---

Workflow Architecture

Complete workflow contains 16 Make.com modules:

1. RSS/Notion Trigger - Get article topic
2. Exa AI Retrieval - Search related materials
3. ChatGPT Processing - Integrate materials, extract keywords
4. Pexels Search - Get matching images
5. ChatGPT Generation - Generate complete article
6. Text Parsing - Handle empty image placeholders
7. Notion Save - Store articles and notes

---

Step 1: Configure Exa AI Retrieval

API configuration interface and parameter description

Add HTTP module in Make to call Exa AI:

Request Configuration:

• URL: https://api.exa.ai/search
• Method: POST
• Headers:
  • x-api-key: {{your_api_key}} (Must be in header!)
  • Content-Type: application/json

Request Body:

{
  "query": "{{article_topic}}",
  "numResults": 5,
  "contents": {
    "text": true
  }
}

Common Error: Incorrect API Key placement causes 401 unauthorized errors.

---

Step 2: Configure ChatGPT Content Generation

Use ChatGPT to process Exa AI returned materials:

First Call - Extract Keywords:

Please extract 3-5 English keywords suitable for image search from the following materials:

{{exa_results}}

Second Call - Generate Article:

Please generate a 1500-word article based on the following materials:

Requirements:
1. Markdown format
2. Include title, introduction, body, conclusion
3. Insert image placeholders at appropriate positions
4. Language easy to understand

Materials: {{exa_results}}
Image links: {{pexels_images}}

---

Step 3: Configure Pexels Image Search

Add HTTP module in Make to call Pexels API:

Request Configuration:

• URL: https://api.pexels.com/v1/search?query={{keywords}}&per_page=5
• Method: GET
• Headers: Authorization: {{pexels_api_key}}

---

Step 4: Handle Empty Image Issues

If Pexels finds no matching images, article will have empty placeholders.

Add Text Parser module, use regex to delete empty images:

!\[\]\(\)

This ensures final article cleanliness.

---

Step 5: Save to Notion

Save generated article to Notion database:

• Title - Article title
• Body - Markdown content
• Social Post - Simplified version content
• Source Topic - Original keywords

---

Error Handling Tips

Use error handling assistant to quickly locate issues

Make.com has built-in error handling assistant (top right star icon) that helps:

• Analyze API call error causes
• Provide solution suggestions
• Quickly locate configuration issues

Common Errors:

1. API Key placement error
2. Missing required Headers
3. Incorrect JSON format
4. Insufficient permissions

---

Final Effect

Processed final article with correctly embedded images

After workflow runs, you’ll get:

• 1000-2000 word Markdown format article
• Automatically illustrated image-rich content
• Simultaneously generated social media posts
• All content auto-saved to Notion

---

Important Notes

Things to note during setup:

1. API Key Position - Exa AI’s x-api-key must be in header, not query string

2. Image Matching Accuracy - Pexels has limited matching for specific keywords, may need manual optimization

3. Empty Image Handling - Need additional Text Parser module to delete empty image placeholders

4. Workflow Complexity - 16 module configuration is complex, need understanding of arrays, collections, etc.

5. Function Usage - Need to master functions like ifempty to handle null values

---

Use Cases

Recommended For

• Content Operators - Need efficient batch generation of original articles
• Make.com Users with Foundation - Can understand multi-module configuration and API calls
• Creators Pursuing Content Timeliness - Exa AI real-time retrieval ensures latest info
• Users Needing Systematic Content Management - Notion knowledge base for easy management and reuse

May Not Suit

• Complete Make.com beginners
• Users with extremely high image matching accuracy requirements
• Users unwilling to invest time learning and maintaining workflows

---

FAQ

How does Exa AI differ from traditional search engines?

Exa AI is designed for AI with semantic Neural Search understanding query intent, per-minute data updates, and automatic SEO spam filtering, returning clean high-quality information.

What’s Exa AI’s pricing?

About $5/1000 searches, each returning 3-5 results, pay-as-you-go with good value.

How accurate is image matching?

Pexels is a free library. For specific keywords (like H100 model), it may only match broader terms, so image accuracy may vary.

How many modules in the workflow?

Complete workflow contains 16 Make.com modules. Configuration is complex, requiring automation foundation.

---

Next Steps

After learning the basic workflow, you can try:

• Integrate more content sources (Weibo trending, Zhihu hot list, etc.)
• Add AI image generation module to improve image matching
• Set scheduled triggers for daily auto-updates
• Add multilingual translation functionality

Feel free to leave comments if you have questions!