Evaligo
Blog/Tutorial
Twitter Posts Icon

Website to Twitter Posts Automation

Build an automation that scrapes any website and generates up to 9 engaging Twitter posts with images. Perfect for promoting products, blog posts, and launches without manual content creation.

18 min read•Intermediate•December 26, 2025
Website to Twitter Posts Workflow
9
Twitter Posts Per URL
45s
Processing Time

Two-Layer Architecture

Flow Playground

The crawling and content generation engine. Analyzes websites, extracts key pages, and generates Twitter posts with images.

We'll build this first

App Playground

The user interface. Simple URL input that displays generated posts in an image gallery with titles and bodies.

We'll build this second

Part 1: The Website Crawler & Twitter Generator Flow

This flow crawls a website, identifies the 3 most marketing-relevant pages, and generates up to 3 Twitter posts per page (9 total). Uses GPT-4.1-mini for cost efficiency.

Complete Flow (Left to Right)

API Input
url: string
URL entry
→
Crawler
Site mapping
Discover pages
→
★ Extract URLs
Top 3 pages
AI analysis
→
↻ Split
3 threads
Parallel
→
Fetch HTML
Get page
↓
Extract Text
Clean text
→
⚡GPT-4.1-mini
★ Twitter Posts
3 posts/page
AI content
→
Flatten
Combine all
Merge
→
↻ Iterator
Per post
Loop
→
Dataset (Write)
Save posts
Storage
Data flows left to right. The flow crawls → analyzes → splits → fetches → generates → stores.

Breaking Down Each Node

API Input
Field:
• url (string, required)
What it does:

Receives website URL from the app form.

Example:
{ "url": "https://yourproduct.com" }
Crawler
Settings:
Mode: Auto
Discovers pages
Extracts metadata
What it does:

Crawls the website and maps its structure with page metadata.

Output:

Tree structure of all URLs with titles and descriptions.

★ AI Prompt
Extract URLs
Model: GPT-4.1
Temp: 0.7
Output: Top 3 URLs
What it does:

AI analyzes the crawl tree and picks the 3 most marketing-relevant pages.

Looks for:
✓ Include:
• Product pages
• Blog posts
• Announcements
• Features
✗ Exclude:
• Terms, privacy
• Signup/login
• Utility pages
Output:

Array of 3 URLs with summaries.

↻ Array Splitter
Input: Array of 3 URLs
Creates 3 parallel threads
What it does:

Splits 3 URLs into separate threads. Processes all pages simultaneously.

Sequential: 3 × 15s = 45s
Parallel: ~15s total→ 3x faster!
Fetch HTML
Gets raw HTML for each URL
↓
HTML Text Extractor
Extracts clean text content
What they do:

Fetch HTML → Extract clean text. Removes tags, scripts, styles.

Why:
  • • Raw HTML confuses AI
  • • Clean text = better posts
  • • Lower token costs
Output:

Clean text ready for AI analysis.

⚡GPT-4.1-mini
★ AI Prompt
Twitter Posts
Model: GPT-4.1-mini
Temp: 0.7
Max: 512 tokens
Output: 3 posts/page
What it does:

Creates up to 3 Twitter posts per page. Max 280 chars each. Includes image URLs when available.

Input:
• html Clean text
• url Source page
Output:
• Title
• Body (≤280 chars)
• Image URL (if found)
Key features:
  • • Extracts image URLs from page
  • • 280 character limit (Twitter)
  • • Prioritizes posts with good images
  • • Avoids logo-only images
Array Flatten
Combines nested arrays into single array
What it does:

Combines 3 arrays into one. [[3], [3], [3]] → [9 posts]

Why:

Iterator needs a flat array, not nested ones.

↻ Iterator
Loops through all posts one by one
What it does:

Loops through all 9 posts one by one.

Why sequential:

Dataset writes are fast (~50ms). 9 writes = 0.5 seconds. No need to parallelize.

⚈ Dataset (Write)
Post Collector
Saves posts with title, body, imageUrl
What it does:

Saves all posts to cloud storage.

Schema:
{ "title": "string", "body": "string", "imageUrl": "string" }
Why:
  • • Display in Image Gallery UI
  • • Post library for reference
  • • Ready-to-use Twitter content
  • • History of all generated posts
Flow Summary

Input (URL) → Crawler (map site) → Extract URLs (top 3 pages) → Split (3 threads) → Fetch + Extract (per page) → Generate Posts (3 per page, 9 total) → Flatten → Iterator → Dataset (save all posts)

Time: ~45s
With parallel processing
Output: 9 posts
Ready for Twitter

Part 2: The App Playground (User Interface)

Now that we have the flow built, let's create the UI that lets users input a URL and view generated posts in an image gallery.

Complete App Structure

Input Layer
Text Input
Website URL
↓
Flow Execution Layer
⚡ Flow Execution
Crawler & Generator
Calls Part 1 flow
↓
Display Layer
Dataset Connection
Post Collector
→
Image Gallery
Shows posts

How It Works Together:

1. User Input: User pastes a website URL (e.g., product page or blog post)
2. Flow Execution: Triggers the crawler & generator flow from Part 1
3. Processing: Flow crawls site, analyzes pages, generates 9 Twitter posts with images
4. Storage: Posts are saved to "Post Collector" dataset
5. Display: Dataset Connection feeds posts to Image Gallery component
6. Result: User sees all 9 posts in a visual gallery with title, body, and images
Complete User Journey:
  1. 1. Paste URL: "https://myproduct.com/blog/new-feature"
  2. 2. Click submit button
  3. 3. Wait ~45 seconds (progress indicator shows)
  4. 4. View 9 generated Twitter posts in image gallery
  5. 5. Copy posts and images to share on Twitter

Deployment

REST API

Deploy as an API endpoint

POST api.evaligo.com/app/your-app-id

Embed Widget

Embed the App UI in your website

<iframe src="app.evaligo.com/embed/..." />

Ready to Automate Your Twitter Marketing?

Start building with Evaligo's visual workflow builder. No coding required.

Get Started FreeView Examples
No credit card•Free tier•Deploy in 1-click

Related

Website to Reddit Posts

Generate founder-friendly Reddit content

More Tutorials

Other AI workflow tutorials