AI Tech Blog Harvester
A daily AI content automation pipeline that monitors technology news through RSS, checks existing portfolio blog posts, selects a new article, fetches the source page, extracts the original image, generates bilingual English/Arabic content with AI, validates structured JSON output, and publishes directly to a protected Next.js blog API.
Core Tech Stack
- n8n
- OpenAI
- RSS
- Next.js API
- JavaScript
Performance
- Runs on a daily schedule instead of polling continuously
- Stops early when no new article is available
- Checks duplicates before calling the AI model
- Builds compact JSON payloads for API publishing
- Uses source metadata extraction before generation to reduce manual review
Security
- Protected blog publishing API
- No public exposure of API keys or webhook secrets
- Validation layer prevents malformed AI output from being published
- Server-side workflow execution on a private n8n instance
- Source URL and image validation before publish
Key Features
Daily scheduled content automation workflow
RSS feed monitoring for technology news
Existing blog check through a custom Next.js API
Duplicate prevention by generated slug and title comparison
Source article HTML fetching for richer context
Open Graph and Twitter image metadata extraction
AI-generated bilingual English and Arabic blog content
Strict JSON-only model output requirement
Payload validation before publishing
Protected API publishing using server-side headers
Automatic tags, excerpts, titles, and HTML article bodies
Production execution history with successful daily runs
Project Architecture
Daily 10AM Trigger ├── TechCrunch RSS reads latest source items ├── Get Existing Blogs calls portfolio API ├── Select New Article prevents duplicates ├── Fetch Source Article retrieves source HTML ├── Extract Original Image reads metadata ├── Content Writer generates bilingual JSON ├── Build Blog Payload validates and normalizes data └── Push to Blog publishes to the Next.js API
Engineering Strategies
- Used RSS as the discovery layer and the portfolio API as the source of truth for duplicates
- Generated deterministic slugs from article title and source data
- Validated required AI fields before sending any content to production
- Required source image extraction to avoid publishing weak visual content
- Kept AI output constrained to JSON for reliable downstream processing
- Separated article discovery, writing, validation, and publishing into clear workflow stages
- Used a protected API endpoint instead of direct database writes
Visual Interface
