Pentagon Defence News is a live news site covering World, Business, Tech, and Lifestyle stories, publishing on a rolling basis throughout the day. Most of its posts are created and published automatically — CodeixLab built and runs the n8n pipeline behind it, and the site is a real example of what a production content-automation workflow looks like once it’s past the demo stage.
The Problem We Were Solving
A news site that wants continuous, multi-category coverage across World, Business, Tech, and Lifestyle either needs a large editorial team publishing around the clock, or a pipeline that can reliably source, structure, and publish content without a human touching every article. Pentagon Defence News needed the second option — a system that could run unattended and still produce a coherent, correctly categorized site.
How The Pipeline Is Structured
The workflow polls source content on a schedule, checks each item against what’s already been published to avoid duplicates and near-duplicates, and routes it through category-matching logic that maps incoming stories to the site’s taxonomy (World, Business, Tech, Life Style, and their subcategories like Games and Travel). Each item is formatted to match the site’s expected post structure — title, body, featured image, category, and byline — before being pushed into WordPress through its REST API. Published posts on the site are attributed to the automation account, which is a deliberate choice: it keeps automated and human-authored content clearly distinguishable in the WordPress admin.
Where The Real Engineering Work Was
The visible part of a pipeline like this — fetch, format, publish — is the easy 20%. The work that actually makes it production-grade is everything around that: handling source feeds that go stale or change format without warning, catching near-duplicate stories (the same event covered with different headlines) that a simple URL check would miss, respecting rate limits on both the source and the WordPress API, and making sure a single malformed item doesn’t stall the entire run. Each of these needed explicit handling in the workflow rather than being left to fail silently.
Monitoring an Unattended Pipeline
Because nobody is watching every run in real time, the pipeline logs what it fetched, what it filtered out and why, and what it published on every execution, with alerting if a source stops returning data or the publish step starts failing. That visibility is what turns “automation that mostly works” into automation you can actually rely on.
The Broader Pattern
Nothing about this pipeline is specific to news — the same shape (source, dedupe, structure, publish, monitor) applies to any business that needs a steady stream of structured content going into a CMS, whether that’s a content site, a product blog, or an internal knowledge base. If a similar pipeline would save your team real editorial time, that’s a conversation worth having with our AI automation and integration team.