Every day, thousands of news items are published across the world, and no single person can read them all. LAS News exists to make that flow easier to follow for readers in Nigeria, Africa and beyond. It does this with an automated pipeline that gathers reports from public sources, groups them by story, writes an original summary, and credits the sources it drew from. This article explains, in plain terms, how that pipeline works and where its limits are.
Step one: gathering from public feeds
The pipeline starts with RSS. Most news publishers offer an RSS feed, which is a machine-readable list of their latest headlines, links and short descriptions. Publishers create these feeds so that other services can discover and link to their work. LAS News keeps a list of feeds across its sections: Top stories, Entertainment, Sport, Games, Tech and Politics. On a schedule, the pipeline reads those feeds and collects the newest items.
Starting from feeds, rather than scraping whole web pages, is a deliberate choice. A feed is the publisher's own invitation to be read by software. It tells us the headline, the link and when the item was published, and it keeps us pointed back at the original reporting.
Step two: grouping reports into stories
When something important happens, many outlets report it at once. If the pipeline treated each report as separate, the site would fill up with near-duplicates. So the next step is clustering: items that appear to cover the same event are grouped together. A cluster might contain a report from one outlet and two others that add detail. Working from several sources about the same event also helps, because a claim that appears in only one place can be treated with more care.
Step three: writing an original piece
For each cluster, an AI language model is asked to write a new article in its own words, based on the material gathered. The goal is a short, clear story that a reader can understand without clicking through five different sites. The instruction is not to paraphrase a single article sentence by sentence, but to summarise what the sources together say.
Language models can make mistakes, which is why writing is only part of the process and not the end of it.
Step four: quality checks
Before a story is published, it goes through automated checks. These include checks on length, so that very thin pieces are not published, and checks for copying, so that long runs of words are not lifted from a source. There are also checks on quotation, so that quoted material is handled carefully rather than invented. Stories that fail are not published. There are also limits on how many new articles and updates can be produced in one run, and how many can appear per category, so that one busy topic cannot crowd out everything else.
Step five: crediting sources
Every story on LAS News is meant to point back to the reporting it is based on. Source credit matters for two reasons. First, fairness: the journalists and outlets that did the original work deserve to be named and linked. Second, trust: a reader who doubts a summary should be able to go straight to the source and judge for themselves. We would rather send a reader to the original report than ask them to take our word for it.
Step six: publishing every 30 minutes
The site is rebuilt and published automatically every 30 minutes. That rhythm is a compromise. It is frequent enough that readers see developing stories within the hour, and slow enough that each run can be controlled and checked. Because the output is a static site served from Cloudflare, pages load quickly even on slower mobile connections, which matters to many of our readers.
What automation cannot do
It is worth being honest about the limits. An automated system does not attend events, interview people or verify claims on the ground. It summarises what others have already reported. If a source is wrong, a summary can inherit that error. That is why LAS News is best used as a way to follow the news and then follow the links, and not as a replacement for primary reporting.
This is also why we built the Eyewitness tab. Eyewitness lets readers submit reports from where they are, adding a human, first-hand layer that an automated pipeline cannot supply. You can read more about how to use it in our FAQ.
How to read LAS News well
- Check the source links under a story, especially for big or surprising claims.
- Compare how the story is told elsewhere when it matters to you.
- Treat developing stories as unfinished: details can change within hours.
- Tell us when something looks wrong, so it can be looked at.
LAS News is owned and operated by LAS G Concept, and you can see how it fits with our other products on the how it works page. If you want to read it for yourself, start at news.lasgconcept.com.
Written by Lawal Ayo. LAS G Concept is a Nigerian technology and media company. Questions? See the FAQ or contact us.