LavX Managed Systems always has a corner for experiments. We use that space to try new technologies, validate ideas in real conditions, and learn what actually works when people use it. That is exactly what news.lavx.hu (LavX News) is.
LavX News: a tech demo that grew up
LavX Managed Systems always has a corner for experiments. We use that space to try new technologies, validate ideas in real conditions, and learn what actually works when people use it.
That is exactly what news.lavx.hu (LavX News) is.
The first version: a one year beta built to prove the idea
The original news.lavx.hu ran as a Beta for about a year. It was a tech demo that focused on integration and workflow, not on polished visuals.
It combined microservices with different kinds of LLMs and supporting systems, and it already had a clear goal: handle the full path from discovery to publishing.
Beta stack
- Python
- PostgreSQL
- Redis
What the beta proved
- A workflow that processes news from sources, prepares it for readers, and pushes it out to social channels
- A working integration between services and automation components
- A baseline that showed where the real bottlenecks and edge cases are

December 2025: rewritten from scratch
In December 2025 we rebuilt the whole project from zero. This was not a cleanup pass. It was a rewrite with the goal of moving from “beta lab” to something stable enough to run like a product.
What changed
- We shifted from Python to TypeScript and started using more mature building blocks
- We kept PostgreSQL and Redis
- The workflow evolved into a more controlled and robust system
- We started generating images for articles to improve the reading experience
The design and layout also moved closer to what you expect from an online magazine: clearer structure, better hierarchy, and a smoother browsing experience.

From beta to production: what “grown up” meant for us
The biggest difference between beta and production was not the language switch. It was how we treated reliability and control.
In beta, it was enough that the pipeline worked most of the time. In production, “most of the time” is not acceptable.
So we focused on things like:
- More predictable processing steps with clear states and fewer surprises
- Better guardrails around what gets published and when
- More robust handling of failures, retries, and odd inputs from sources
- A workflow that is easier to operate and easier to improve without breaking it
In short, we kept the experimental spirit, but we stopped treating it like an experiment in day to day operation.
Why we run LavX News, and why it is free
We still treat the site as a tech demo platform, because it lets us try new workflows in a real environment. The difference is that it is no longer beta software. It is robust enough to be a product.
At the same time:
- We are not a publisher
- We do not want heavy moderation
- We want a clean reading experience
That is why the content is free, with no advertisements, no paid articles, and no sponsored placements.
If you find something odd, abusive, or incorrect, tell us and we will gladly correct it.
And if you like the site, we would love it if you register, comment, and share articles you find useful.

Comments
Please log in or register to join the discussion