Sportopod: a news site written by our LLM layer

A sports-news site we built to show our LLM layer and smart router at work. Since July 2026 it stands as a frozen demo archive.

We wanted a hard, public test for the AI plumbing we build for clients, so we gave it a newsroom. News is fast, comes in several languages, and a wrong article is visible to everyone. What you see on the site is what came out.

Sportopod home page: the lead story and the trending stories, each with its photoA Sportopod article: the story, why it matters, the questions readers ask, and its sourceAnother Sportopod article, built the same way from its source
A Sportopod article on a phoneAnother Sportopod article on a phoneSportopod home page on a phone

Page by page

  • A Sportopod article: the story, why it matters, the questions readers ask, and its source
  • Another Sportopod article, built the same way from its source

On every screen

  • Sportopod home page on a phone
  • A Sportopod article on a phone
  • Another Sportopod article on a phone

Twelve sports in one place

A reader found scores, standings, trade rumors, injuries, recaps, previews, rankings and opinion for 12 sports. It was never meant to be a media business.

One event, one article

When many outlets wrote about the same event, in several languages, the reader got one article: with its source named, a short "why this matters" and the questions a reader would ask.

Nothing went straight to the page

Our router chose the model for each task by itself, and every answer was checked against a schema before it was published.

One codebase, six branded sites

The same codebase ran six branded sites. The site published a sitemap, a news sitemap, a feed and an llms.txt, so search engines and AI assistants could read it. It is the kind of layer we build under client products that put AI to work on real content.

Frequently asked questions

What is it built with?

The pipeline runs on PostgreSQL with pg_cron and pgvector. The models are reached through our own LLM gateway, written in TypeScript on Fastify 5.

What does "checked against a schema" mean?

Every answer from a model had to arrive in an agreed structure: headline, summary, body, source and so on. An answer that did not fit that structure was not published.

Can you build this kind of pipeline for us?

Yes. The gateway, the router and the checks are what we reuse. Tell us what content or data you need processed and we will say honestly what it takes.

Get in touch

Tell us what you want built. We tell you where to start and what it takes.

One conversation. A clear plan.

You leave with

  1. A clear plan for the build
  2. An honest answer on cost and time
  3. Concrete next steps

ByFounder and Imaginator

Pineido, based in Strumica, North Macedonia, is a software company that builds websites, online shops, web and mobile apps, business automation and AI for companies at home and abroad. Work ships in 30-day sprints at a fixed price, and the client owns the code.

Published