Q24
Many sources, one story.The world, aggregated in Romanian.
/ About the project
The same story, told differently by ten newsrooms, each with its own angle. To really get it, you'd have to read them all, compare and filter. Q24 does that work for you: for every topic, one synthesis drawn from several newsrooms, with the facts laid out in the open. You draw your own conclusion.We built it end to end: aggregation, AI processing, media generation, automatic social media posting. Q24 was designed to run on its own, on a predictable operating budget.
View live/ Project details
Name:Q24, international news platform in Romanian
Client:NextApp Labs
Industry:Media & news
Timeline:Since January 2026
Outcome:Zero manual posts since launch: aggregation, synthesis and social media publishing were built to run without human intervention
Live:q24.ro
/ The challenge
Hundreds of sources, from Romania and abroad, reporting the same event from different angles. Some with a visible editorial agenda, others neutral. Plus big topics that show up fragmented across dozens of separate articles, without the reader realising they're all about the same thing.Three problems that, left unsolved, make the synthesis useless.
/ What we built
Multi-source aggregation
We collect the versions of the same story from Romanian and international press. On a single page you see at once what each newsroom says about the topic.
Editorial balance
For every topic we combine sources with different editorial profiles. The synthesis keeps the shared facts without adopting any single newsroom's angle.
Story clustering
The system recognises when ten seemingly different articles are actually about the same event. It merges them into one topic, so you don't read the same story split in ten.
/ How it works
Q24 continuously follows what newsrooms in Romania and around the world publish. When several sources report the same event, the system groups them and condenses them into a single story.The reader gets a coherent version of the event, with the sources shown for anyone who wants to check for themselves.
/ Tech stack
Next.js & React
A front end built for speed and for Google. New pages appear instantly when a topic is published, get indexed right away and load fast for the reader.
Python backend
For the processing pipeline we chose Python. It has the most mature libraries for working with AI models and handling data at scale, plus direct integration with the external services we rely on.
AI models
The models generate the summaries, identify quotes and extract the key facts from articles. Video content for social media is produced automatically, with no human editor.
