ntv.de & ML6 - Revolutionizing digital journalism
Impact
ntv.de, an RTL brand, a leading German news portal, collaborates with ML6 to optimise its editorial process. By strategically deploying machine learning, they enhance workflows and improve the quality of content. This leads to greater efficiency, considerable time savings for editors, and an increase in daily news production. Experience the significant transformation as ntv.de leads the way in the evolution of modern journalism.
Intro to the customer
Ntv.de is the digital news offering that belongs to the broadcasting channel ntv, a brand of RTL Deutschland. ntv.de is one of Germany’s most visited digital news offerings holding one Grimme Online Award and has already been awarded Germany’s biggest news portal.
‍
Together with its news portal ntv.de operates the ntv Nachrichten app, one of Germany’s most downloaded News apps (Top 10 downloaded News apps Germany, iOS & Android, April 2024)
‍
The editorial office of ntv.de is located in Berlin. In this newsroom the journalists create self-researched news as well as articles based on news agency imports from dpa, Reuters, AFP & sid.
Challenge
The media industry faces significant challenges. Traditional journalism is being disrupted by the dynamics of social media and the evolving news consumption habits of younger generations. Outlets like ntv.de are striving to secure their spot in Germany's news landscape by adapting and expanding their digital offerings and operations for this new era.
‍
For ntv.de, the challenge is to guarantee the speed and quality of their news production to maintain their standing and continue growing their advertising revenue.
Solution
To streamline their publication timeline and enhance article quality, ntv.de has teamed up with ML6 to complete the editorial production process using Large Language Models (LLMs).
Through intensive collaboration and joint efforts, the project team—which includes editors, journalists, and machine learning specialists—developed a solution that combines journalistic expertise with cutting-edge machine learning automation. This system employs various fine-tuned LLMs. Initially, incoming news articles from and in collaboration with the German News agency dpa are assessed for their relevance according to rules previously defined by ntv.de journalists. An editor can then select an article to proceed with its generation in ntv.de writing style. What appears as a simple click for the editors involves intricate workflows and prompt chains operating behind the scenes, culminating in a suggested article draft. However, the final decisions on article selection and any necessary rewrites remain with the editors, ensuring a human-in-the-loop approach.
By
Results
The solution developed by ntv.de and ML6 significantly streamlines the editorial process. It achieves this by reducing the time an editor spends searching for a topic and suggesting text elements that save time per article and benefit time-to-publish.
‍
As a result, the daily output of news is increased, while maintaining or possibly even enhancing the number of clicks and reads on the platform. This efficiency boost not only expedites the news cycle but also optimizes resource allocation across the editorial team. It gives journalists more time to produce unique content and allows them to devote more time to creative tasks.