The hackathon is the highlight of the year for the AI for Media Network. This year, it was themed “Conversational Products and Video Production with AI.” A total of 86 participants, divided into 14 teams, dedicated over 1,000 hours to develop demo versions of their projects. In the end, the team “SchnittmAIster” won the hackathon. They developed a tool where AI supports video production, including editing and voice-over.
The second AI for Media Hackathon wasa joint event by the AI for Media Network, pub.Tech, and Google Cloud.14 teams within the media industry participated, including ARD broadcasters BR, SWR, MDR, and NDR, the news agency dpa, as well as publishers like Ippen Media and Medienholding Süd. Some participants even traveled in from abroad.
BR Director General Dr. Katja Wildermuth praised this cross-media collaboration in her opening remarks: “It’s not often that public media, private media, industry and academia discuss and develop joint projects together. This is exactly what happens in the AI for Media Network. And that’s great.“
Interdisciplinary Teams
In the event space of Google’s Munich office, teams had one and a half days to develop a prototype leveraging AI, either for video production or for building products that users can directly interact with. The teams were interdisciplinary, as is typical for a hackathon: developers, (UX) designers, journalists, data experts, and product managers. They could utilize various AI models from Google, including the LLM Gemini 2.5.
AI-Supported Video Production Pipeline Takes the Lead
On the second day, each team presented their project to the jury and other participants in a five-minute pitch: outlining the problem, the AI solution, and showcasing a demo version. Ultimately, the six-member jury (Lilian Dammann, Director of Analytics & Data Platforms at pub.Tech; Regine Gatzka, Head of Enterprise Sales Media, Entertainment, Software & Telco at Google Cloud; Uli Köppen, Chief AI Officer at Bayerischer Rundfunk; Steven McAuley, Head of AI Customer Engineering at Google Cloud; Ole Reißmann, Head of AI at Der Spiegel; and Paola Sunna, Senior Technology Innovation Manager at Eurovision Italy) was most impressed by the “SchnittmAIster” team.
Their concept involves a prompt where the editorial team specifies how raw video material should be processed—whether it should become a news film with voice-over, a highlight reel from a soccer match, or key segments from an interview. Google’s Gemini 2.5 Pro AI model analyzes the video material and provides an editing suggestion. This generates an XML file containing the necessary cuts and all metadata for the video. The editorial team can then open this cut list in Adobe Premiere for final adjustments if needed. Additionally, an AI creates a voice-over for the video. Uli Köppen, Chief AI Officer at BR and hackathon jury member, remarked:
“You can tell the pace of advancements in the AI world: solving real problems instead of moonshots. I am confident that we will see the winning project implemented in a newsroom workflow in the coming months. Congratulations to the SchnittmAIster team!”
Rohcut AI and Interactive Podcast Secure Second and Third Places
Second place went to the “Rohcut AI” team from the news agency dpa. They demonstrated how to create social videos in a 4:5 format from raw video material in 16:9 format. By inputting a thematic focus, the AI selects appropriate scenes from the video, adds subtitles, and produces a finished social post.
Third place was awarded to “Interactive Podcast.” The team with members from BR, RTL and The Economist showcased how to make an AI-powered podcast interactive: users can pause an episode and ask a question about the content via voice. An artificial voice of the host then answers the question based on the podcast script, and the episode continues. Additionally, there were eleven other valuable use cases, ranging from an adaptive user interface that adjusts to desired content, to audio summaries of news, and voice chats that bring historical diaries to life.
AI for Media Network Now Boasts Over 900 Members
Founded in 2024 by BR and Ippen Digital, the AI for Media Network is dedicated to exploring how AI can enhance journalism. To this end, the interdisciplinary network promotes exchange among AI experts at its own events, demonstrates editorial AI use cases, and collaboratively develops AI solutions for journalism, as seen at this hackathon. The goal is to build a value-based AI ecosystem. The network now has over 900 members, primarily from Germany, Austria, and Switzerland. The next AI for Media Network meeting will take place on July 10 at the BR headquarters.
