Mediahuis produces large volumes of video news every day. But without rich metadata, reuse remained limited. Together with Mediahuis, Dawn Technology developed Media Miner: an AI solution that, in a production-ready pilot, demonstrates how more than 400,000 videos can be automatically enriched and made instantly discoverable within the existing workflow.
Discover our sector Contact us
Mediahuis is an independent news publisher with a clear mission: to tell reliable and diverse stories across text, audio, and video. With leading news brands such as De Telegraaf, NRC, Dumpert, and Metro, Mediahuis reaches a large audience every day. Video plays an increasingly important role in this. Collectively, Mediahuis’ news brands generate around 65 million video views per month. Every day, large numbers of videos are produced, stored, and distributed.
At the heart of this workflow is Zebra: the in-house developed Media Asset Management system. Technically, Zebra integrates well with the rest of the chain, from CDN and CMS to frontends and players. However, its functional value depends heavily on input and that is where the challenge lies.
Videos are stored with minimal metadata. Usually a title, sometimes a short description, but often little more than a file name. Additional context such as topics, people, locations, or relevant fragments is missing. Not because the information doesn’t exist, but because manually enriching content under editorial time pressure is simply not feasible.
The result is a rapidly growing database of more than 400,000 video assets that are difficult to search. Existing footage is hard to find, leading editorial teams to re-film, purchase external content, or leave valuable material unused. This costs time, money, and journalistic impact.
Mediahuis wanted to break this pattern, without adding extra pressure on the editorial team. Manual tagging was not an option. External SaaS solutions proved costly and introduced risks in terms of data, privacy, and dependency. The need was clear: a solution that enriches content automatically, integrates with the existing workflow, and runs fully in-house.
With the rise of AI, a new approach became possible. By automatically analyzing video, audio, and speech, relevant context can be generated, while still allowing for human control and refinement. Mediahuis was looking for a partner who could translate this technology into a practical and secure solution.
Dawn Technology started with a Proof of Concept, focused on a single brand and without impacting the production environment. Based on the insights from this phase, the solution was further developed into a production-ready pilot.
Together, clear requirements were defined:
The choice was deliberately made for open-source technology, ensuring full control over data, costs, and further development. The result is an on-premise AI solution that automatically analyzes and enriches video content with editorially relevant metadata.
Media Miner is built as a multi-agent AI application, where specialized components work together:
Within the pilot, videos are automatically enriched and linked to existing assets in Zebra. The solution is designed to integrate seamlessly into the Mediahuis ecosystem, without adding manual steps for editorial teams.
A strong focus was placed on relevance. Not everything that is technically possible is journalistically valuable. That is why the solution was developed iteratively, in close alignment with real-world use.
In its pilot phase, Media Miner demonstrates how the Mediahuis video library can become significantly more accessible. Videos are enriched based on image, audio, and text, making it easier to find relevant content that was previously hidden.
The first results show clear potential:
The solution is fully designed with digital sovereignty in mind. Data remains in-house, costs are predictable, and the technology can scale towards full production implementation.
Media Miner shows how AI, when applied thoughtfully, can strengthen journalistic processes. Not by influencing content, but by unlocking access to what already exists. That is where its real power lies.
What our clients say
We want to tag and enrich our content, but time and cost are the main obstacles. With the rise of AI models, new opportunities are emerging. With the open-source, multi-agent AI solution Media Miner built by Dawn Technology, we can automatically analyze context, scenes, speech, and objects and generate editorially relevant metadata.
RTL reaches millions of people every day. Not only through television broadcasts, but also through its digital platforms and apps. We are RTL’s trusted partner for consultancy and development of their digital solutions.
Anexo helps organizations manage complex processes and data. From contract management to the execution of technical measurement assignments: accuracy and reliability are key. To support further growth, Anexo asked Dawn Technology to help digitize their mutation data, scheduling, and contract analysis. The goal: save time, improve quality, and achieve scalability without compromising on security or human touch.
Curious how AI can make your content more discoverable, without adding pressure to your team? We’re happy to explore it with you.
