How Rheinische Post built an AI governance that aligns with its digital subscriber goals
As publishers position themselves for the AI future, two priorities should guide them. First, they need to invest in building a strong brand and relationships with their customers. And they must differentiate themselves from AI by being human, approachable, and having an editorial voice.
This was the advice from Margret Seeger, Group Head of AI at Rheinische Post Mediengruppe (RPM). She joined WAN-IFRA’s recent Frankfurt AI Forum to outline the structure that the German publisher has created for its use of AI.
Headquartered in Düsseldorf, the Rheinische Post Mediengruppe’s biggest title is its flagship newspaper with a daily circulation of 208,000. The company also holds a majority stake in several local titles, including General-Anzeiger (Bonn) and Saarbrücker Zeitung (Saarland).
The group announced in March that it was acquiring Westfälische Medien Holding, the publisher of Westfälische Nachrichten among other local titles. The deal was recently approved by regulatory officials.
Like many publishers, RPM began exploring generative AI at the end of 2022 with the release of ChatGPT. Initially, there were no company-wide rules or strategy for AI use, Seeger said.
When she took on her current role in 2024, this included overseeing RPM’s AI efforts, developing a clear strategy, and ensuring the whole organisation was aligned accordingly.
Crucially, Seeger decided not to create a separate “AI strategy,” instead aligning the group’s diverse AI activities with its broader digital transformation strategy. This meant that instead of having distinct targets for AI use, AI efforts would need to support the group’s overall goal of growing its digital customer base to 250,000 digital subscribers.
“AI can help us achieve this goal faster, better, with higher quality, and sometimes also at a lower cost,” she said.
Governance structure built on three pillars
The group also clarified internal roles and set up an explicit structure for AI governance – a step Seeger called a “crucial factor of success.”
“If you want to grow with AI, you have to be clear about who’s allowed to do what, when and where,” she said.
A three-part framework now guides RPM’s AI initiatives:
Group head of AI: Every AI initiative has its own project lead, but in her role, Seeger supervises all initiatives, their planning and resourcing, and helps keep them on track.
AI Taskforces: Group-wide taskforces (covering all departments, from editorial to logistics) have the operational responsibility and oversee bottom-up implementation.
AI Council: Comprising top management, AI taskforce leads and directors of business units, the Council ensures the strategic orientation of AI projects and their alignment to broader company goals.
To support AI adoption within the company, RPM also set up an AI Academy, a centralised structure for tools and training – and “one of the biggest investments” around its AI efforts, Seeger said. The new structure allowed RPM to provide AI training to more than 1,000 employees.
Investment in internal AI capacity has been a key enabler for scaling AI within the organisation, Seeger said.
“If you haven’t started [AI training] yet, think about making it at least somewhat mandatory. Because the problem isn’t AI or the technology, but the problem is that our organisations aren’t mature enough to deal with the technology,” she said.
She added that RPM’s AI Academy is now shifting its focus more specifically to the top management: “Our leadership team sets the direction and needs to fully understand, on the one hand, which options there are with AI, and on the other hand, how to lead in an AI world.”
For example, AI has helped reduce product prototyping cycles from months to days. As processes speed up, it’s “going to change the way we work as an organisation. So getting the leadership on board to shape not only our products but also our organisation is going to be very important,” Seeger said.
AI projects across the value chain
To further enable AI experimentation, RPM initially built their own solution for internal AI use before switching to Langdock, an AI integration platform, at the end of last year.
Seeger said that after only a few months, about a thousand people are already using the platform to create their own tools: “This has been the fastest adoption I’ve ever seen of a tool in our company.”
RPM currently has more than 20 active AI projects, placed across the whole value chain of the company. In fact, some of the highest returns-on-investment (ROI) have come from projects that don’t involve the newsroom.
Automated creation of print pages has been a big success, for example, as have been AI projects within customer service. Some customer service processes are already 80% automated.
Seeger said that if someone in the company wants to propose a new AI project, they first need to answer questions such as: What customer problem are you measurably solving? What data and processes do you need? And what does adoption look like in everyday work?
The last question is crucial, she said: “Pilots can be very fancy, but when it comes to getting them across the newsroom and our 600 journalists, you really have to have a system in place that helps you make it work in real life.”
Not all AI projects have been successful. One initiative that RPM tried but eventually dropped was cloning voices from the newsroom for audio versions of articles. The usage of the feature was limited compared to costs, so it was abandoned.
In 2026, focus on ‘lighthouse initiatives’
More recently, RPM has shifted its focus from experimentation to “lighthouse initiatives” – bigger, more specific AI-powered projects.
For editorial, this means initiatives such as a research assistant that helps analyse local public documents, as well as new data tools. Meanwhile, sales and customer service have doubled down on using LLMs and agents to streamline their processes.
To support its B2B sales, RPM is working on a tool that will crawl the web (specifically Meta’s and LinkedIn’s platforms), identify branded posts as potential clients, and create leads for sales representatives. This tool, currently in an alpha phase, is being developed by an external agency.
Going forward, measuring ROI will be a central focus: Seeger said that 80 percent of their AI projects have specific objectives. The remaining 20 percent are projects that are deemed as necessary even if their direct ROI is difficult to quantify.
“AI is coming into our lives with a massive force,” Seeger concluded. To prepare for this, news leaders will need to act fast not only to identify the right AI projects, but also to “ensure that our organisations are ready – in terms of skill profiles, processes and decision-making.”
