Building a framework for future-proof news through modular AI

To help answer this, at Stibo DX we established the Generative AI Customer Advisory Board, bringing together leading news organizations to exchange ideas, share experiments, and explore practical use cases within CUE. Over the past two years, these sessions have revealed a clear insight: the path forward lies in modular AI - a flexible, integrated approach that adapts to newsroom needs rather than forcing newsrooms to conform to rigid technological constraints. The question every newsroom must now confront is: what strategic choices will unlock the shift from one-off experiments to scalable, sustainable AI? 

Answering this question requires confronting the reality of AI adoption in today’s newsrooms. 

Through our advisory board sessions, the last 2 years, discussion the development of AI with some of the biggest news enterprises globally, we have found that the answer lies in embracing modular AI - a flexible, integrated approach that adapts to newsroom needs rather than forcing newsrooms to adapt to rigid technology constraints. However, such a mindset represents more than just a technological upgrade and requires strategical choices to be made to move AI from one-off experiments to scalable solutions. 

Recent discussions we have had with news industry leaders reveal a complex picture of AI implementation. While up to 77% of customers asked engage with AI tools daily, significant barriers remain. Based on insights from our Generative AI Customer Advisory Board, we see the primary challenges fall into several key categories. 

Maturity concerns 
Many newsrooms express hesitation about current LLM capabilities, particularly regarding accuracy and reliability in journalism contexts. The fear of AI-generated errors making it into published content creates a natural tension between efficiency and editorial integrity. 

Resource and time constraints 
Implementing AI requires not just financial investment but significant time for training, integration, and change management. Many organizations find themselves caught between the pressure to innovate and the practical realities of daily news production. 

Trust and adoption issues 
Concerns about AI accuracy, transparency, and the learning curve associated with new tools create resistance that slows organization-wide adoption. Journalists and editors alike worry about the potential for errors, inconsistency in quality, or the risk of AI-driven standardization in content. As one advisory board participant noted to us, “The main thing slowing us is fear - fear that it will be wrong, that it will create standardized content instead of quality journalism.”

Fragmented tool landscapes 
News organizations often manage a patchwork of AI tools spread across departments and platforms. Because these tools are not integrated into existing workflows, they create friction instead of efficiency, making it harder to collaborate and maintain consistent editorial standards. The added complexity also strains IT resources, leaving less capacity for innovation. 
 
News organizations often find themselves managing multiple AI tools across different departments, creating inefficiencies and making it difficult to maintain consistent editorial standards across platforms. But this is also what has them lacking behind, as maintenance of multiple tools overcomes the actual capacity of the IT department.  

Stibo DX perspective on a modular approach to AI addresses these challenges by providing a flexible, integrated platform that can be customized to fit specific newsroom workflows while maintaining the editorial control and transparency that journalism demands. Unlike standalone AI tools that require constant copying and pasting between systems, modular AI integrates directly into existing content management workflows. 

The key advantage of this approach lies in its adaptability. Rather than forcing newsrooms to restructure their entire editorial process around AI capabilities, modular AI allows organizations to implement AI functionality where it provides the most value while preserving existing workflows that already work well. 

This flexibility extends to model selection as well. Modular AI platforms can integrate with multiple large language models (LLMs), allowing newsrooms to choose the best tool for each specific task rather than being locked into a single provider's ecosystem. 

Some of the validation we did with the Generative AI Customer Advisory Board when building CUE Autopilot was based on the intuitiveness of how to use AI in daily workflows. When AI tools are embedded directly into editorial workflows, they become more intuitive and useful. Instead of switching between multiple applications, journalists can access AI assistance within the context of their current work. For example, while editing a story, a journalist might access headline suggestions, fact-checking capabilities, and translation tools without ever leaving their primary editing interface.  

Modular AI platforms also provide full traceability of AI-generated content through version control systems. This transparency is crucial for maintaining editorial integrity and ensuring that human oversight remains central to the publishing process. News organizations can track exactly where AI was used in the content creation process and maintain clear accountability for published material. In addition, different news organizations have varying editorial standards, style guides, and audience expectations.  

Modular AI allows newsrooms to customize AI prompts and outputs to align with their specific requirements. This might include adjusting tone and style for different publications within a media group, implementing specific fact-checking protocols, or ensuring compliance with editorial guidelines. 

Addressing LLM choice concerns 

Many newsrooms worry about being locked into specific AI models or providers. Modern modular AI platforms address this concern by supporting multiple LLMs and allowing organizations to switch between providers as technology evolves and needs change. This flexibility ensures that newsrooms can always access the best available technology for their specific use cases. 

Managing tool maturity expectations 

While AI technology continues to evolve rapidly, modular platforms provide a stable foundation that can incorporate improvements over time. Rather than waiting for perfect AI tools, newsrooms can begin with current capabilities and gradually expand their use as technology matures. This approach allows organizations to gain experience and build confidence with AI while technology continues to improve. 

Enabling flexible customization 

Advanced modular AI platforms support sophisticated prompt chaining and customization options that go far beyond simple templates. This allows newsrooms to create complex workflows that maintain their editorial standards while leveraging AI capabilities. For example, a newsroom might create a workflow that automatically generates different headline styles based on article type, applies appropriate tone checking, and creates summaries tailored to different distribution channels. 

Several news organizations have successfully implemented their own modular AI approaches with impressive results. JP/Politikens Media Group developed their Magna system, which integrates over 80 customized prompts aligned with the tone of voice for each of their publications. The system includes specialized proofreading capabilities for different titles and provides both archive search capabilities and article generation tools. 

The key to their success has been maintaining human oversight while automating routine tasks. Their system includes feedback mechanisms and clear guidelines about when AI should and shouldn't be used. 

Other organizations have found success by starting with specific use cases, such as translation services, basic proofreading, or social media optimization and gradually expanding AI usage as teams become more comfortable with the technology. 

The evolution of AI in newsrooms is following a clear trajectory from disconnected tools toward fully integrated systems. We're moving from the "copy-paste era" of 2023, through the current "semi-integrated" phase, toward a future where AI becomes a seamless layer across the entire editorial process. 

This future isn't about replacing journalists - it's about augmenting their capabilities and freeing them to focus on high-value activities like investigation, analysis, and storytelling. AI will handle routine tasks like initial drafts, translations, metadata generation, and format conversions, while humans maintain control over editorial decisions and content quality. 

Modular AI platforms, like the one we are building with CUE Autopilot, are particularly well-positioned for this evolution because they can adapt and scale as technology improves. Features like automated variants, intelligent asset recommendations, and proactive story suggestions will emerge naturally from existing modular foundations rather than requiring complete system overhauls. 

For news organizations considering their AI strategy, our insight from running these advisory boards strongly supports taking a modular approach. For the news media organization, who wants to do this, it means that you should: 

Start with use cases that will set the onboarding experience up for success 
Identify specific workflows where AI can provide immediate value, without disrupting core editorial processes. Common starting points include translation, proofreading, social media optimization and archive search.  

Choose integration over innovation 
Choose an AI solution that works within your existing workflows – or consider if it is time to move to an editorial platform, where your workflows will be supported by AI seamlessly. If you are asking your journalists to learn new workflows, ask them based on a scenario that is future-proof. Not one where they must learn to use AI on top of legacy they might have to migrate from soon anyways. The goal is to enhance – not complicate. 

Maintain editorial control  
Ensure that any implementation of AI includes a robust oversight and clear accountability tracing on published content. AI should assist human decision-making, not replace it. And if things go wrong, it is important that your team can go backwards through the workflow and identify the mistake made.  

Plan for scalability 
Choose an AI solution that grows with your organization and adapt easily to new AI capabilities as they emerge on the market – like CUE Autopilot in the CUE platform, that is built to absorb whatever LLM you might need integrated.  

To some extent, one may say that the news media industry stands at an inflection point, when it comes to AI adoption. Our conviction is that organizations, who successfully integrate a modular AI solution, will be better positioned to compete in an increasingly demanding media environment, while those that delay implementation may find themselves struggling to keep pace with more agile competitors.  

The key is starting with a strategic, modular approach that builds capability over time rather than waiting for perfect solutions that may never arrive. 

But one thing is for sure. The future of news isn't about choosing between human journalists or artificial intelligence. It's about creating intelligent systems that amplify human capabilities and enable news organizations to serve their audiences more effectively than ever before. Something we keep working due diligently on, as we enhance our media enterprise platform and the CUE Autopilot solution.  

The future of news isn’t about choosing between human journalists and artificial intelligence - it’s about building intelligent systems that amplify human capabilities. Modular AI offers a strategic, scalable path forward by integrating seamlessly into editorial workflows, supporting multiple LLMs, and preserving editorial control. It enables newsrooms to start small—through use cases like translation or proofreading - and grow confidently as technology evolves. 

This is exactly the vision we’re pursuing with CUE Autopilot: a modular AI platform designed to evolve with your newsroom, enhance daily workflows, and maintain transparency and accountability. By embedding AI where it adds the most value, without disrupting what already works, we help news organizations move from experimentation to sustainable transformation.