Imagine this: A reader asks ChatGPT about a local election in Bavaria. Instead of an anonymous summary stitched together from unnamed sources, they get a branded news carousel from their regional paper. They swipe through headlines, tap into a full-screen article, and ask a follow-up question. The AI answers exclusively from the verified source material. They never leave the conversation. The journalist is visible. The newsroom is present. The source is not a footnote; it is the experience.

This is NewsConnect, built by The.Lab, Ippen Digital’s Human Experience Lab. Here is what we built and why it matters.

The quiet disappearance

A growing share of audiences no longer search for news; they ask for it in conversations with AI. Over 60% of users aged 18 to 29 already use ChatGPT. They expect answers, not links.

And when they ask, the AI answers. It summarises. It paraphrases. It delivers. But it rarely names the journalist. It almost never shows the brand. The reporting gets consumed. The reporter becomes invisible.

For a local newsroom that has spent decades building trust in its community, this is not a technology trend to watch. It is an erosion already underway. If your journalism only lives on the Web, it is quietly disappearing from the places where your next generation of readers already are.

Bringing journalism into the conversation

We decided not to wait. We built NewsConnect, an AI-native distribution infrastructure that embeds trusted journalism directly into conversational interfaces.

The idea is deceptively simple. We take existing publisher content and standard RSS feeds, transforming them into interactive, branded “content apps” that live natively inside AI platforms like ChatGPT. News carousels for browsing. A fullscreen reader for immersive reading. Audio podcasts for listening. Event maps for local stories. All rendered inside the chat, all clearly attributed, all editorially controlled.

The full article text is available to the AI, which means follow-up questions are grounded in verified reporting rather than hallucinated from training data. The result: less hallucination, more context, and a reader experience that feels natural rather than bolted on.

From one pilot to 80 brands

We built NewsConnect on the Model Context Protocol (MCP), a standard for connecting AI models to external tools and data sources. It was so new that bugs, missing features, and breaking changes were part of every sprint. We engineered custom abstractions and fallback mechanisms to make it reliable anyway.

The breakthrough was not making it work; it was making it scale.

Starting with a single pilot, serienjunkies.de, we abstracted editorial logic, transformation rules, and interface components into reusable infrastructure. Then we scaled from one publisher to more than 80 brands. Not in months. In minutes. The design system handles brand-specific styling per publisher, so high-quality AI-native journalism does not have to be bespoke. It can be industrialised.

The numbers behind the infrastructure: 80+ publisher brands, over 2,000 journalists, 29 million readers per month, 4,000 articles published monthly. A dedicated control environment enables editorial teams to preview, iterate on, and govern their content apps before they go live. That operational backbone is what makes this viable at network scale.

Why this matters now

NewsConnect protects against the silent erosion of publisher visibility. Journalism becomes a branded destination inside the conversation, not anonymous training data feeding it. Every article retains its source, its author, its portal identity. It is 100% citation at distribution scale.

But it also redefines what a publisher can be in the AI ecosystem. Not just a content provider. A context provider. Essential infrastructure for trustworthy AI. When AI platforms need grounded, verified information, publishers who built the pipes will deliver it.

What started as a single pilot became an infrastructure that scales editorial trust across platforms that think in tokens, not headlines.

What comes next

NewsConnect marked a shift from Web-first to AI-native publishing. We are now building an onboarding interface that lets any publisher go from sign-up to live preview in minutes. The infrastructure is ready. The channel is open.

Now we need to fill it with the journalism that communities depend on. And we need to do it together. Because the question is no longer whether AI will reshape distribution. It is whether publishers will be inside that conversation, or outside it.

Before, AI summarised. No sources. No brand.

Now, publishers speak inside AI. Branded. Interactive. Trusted.

The source is no longer a footnote. It is the experience.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .