Why Google, LLMs, and AI agents still don't show you the whole world

Imagine walking into the world's largest library. Millions of books, in every language, on every subject imaginable.

Now imagine that every time you visit, the librarian takes you to the same two shelves.

The books are excellent. Some are among the best written. But the rest of the library remains invisible.

In many ways, that's how we consume news.

We have Google, Google News, ChatGPT, Gemini, Perplexity, and a growing wave of AI agents that promise to monitor the world for us. The assumption is simple: if an AI agent is searching the web on my behalf, surely, I'm seeing more of the world's journalism than ever before.

The evidence suggests otherwise.

A 2025 study conducted at Northeastern University analyzed more than 24,000 real user conversations and over 366,000 citations from OpenAI, Google, and Perplexity. The conclusion was striking: All three relied heavily on a surprisingly small group of news organizations.

OpenAI showed the highest concentration. 67.3% of all news citations came from just 20 publishers. Reuters alone accounted for 22.8% of all news citations, followed by Associated Press (12.2%), Financial Times (7%), Axios (6.7%), and TIME (3.9%). Google and Perplexity showed different preferences, but the same pattern: a relatively small group of major international publishers dominated what their AI systems surfaced.

This means that the picture users receive is often built from a surprisingly narrow slice of the global news ecosystem.

But this isn't a flaw in AI search.

It's a consequence of what these systems were designed to optimize.

The discovery blind spot

Search engines optimize retrieval. Large language models optimize understanding.

Neither was designed to continuously discover emerging stories across thousands of local newsrooms, dozens of languages, and hundreds of countries.

This is what we call the Discovery Blind Spot.

Most people assume AI agents solve this problem.

AI agents automate your workflow. They don't necessarily expand the coverage.

Many journalists now build AI agents to monitor industries, companies, or breaking news. It's a smart evolution. But there's a hidden limitation.

Most AI agents search the same search indexes and the same familiar sources. Instead of escaping the information bubble, they mainly make it more efficient.

The echo chamber hasn't disappeared.

We still haven't solved discovery: Search answers questions; Discovery reveals stories nobody knew existed.

Those are fundamentally different challenges.

AI agents usually search the same familiar indexes, retrieval systems, and the same publishers we have relied on before. They automate the workflow, but they don't necessarily broaden it. Instead of escaping the information bubble, they can make it more efficient.

The discovery bottleneck

Every journalist understands where this becomes a problem.

The biggest stories don’t always begin at the news agencies, the BBC, or The New York Times. They often start in a local newspaper in Italy, a regional business publication in Japan, or a municipal website in Indonesia. Only after someone notices the story does it travel through news agencies and international media before eventually reaching Google and AI search.

If nobody picks it up, the story doesn't disappear.

It simply remains invisible to the rest of the world.

Working as a news researcher or a general news reporter over the past 20 years would mean that mornings begin with scrolling over five to ten news websites, social media, and newsletters - those they know and are used to reading.

Now, we have LLMs capable of summarizing almost anything in seconds. Yet the hardest part of journalism hasn't changed.

Discovery is still the bottleneck.

 The LLMs cannot solve the discovery blindness, and that’s why we’ve built an ultimate AI layer that finds out what you don’t know you were looking for, starting with the global news ecosystem itself, analyzing daily more than 100,000 news stories across thousands of publishers, languages, and countries.

The result isn't simply more articles. It's broader visibility, more perspectives, context, and often, stories that would otherwise remain invisible.

Journalism has many responsibilities. One of the most important is helping us see more of the world. We are here to support that goal.