I think we all understand that a lot of the news we read, while factually true, is still presented with sensationalism, marketing tactics, political influence or financial interests. Clickbait. And that’s the real problem. The facts might be correct, but the way a story is written, framed and presented can completely change how people perceive it and ultimately influence public opinion. So I decided to build a news website using multiple AI agents with one main goal: separate the facts from the framing. First, the system checks whether the story is actually true and relies on reliable sources. Then, other agents check for hallucinations, unsupported claims and anything the model might have added that isn’t actually supported by the evidence. Finally, the article is rewritten in a clean and neutral way, without trying to push the reader towards a particular opinion. Today was a big day for the project. I managed to get several agents working together and consistently delivering the same result without errors. The results are looking really promising. I’m also thinking about adding a feature that gives each story a Sensationalism Score. The idea would be to analyze how much the original media coverage relies on sensationalism, emotional language, clickbait or marketing techniques, and then show the reader a cleaner version of what the story actually says. Basically: What the media says → What the evidence actually shows. I’m still figuring out exactly how this should work. What features would you add?