Writer: Hassan Yehia
I built a working AI newsroom, alone, and it already produces work I can say confidently is good enough to be published. That was not the plan. The plan was a test: can different AI models give the same good quality, or does everything change when you change the model? I ran the test across three projects, with Qwen, ChatGPT, and Google Gemini. The result held across all three. Same standard, same editorial discipline, same quality bar, no matter which model was underneath. Inside that one newsroom, there are different editors, with different voices, different tonality, and different roles. And there are different sections doing different jobs: the researcher, the editor, quality control, and the publisher. Here is the moment that told me it worked. A friend evaluated the work. His own AI agent told him the editors were AI agents. He argued back. He told his AI agent it was wrong, that those editors were real people, not AI agents. That is the standard I was testing for. Not work that looks like AI. Work that reads like people did it. And one rule stays, whatever the stage: nothing is published without a human. In the first stage, the human approves the ideas and approves the final draft. In the next stage, the human will approve only the final draft, and even the decision on ideas will go through the AI agents. The human still has the last word, but the newsroom does more of the work. It is not a demo, and it is not a promise of what AI might do one day. It is already built, already running, and already tested against real editorial work. The lesson for me, as a journalist, is not that AI will replace humans. The hardest thing in this project was making the AI take the right decisions. That was the real test: to produce a very good amount of news, very well written news and reports, and at the same time to adjust and choose what is really interesting for our target audience, the audience I already put in the newsroom. That is what I take from this experiment. Writing itself is not what you need to master. What you need to master with AI, and what is changing with AI, is how the workflow works, and what the role of the journalist is. The role of the journalist today is becoming more and more about supervising AI, instead of AI replacing the journalist. That is what I think, and that is what I think it should be. This is the kind of work I do: journalism, building with AI, and training people to use AI in their own work. So here is my question for you: if a newsroom like this already exists, what would you want it to prove to you before you trusted it? Hassan Yehia, Business journalist, founder of Clawser.