Kagi solves web search for now. LLMs are incapable of determining what’s important. They are excellent at determining what’s common. That doesn’t connect with news summarization in the way we’d ideally want it to unfortunately. I don’t care for low fidelity news.