Search that answers the question

Most search tools give you a list of links and a block of text scraped from the top result. You still have to piece together whether that text actually addresses what you asked. Mu now does something different: it synthesizes web results in light of your exact query. If you ask “What is the flight time from New York to London?” it does not just fetch a generic summary of flight times — it reads the available sources, picks the number that matches your question, and tells you the typical range. The answer is grounded in the web content, but it is shaped by the intent you expressed.

The key refinement is that the synthesis step starts with the question, not the page. Mu looks at the snippets and full text from search results, then extracts only the parts that are relevant. If multiple sources disagree, it says so. If the most recent information is a few days old, it leads with that caveat before giving the answer. The goal is to save you from scanning five tabs just to confirm a simple fact. When the web is uncertain, or when the answer depends on unstated assumptions, Mu tells you that too.

This is not a silver bullet. If your question is vague or subjective, the synthesis will be limited by what is on the web. Mu does not fabricate an answer when the evidence is thin — it says “I don’t know” or “The available sources are conflicting.” That honesty is part of the design. The trade-off is that you get a direct answer, but you lose the serendipity of browsing. For many everyday searches, that trade-off is worth it. You ask, you get an answer, and you move on.