wants to redo the search engine

Google Search organizes how billions of people think about facts and data, and for years it’s been organized around a principle dubbed “one true answer”: the idea that most people are looking for something best answered with a succinct factual snippet. But that’s not the only way people can browse the web, and today a company called is trying something different: a search engine built around sorting and comparing results., founded by two former Salesforce employees, opens in public beta today and announces a $20 million fundraising round led by Salesforce CEO Marc Benioff. The service ditches the linear list of links you’ll find in most general-purpose search engines, opting for a grid of answers organized by source. Sources include generic categories such as “web results” and “news”, but also specific sites such as StackOverflow, Wikipedia, Twitter, Amazon, LinkedIn and individual news sites such as The New York Times.

In addition to this organizational change, the big differentiating feature of is that it allows people to influence the sources they see. You can “upvote” and “downvote” specific categories, so when you search you’ll see favorite sources first, neutral searches next, and downvoting sources last.

When I searched for “Section 230” on’s pre-launch beta, for example, it by default displayed a general “web results” box, including links to Cornell University and the ‘Electronic Frontier Foundation. But I could also choose to see something like an excerpt from Wikipedia or a series of Reddit results first. Categories are also influenced by the context of your search. If I searched for “Moonfall” it would favor an IMDb-powered “What to Watch” box for Roland Emmerich’s next film, while searching for “infrastructure bill” would prioritize a general information grid and media coverage from many different outlets.

Some of these searches, like the one in section 230, end up yielding results quite similar to Google’s. But the interface encourages searching through a range of sources rather than clicking on the first link or two. It also includes interesting tools for specific use cases. Searching for “for loop javascript”, for example, will invoke listings of reference pages on Google, but will show plain-text syntax snippets from sources like W3Schools that you can easily copy and paste. is not optimized for answering basic questions like Google is, especially for queries that require guessing what people to want instead of what they literally type. It’s much more willing to encourage clicks to other pages – if you want “the actor who plays Sherlock Holmes”, will provide links to listings while Google will provide you with a movie star grid. It does not include small features like offering answers in the search bar.

And is generally less good at guiding your gaze to the most relevant information. If you’re not used to perusing the text of search results, it seems overwhelming and somehow cluttered. I only used a desktop version of the pre-launch beta, which looks like the way it is Assumed be experienced, but that’s not necessarily where most people want to search the web.

But the service seems much more honest about its own limitations than Google. Google’s text snippets are arguably the platform’s worst feature, granting a false sense of authority to inaccurate or offensive information or even summarize a correct answer in a dangerously wrong way. (Full disclosure: I was only able to find this last link via Google, where I got it by searching for “google grab snippet”, while offered general information about snippets and entries.) offers a “Quick Facts” box for queries like “distance from Earth to Moon”, but it comes with a bunch of other results. comes very close to being an engine for experienced web searchers who like to compare multiple sources of information. It’s still missing key features like limited date range search, something co-founder and CEO Richard Socher says will come later, and it’s not as versatile as Google. (Clicking the Maps icon in the closed beta actually takes you to Google Maps.) But that’s part of the appeal — Socher reasonably describes as an “unbundled” search of a Google-style web empire .

Oddly, the company doesn’t do this pitch very well in its launch announcement. He describes his system as “summarizing results from across the web”, which sounds almost precisely retrograde. Socher says the phrase refers to grouping results by source, but compared to Google, it has virtually no explicit editorial summary of the information.

Like DuckDuckGo, Brave Search, and other services trying to reduce Google’s overwhelming dominance in search, emphasizes the idea that it’s more private and less tracked. It’s not ad-supported and tied to countless related products like Google is, and there’s an incognito mode that says hides your IP address. Among other things, it promises that it will never target personalized ads to users. The service, unfortunately, doesn’t have a business model at all yet, so it’s unclear what kind of other compromises it might have to make in the future. probably won’t appeal to everyone, but it does offer an interesting and quite unique set of features. It’s one of the few non-subject-specific search engines that avoids feeling like a more grounded but practically inferior version of Google – encouraging to view the web as a true network of websites, not just water for an answer box.