Exa has introduced a new search product called "Websets," which significantly outperforms Google in handling complex queries by leveraging a custom embedding-based search algorithm. The company claims that Websets retrieves over 20 times more accurate results than Google on a benchmark specifically created to evaluate complex searches, such as identifying software engineers in the Bay Area with blogs or NYC startups with over ten employees working on futuristic hardware. The benchmark was designed to avoid bias by using a one-shot sampling method for query generation. During testing, Websets demonstrated superior performance compared to Google and OpenAI's Deep Research by returning a higher number of correct results and offering flexibility in the number of returned results based on the compute power used. Although Google's keyword matching algorithm struggles with complex queries, Websets' advanced algorithm allows it to achieve impressive precision, showing that it can efficiently handle entity searches with multiple criteria.