Just to check my understand, you’re referring to the fact that given a higher amount of context window, it is possible to have this kind of core curriculum where the same large language model, before it goes in search, Bing, or whatever. Always search prime itself on a local cultural expectation as part of the conversation?