Context is the set of information supplied to the model at request time: conversation history, documents, system instructions, business data. Since models do not know your company, the quality of the context makes the quality of the answer; that is the whole point of RAG architectures.
In practice at Gensai
Much of Gensai's work is building the right context: which data, in what order, with which instructions.