Exemplars vs. Conversation History

Currently, exemplars are specified as previous chat history items. For example:

exemplar_1 = {user: “what is the capital of France”}, {assistant: “Paris”}

This serves as both an exemplar (eg: the agent should respond tersely), as well as an actual previous message pair in the chat. However, this limits the applications of exemplars to single input/output pairs. Is there any way to specify exemplars and chat history messages separately?

chat_message_1 = {user: “Hello”}, {assistant: “How can I help you today”}
chat_message_2 = {user: “What is the capital of France”}, {assistant: “Paris”}
exemplar_1 = {user: “What is the capital of England”}, {assistant: “London”}

In this example, the exemplar can be applied at a given point in the chat history, and is separate to the actual chat history messages. In particular, how could I use exemplars to specify how the model should respond to a particular message given a previous conversation?