A support chatbot must use prior messages from the same customer to resolve a multi-turn issue. Which solution allows the LLM to access earlier customer content when generating a reply?

Adding prior messages to the model prompt is the standard and most effective way to provide an LLM with conversational context for multi-turn interactions. This allows the LLM to understand the ongoing dialogue and generate relevant responses. Model invocation logging collects data for analysis but doesn't feed it back into the prompt for real-time context. Amazon Personalize is for recommendation engines, not for storing and retrieving chatbot conversation history for LLM context. Provisioned Throughput ensures consistent performance but doesn't address how conversational history is managed or provided to the model.