[coredog64]

A customer service chatbot can require more than one LLM call per response to the point that latency anywhere in the system starts to show up as a degraded end-user experience.