out = await response_generator.acall(
    transcript=past_messages,
    question=question,
)

response = out.response # the response
save_memory = out.save_memory # the LLM's decision to save memory or not

past_messages.extend(
    [
    {"role": "user", "content": question},
    {"role": "assistant", "content": response},
    ]
) # update conversation stack

if (save_memory): # Update memories only if LLM outputs this flag as true
    update_result = await update_memories(
        user_id=user_id,
        messages=past_messages,
    )