I am doing a chat app and I need to generate answers to 1000 prompts. Can anyone recommend the simplest method to achieve it without Kubernetes and training my own model etc?