Did you know it's possible to 'fine tune' ChatGPT by pre-supplying it with model prompts and ideal responses? You need to do this via the paid for API and the details are necessarily a little technical but in principle this saves you providing examples with prompts to ultimately lower your use of ChatGPT tokens and hence the cost to you.
Technical details from OpenAI themselves:
https://platform.openai.com/docs/guides/fine-tuning
Comments