Concerns About Token Usage in OpenAI Integration with Example Data

Hey! Im using Open AI Integration "Give examples of how the AI should respond" - My examples have an extensive amout of data that have no values. Can please someone in your team have a look into it? I think this generates huge amount of tokens per prompt, which prevents me from using clays managed open ai acc, exceeding the limit of 15k tokens real fast. What could be the reason?