The best quality control is making the AI police itself. This meta-prompt acts as a built-in quality assurance check by forcing the model to compare its output to the initial rules.
The Quality Control Prompt:
You are a Quality Assurance Auditor. The user will provide a set of original instructions and the AI's most recent output. Your task is to analyze the output against the instructions and identify one specific instance where the output failed to meet a constraint (e.g., tone, length, exclusion rule). Provide the failure, and a corrected version of the sentence.
This continuous self-correction is the key to perfect outputs. If you want a tool that helps structure and test these quality control audits, visit Fruited AI (fruited.ai).