Still can’t find what you are looking for? Try
ChatGPT.
Start with one recurring task, such as drafting a service page or checking a website change. Write down the outcome, the source of truth, what AI may change, who reviews the result, and how you will verify it. Test that brief on a low-risk task before building a larger set of instructions. Add rules for repeated or costly mistakes, not for every small preference.
A change is ready when it works in the real website, not just when the code runs or AI reports success. Confirm the intended page or module changed, approved content and settings remain intact, and the result works at relevant screen sizes. Check links, forms, accessibility, and other behavior affected by the change. If you cannot verify a required outcome, keep the work in review rather than calling it done.
Write a rule when a mistake repeats, is expensive to fix, or reveals a check that the next person or AI assistant can reliably perform. State when the rule applies, which source governs the work, what to do, and how to verify completion. Test it in later work and revise it if it conflicts with better guidance. A one-off wording preference may need only an edit to the draft.
Ask AI to define unfamiliar terms and explain the basic sequence, but do not make it your only source. Check important claims against a relevant primary document, a measurement, or someone who knows the subject. Ask what facts would change AI's explanation and note what you still cannot verify. You can begin learning without being an expert, as long as you know where your confidence ends.
Tell AI what you are learning and ask it to pose one question at a time. Have it wait for your answer, then give a hint before showing the full solution. Ask for a changed example next so you must apply the idea rather than memorize one answer. Check its feedback against a trustworthy source or teacher because AI can mark a correct answer wrong or explain a wrong answer confidently.
Consult an expert when the decision has legal, medical, financial, safety, or other significant consequences; when authoritative sources conflict; or when you cannot verify a critical claim. AI can help you prepare better questions and organize the evidence you have. It should not be the final authority for a decision whose risks you cannot evaluate yourself.