Your first useful ChatGPT task: make a working document

Choose one task you already understand well enough to check. That might be turning rough notes into an agenda, organizing a repair list, or drafting a welcome sheet. A familiar task lets you learn how to steer the AI without also having to judge an unfamiliar technical subject.

Bring a source and a purpose

Use a synthetic example or material you are authorized to share. Tell ChatGPT what the source represents, who will read the output, and what form you want. If your current account supports the relevant file workflow, provide the file through that supported interface. Otherwise, a short pasted example can be enough to begin.

Try this brief: “Use the notes below to create a one-page weekly repair review. Include the issue, current status, next action, and missing facts. Keep uncertain dates uncertain. This is for my review, and nothing should be sent or changed elsewhere.” The request defines useful work and leaves an obvious way to inspect it.

Review in two passes

First check substance: are the facts correct, are open questions visible, and has any information been invented? Then check presentation: can the intended reader use it quickly, and is the language clear? Separating those passes prevents polished formatting from distracting you from a factual problem.

Make a specific revision request

• Name the section that needs improvement.
• Explain what is wrong or missing.
• Supply the correct source fact if you have it.
• Describe the desired replacement or constraint.

For example, “The plumber has offered a window but the visit is not confirmed. Change that row to awaiting confirmation and keep the offered date as a possibility.” This correction teaches the immediate workflow more effectively than simply asking the model to be more accurate.

Keep the result

Save or export the useful deliverable through the available product controls and inspect the actual file when one is produced. Note the brief that worked so you can reuse it with fresh sources. The goal of the first exercise is not total automation. It is one piece of work that is easier to produce, easy to verify, and genuinely useful the next time you open it.