Write the first useful request
Give ChatGPT a job, source, and finish line
After this, you'll be able to write a first request that gives ChatGPT enough context to produce a specific, checkable output.
Before you start
Complete What ChatGPT is good for first.
The idea
A good ChatGPT request has three parts: job, source, and finish line. The job says what work to do. The source says what information to use. The finish line says what done looks like.
- The rewritten request namesThe rewritten request names one job, not five
- tells ChatGPT to use theIt tells ChatGPT to use the resume draft and job posting
- defines a visible outputIt defines a visible output format
- Write one request with aWrite one request with a job, source, and finish line, then verify the answer can be
Here is the before and after: before, ChatGPT is answering a loose request. After, you can write a first request that gives ChatGPT enough context to produce a specific, checkable output.
Now try it use the exercise prompt on one real task from today. Keep the answer small enough to check before you reuse it.
You are ready when the task, source, surface, and proof are all visible.
Try it (8 min)
Watch out for
- Writing a giant instruction block before you know what ChatGPT will miss.
- Asking for a polished final when you still need diagnosis.
- Leaving the source implicit because you assume ChatGPT knows which file matters.
- Using vague finish lines such as 'make it better' or 'give me insights.'
Paste this into Claude
Rewrite this weak ChatGPT request into a strong one: Weak request: "Help me with my resume." Use this structure: Job: [what ChatGPT should do] Source: [what information it should use] Finish line: [what the output should look like] Then give me the final prompt in one paragraph. Keep it realistic for someone who has a resume draft and one job posting.
What a good response looks like
Use my resume draft and the job posting below to find the five strongest edits. Return a table with Current line, Suggested line, Why it matches the job, and Missing proof. If a claim is not supported by the resume, mark it as missing proof instead of inventing a metric.
What good looks like
- The rewritten request names one job, not five
- It tells ChatGPT to use the resume draft and job posting
- It defines a visible output format
- It asks ChatGPT to mark missing information instead of inventing it
When this breaks
- Breaks when the job is broad because ChatGPT cannot tell whether you want diagnosis, drafting, editing, or planning.
- Breaks when the finish line is invisible because you cannot judge whether the answer is complete.
AI can help with this
Use ChatGPT to help you you can write a first request that gives ChatGPT enough context to produce a specific, checkable output. Start with the exercise prompt and your real input. Ask for one draft, then check it against this proof: The rewritten request names one job, not five. Accept only the version you can verify yourself.
You can now
You can name the job in one sentence
- You can name the source ChatGPT should use
- You can define the output shape before sending
- You can steer one miss without rewriting the whole prompt
Key takeaways
Good first requests are not fancy. They tell ChatGPT what work to do, what to use, and how you will know it is done.
- Job, source, and finish line are the foundation of usable ChatGPT work.
- A narrow prompt is easier to improve than a broad prompt.
- Tell ChatGPT to mark missing information instead of filling gaps.
- Follow-up turns should fix one visible miss at a time.