How to Write Better AI Prompts for Recruiting Work

Start from outcomes

Good prompts answer who consumes the output, what format, and what evidence counts.

Before typing into ChatGPT or Claude, finish this sentence: When this works, I will paste the result into …

If the answer is nowhere, you are still brainstorming — which is fine — but switch modes: ask for options and critiques, not final copy.

Techniques recruiters reuse daily

Few-shot prompting

Paste two anonymized examples of emails or summaries you like. Ask for a third in the same shape.

Structured outputs

Ask for bullets, tables, or JSON matching your ATS fields. Structure beats paragraphs when you move fast.

Constraints

State length limits, tone (direct vs warm), and phrases to avoid (“world-class”, “rockstar”, vague superlatives).

Self-critique (sparingly)

Ask the model to list weaknesses in its draft — then fix them yourself. One pass is usually enough; endless loops waste time.

Context blocks beat clever one-liners

Keep a living doc:

• Company elevator pitch (three lines).
• Voice guidelines for candidate-facing copy.
• Role-specific must-haves vs trade-offs.

Paste that block at the top of serious prompts so every generation aligns with the same facts.

Practice loop

Pick one deliverable per week (brief, outreach, screening summary). Track editing time before vs after templating — that is your ROI story.

For structured curriculum and exercises, see courses on the Store.