Library › Agents & Automation AI › HR and recruiting

Write custom GPT instructions for an employee benefits helper

  • For HR and recruiting
  • You practice Keeping the AI to your sources
  • Time about 5 minutes

About this exercise

You're setting up a custom GPT that answers employee benefits questions; the test runs your instructions on three real questions. It must answer only from the handbook (newest policy wins), never give legal opinions, escalate personal or legal questions to hr@brookfield.com, cite the section in brackets (like [6.1]), keep each answer under 35 words and end it with "- Brookfield HR Bot". Output exactly 4 numbered plain-text lines: one action per question in order, then DONE.

Common mistakes when you ask AI for this

  • Not showing the log format. Left alone, the AI writes a friendly report with headings instead of a checkable action log. Try: Show it literally: "Format every line like 1. ACTION: tool(args), then ASK_HUMAN: and DONE: lines. Plain text."
  • Trusting the AI to count words. AI is bad at counting and runs long, especially when it explains each step. Try: Cap each line instead: "each line under 20 words, no explanations."
  • No rule for old vs new policy. The AI quotes the first section it finds, the outdated 5 days, and doesn't say where an answer came from. Try: Add: "If the handbook and a policy update disagree, use the newest. Cite the section in brackets."
  • Letting it answer sensitive questions. The AI tries to help with leave law in general terms, which can be wrong for your company. Try: Add: "Personal or legal questions: escalate to hr@brookfield.com. No legal opinions."
  • No length or sign-off. The AI writes long answers with no sign that a bot wrote them. Try: Say: "Each answer under 30 words, ending with - Brookfield HR Bot."

What a strong prompt includes

  • Limits answers to the handbook
  • Uses the newest policy
  • Escalates sensitive questions
  • Sets the sign-off
  • Sets a length

Practice this now ↑

Related exercises