The problem with prompting…

Most people ops folks use AI like a search engine.

One question in, one answer out.

This is fine sometimes, but it adds little value to your work.

The real change in your outcomes happens when you learn how to ask well, how to build a sequence, and how to challenge what comes back.

That's what separates someone who plays with AI from someone who thinks and designs with it.

This guide takes you through the whole arc:

Foundations first.
Then sequencing.
Then the bit most people skip entirely: making sure the output is actually good before you act on it.

<aside> 🧭

How to use this guide

New to prompting? Start at Section 1. Work through in order.
Comfortable with basics? Skip to Section 2 (Prompt Chaining). That's where the biggest jump in quality happens.
Already chaining? Go straight to Sections 3 and 4. Critical thinking prompts are what stop you shipping something that sounds right but falls apart under scrutiny. </aside>

YOUR LEARNING ARC
═══════════════════════════════════════════════════════════════

  ┌─────────────┐    ┌─────────────┐    ┌─────────────┐    ┌─────────────┐
  **│  Section 1  │───▶│  Section 2  │───▶│  Section 3  │───▶│  Section 4  │**
  │  Prompt     │    │  Prompt     │    │  Critical   │    │  People Ops │
  │  Eng. 101   │    │  Chaining   │    │  Thinking   │    │  Prompt     │
  │             │    │             │    │  Prompts    │    │  Library    │
  └─────────────┘    └─────────────┘    └─────────────┘    └─────────────┘
     **FOUNDATION**         **SEQUENCING**        **CHALLENGE**          **APPLICATION**
     "Ask well"         "Build on it"     "Stress-test"      "Use it daily"

Section 1: Prompt Engineering 101

<aside> 📖

Prompt engineering is the practice of writing inputs to AI that produce useful, reliable, context-aware outputs. It is not about memorising magic phrases. It is about learning how to communicate clearly with a system that knows nothing about your situation unless you tell it.

</aside>

The 5 Building Blocks of a Strong Prompt

Every good prompt draws on some combination of these five elements. You don't need all five every time. But knowing they exist means you can deliberately choose what to include.

Building Block	What it does	Example snippet
Role	Sets the AI's perspective and expertise level	"Act as a senior People Business Partner…"
Context	Gives the AI the background it doesn't have	"We're a 200-person SaaS company with 15% annual attrition…"
Task	Tells it what you actually need	"Draft a 90-day onboarding improvement plan…"
Constraints	Sets the boundaries: format, tone, length	"Keep it under 500 words. Use bullet points. No jargon."
Output spec	Defines the exact shape of what comes back	"Return a table with columns: Action, Owner, Timeline, Success Metric"

Weak vs. Strong Prompts

<aside> ❌

Weak prompt

"Give me some ideas for improving our onboarding."

No role or context
Vague task
No constraints
You will get generic nonsense back </aside>

<aside> ✅

Strong prompt

"Act as a senior People Ops leader at a 200-person tech company. Our 90-day onboarding has a 30% early attrition rate. Draft 5 specific, actionable improvements. For each, include the problem it solves, implementation effort (low/med/high), and expected impact on retention. Format as a table."

</aside>