Skip to content

Prompting with User Stories

From Principle to Practice

You just learned the Three Pillars — Scope, Intent, Structure — and why specificity matters. Now here's a format that delivers all three pillars every time. It's called a user story.

A user story is a simple, structured way to describe what you want built and how you'll know it works. It has two parts: the story (what and why) and the acceptance criteria (how you'll verify it).

The Story: Who, What, and Why

A story follows this format:

As a [who needs this], I want [what they need] so that [why it matters]

Notice that this format starts with a person — not a feature, not a technology, not a screen. That's intentional. Good software is built for humans, and the best way to build something valuable is to stay connected to who you're building it for and what they actually need. A user story is a small act of empathy: you step into someone else's shoes and describe the world from their perspective. When you do that, the thing you build is more likely to be something they'll actually use.

The "As a" tells AI who it's building for. The "I want" tells AI what to create. The "so that" tells AI why — and that "why" is critical. Without it, AI has to guess what you're optimizing for.

Example for your Avalanche Field Guide:

As a backcountry skier planning a trip, I want an interactive danger rating display so that I can quickly assess whether conditions are safe before heading out.

Why "So That" Matters More Than You Think

The "so that" clause is the most commonly skipped part — and the most important one for AI.

Here's why: AI optimizes for completing what you asked for. If you say "I want a danger rating display," AI will build some kind of display. But if you say "I want a danger rating display so that I can quickly assess whether conditions are safe," AI now knows the display needs to prioritize clarity and speed of comprehension — not decoration or detail.

  • Without "so that": AI guesses what "good" means
  • With "so that": AI knows what outcome you're trying to create

Your "so that" is the steering wheel. Speed without steering is just a faster crash.

Acceptance Criteria: How You'll Know It Worked

After the story, you add acceptance criteria — specific conditions that define "done." These use a format called Given / When / Then:

Given [a starting situation], When [something happens], Then [what should be true]

Example:

Given the current danger level is "Considerable" (Level 3), When I open the danger rating display, Then I see an orange indicator, the word "Considerable," and travel advice that says "Dangerous avalanche conditions. Careful snowpack evaluation, cautious route-finding and conservative decision-making essential."

This isn't just documentation — it's your verification checklist. When AI builds something, you check each acceptance criterion. Pass or fail. No guessing whether it's "good enough."

How This Maps to the Three Pillars

Three Pillars User Story Format What It Does
Scope As a [backcountry skier]... Tells AI who it's building for and what domain to focus on
Intent I want [danger display] so that [I can assess safety] Tells AI what to create and what outcome to optimize for
Structure Given/When/Then acceptance criteria Tells AI exactly how the output should behave — and gives you a way to verify it

The Three Pillars are the principle. The user story is the practice. Every time you write a story with acceptance criteria, you're automatically covering Scope, Intent, and Structure.

Try It: Rewrite a Prompt as a Story

Format: Pairs Setup: Break into pairs within your team (groups of two).

Take this vague prompt:

Make me a checklist for avalanche safety gear

Working together with your partner, rewrite it as a user story with at least two acceptance criteria. Use this template:

As a ___, I want ___ so that ___

Given , When , Then ___ Given , When , Then ___

Then send BOTH versions to your AI chat tool — the vague prompt first in one conversation, then your story in a new conversation. Compare the results together.

Start a new conversation for each prompt so the results are independent.

Start a new conversation for each prompt so the results are independent.

Team Activity: Story vs. No Story

Format: Split & Compare Time: ~3 minutes Setup: Split your team in half — Pair A and Pair B.

The task: Build a "what to do if you're caught in an avalanche" emergency response section for the field guide. Each pair starts a new conversation in their AI chat tool.

  • Pair A: Prompt AI with plain English — be as specific as you can, but don't use the story format.
  • Pair B: Write a user story with acceptance criteria first, then send the whole thing to AI as your prompt.

Regroup: Come back together. Compare the two results side by side.

Discuss: Which result is closer to something you'd actually use? Which one was easier to verify — could you point to specific things that are right or wrong? Did the acceptance criteria give Pair B a checklist to verify against?

Key Insight

The Three Pillars tell you what makes a prompt good. A user story is a format that delivers all three pillars every time: "As a" gives you Scope, "I want... so that" gives you Intent, and Given/When/Then acceptance criteria give you Structure — plus a built-in verification checklist. Write your prompts as stories, and you'll spend less time guessing whether the result is right.