Prompting

Garden · not yet reviewed Established: widely used Widely discussed since 2022

Telling an AI what you want in plain words: the task, who it is for, what good looks like and any limits. Clear requests get better results than clever wording.

Also called: writing a prompt · asking AI · giving instructions

Connections

Understand first

Ideas this one builds on.

• Large language model (garden)

Leads on to

Ideas that build on this one.

• Artifacts and canvases (garden)
• Clarifying questions (garden)
• Context engineering (garden)
• Custom instructions (garden)
• Examples in prompts (garden)
• Prompt engineering (garden)
• Steering (garden)

Blueprints that practise it

• Let AI ask before it plans (garden)

Resources that teach it

• AI Fluency: Framework & Foundations (garden)
• AI Fluency for Nonprofits (garden)
• Claude 101 (garden)
• Claude Help Center: Introduction to prompt design (garden)
• Google AI Essentials (garden)
• Gemini API prompt design strategies (garden)
• Gemini for Workspace prompting guide (PDF) (garden)
• OpenAI Academy: Apply AI at Work track (garden)
• OpenAI Academy: AI Foundations (garden)
• ChatGPT 101: A Guide to Your AI Super Assistant (recording) (garden)
• Prompt engineering best practices for ChatGPT (garden)
• Assigning AI: Seven Approaches for Students, with Prompts (garden)
• Wharton Generative AI Labs Prompt Library (garden)
• Getting started with AI: Good enough prompting (garden)
• Instructors as Innovators: A future-focused approach to new AI learning opportunities, with prompts (garden)
• How to be more ambitious: A tutorial for humans with models (garden)

Competency

• Describe the task
• Frame the outcome

Find prompting on the commons map ↗

You may know it from

• workplace communication writing a clear request to a colleague Goal, audience, deadline and format help a person as much as they help an AI.
• operations a work order

In plain words

A prompt is whatever you type or say to an AI to get it working. Prompting well depends on the same things that make any request clear: what you want, why, who it is for, what a good result looks like, and what to avoid.

To test a prompt, imagine handing it to a capable stranger with no background. Would they know what to do? If they would need to ask five questions first, the AI is guessing at those five answers. Adding a sentence of purpose (“this is for a board that skims”) often helps more than adding instructions.

You also don’t have to get it right in one go. A good first prompt followed by a few rounds of “closer, but shorter and less formal” is normal practice.

Example

Weak: “Write a newsletter intro.” Stronger: “Write a three-sentence intro for our monthly volunteer newsletter. Readers are retirees who skim on their phones. Thank them for the food drive (412 boxes) and point them to next month’s sign-up.”

How it connects

• Prompt engineering was the early, trick-focused version of this skill.
• Context engineering widens it from the words you type to everything the AI can see.
• Adding examples and inviting clarifying questions are two quick upgrades.

Try it

Take a prompt you used recently. Write down the one piece of background a stranger would most need. Add it, rerun, and compare the two answers.

---

Read this card as plain Markdown: /concepts/con-prompting.md. Know a parallel from your own field, or a product that implements this? Add it.