Stop 12 · AI & Modern Design ToolsTopics · 4/7

Prompt Engineering for Designers

In short

'Make a nice login screen' gets you something generic. A prompt with the user, context, limits and format gets you something you can actually use.

Writing clear, structured prompts to get useful design-relevant output from an AI tool — providing context, constraints, and examples rather than a single vague sentence.

The jaadu leans into an auto-rickshaw and points down the road, giving the driver directions.

Think of it like this

Giving directions to the auto driver

Say 'Bhaiya, take me to the market' and you might reach the wrong market. Say 'the vegetable market near the bus stand, enter from the back gate, I'm in a hurry' and you get there. AI tools need the same detail.

Watch it happen

The parts of a useful design prompt

Good prompt

Role

Context

Task

Constraints

Output format

1/5A prompt is a brief. Each part removes a guess the AI would otherwise make.

The big ideas

What you need to know

One job per prompt

Break big asks into steps. 'List user goals' first, then 'suggest flows for goal 2'.

Show an example

A reference screenshot or a sample of the style you want beats a long description.

Iterate, don't restart

Reply with specific fixes: 'make step 2 shorter, keep the rest'. Treat it like a conversation.

Keep a prompt library

Save prompts that worked in a doc. Your team can reuse and improve them.

See it in code

Walk through the code

text

1Act as a senior mobile UX designer.2Users: first-year students, many on budget Android phones,3using the app between classes.4Task: suggest 3 different onboarding flows for a5campus events app.6Constraints: max 3 screens each, skippable, works7one-handed, WCAG AA contrast, no account needed to browse.8Format: for each flow give a name, the screens in order,9and one risk to test with users.

1/4Role and context: who's answering and who it's for.

Role, context, task, constraints and format, each on its own lines. The constraints do the most work in steering the answer.

Save yourself the pain

Common mistakes

Oops

Writing one vague sentence and blaming the tool for generic results.

Do this instead

Add users, context, constraints and format. Most bad outputs come from missing information.

Oops

Pasting private user data or company plans into a public AI tool.

Do this instead

Anonymise and check your team's rules on which tools you're allowed to use.

Your mission

Vague vs detailed

~20 min

Quick check

Your AI output keeps ignoring that the app is for older users. What's the best fix?

Go deeper

1Practice iterating on a design-related prompt 3-4 times to see how wording changes shift output.
2Learn to give AI tools specific constraints (style, platform, brand) rather than open-ended requests.
3Build a small library of prompt templates for recurring design tasks.