I asked ChatGPT to help me design — here’s what worked

Understanding the language of AI in design can set you ahead of many designers. I still remember when AI tools started gaining traction in the design world. There was panic, memes, and real fear that designers would soon be replaced by AI. And honestly, it wasn’t entirely wrong to worry. Because here is the thing — some designers won’t be replaced by AI itself. They’ll be replaced by designers who understand the “language of AI” — a term I like using for AI prompts.

Knowing how to craft strong AI prompts has improved my workflow as a designer — whether it’s generating briefs, writing website copy, choosing the right color palettes, or even producing high-quality wireframes. The truth is, AI will give you results based on the input you give it. So, crafting quality prompts will give you quality results.

In this article, I’ll share prompts I use day-to-day as a designer — plus, I’ll give you a simple framework to create your own prompts for any project. And no, you don’t need to be a professional “prompt engineer” to do it.

The AI prompt clarity framework

I start almost all my prompts with ChatGPT. Why? Because it responds really well to structure — especially the framework I’m about to show you. I also use it to shape better prompts for other tools like MidJourney, Relume, and the rest.

Here’s the framework:

• Role — You are a [insert role] with [insert years] of experience working on [insert type of projects or industry].
• Context —  [Briefly describe the scenario, problem, or goal you’re working on]
• Task — [Clearly state what you want the AI to help with — e.g., generate ideas, write a UX brief, critique a wireframe.]
• Tone — [Specify the tone you want — e.g., professional, friendly, concise….]

Role

This defines who or what the AI should pretend to be. Giving it a specific role shapes the type of language, depth, and insight it provides. For example: “You are a UX strategist with 4 years of experience working on mobile productivity apps.”

This helps the AI adopt the right mindset and produce more relevant answers.

Context

Provide essential details about the project, product, audience, or problem you’re tackling. The more context you give, the less guessing the AI has to do. For example, saying, “I have an idea for a habit-tracking app aimed at college students that struggle with time management.” is far more specific than simply saying, “Help me design an app.”

Task

Be specific about what you want. Whether it’s ideas, feedback, or a step-by-step plan, clarity helps the AI focus. For instance, “Help me draft a brief outline for this project…”

Tone

You may want the response to be formal, casual, technical, or friendly, depending on your goals. Specifying tone ensures that the output matches the style you’re aiming for. For example, “Respond in a casual, upbeat tone” or “Use a professional, concise style.”

How to apply the framework (with real prompt examples)

Let’s talk about how this prompt framework actually works. It’s not just about typing random ideas into AI and hoping for magic. The real power lies in structuring your prompt intentionally — assigning a clear role, giving context, setting a task, and guiding it toward a specific result. This structure turns vague instructions into useful, high-quality output.

Here are 8 AI prompts I use following the framework:

1. Crafting UX design briefs

At the start of a project, I sometimes need to create a quick UX brief. Instead of staring at a blank page, I feed some info to ChatGPT and let it generate a structured brief that I can then fine-tune.

Prompt:

• Role — You are a UX strategist with 4 years of experience working on mobile productivity apps
• Context — I have an idea for a habit-tracking app aimed at college students who struggle with time management
• Task — Help me draft a brief outline for this project – include the app’s goal, target users, user needs/pain points, and key features we should include
• Tone — Professional

Result:

It’s not final, of course, but it gives me a strong starting point. In seconds, I get a template instead of a blank canvas.

2. Writing UX microcopy

When I’m working on onboarding flows, buttons, or tooltips, I often use ChatGPT to help me generate friendly, clear UX copy fast.

Prompt:

• Role — You are a UX writer with 3+ years of experience working on mobile banking apps
• Context — The app is for users who want to track spending and manage their money better. I need copy for an error message that appears when someone types the wrong password
• Task — Suggest three variations of helpful error messages that guide the user without sounding harsh
• Tone — Polite, reassuring, and supportive

Result:

This helps me avoid robotic or stiff UX copy. It also gives me variations I can test and refine.

3. Writing website copy

Sometimes, when working on a landing page, designers end up writing the copy too. And when I need a quick tagline or hero message fast, AI helps me move from a blank canvas to a working draft.

Prompt:

• Role — You are a copywriter with 3+ years of experience who specializes in digital products
• Context — I’m working on a landing page for a startup that sells eco-friendly fashion to Gen Z
• Task — Write three hero taglines for the homepage that are short, catchy, and explain what makes the brand stand out
• Tone — Youthful, fun, and bold

Result:

These kinds of quick prompts help me test tone, headline style, and value proposition ideas in minutes, and it’s way better than just dropping in ‘Lorem ipsum.’

4. Generating color palettes based on emotion or industry

When I’m stuck picking colors for a brand or want to explore emotional direction, I use AI to suggest palettes based on feelings or audience expectations.

Prompt:

• Role — You are a senior brand designer with deep color psychology knowledge
• Context — I’m designing a mental health app for teenagers, and I want the UI to feel calming and safe
• Task — Suggest a three-color palette (primary, secondary, accent), and explain why each color fits
• Tone — Clear and professional

Result:

This helps me explore emotional tone in visuals without opening a million Dribbble tabs.

5. Structuring a website from navbar to footer

When I’m designing a site for a new client or industry, I use ChatGPT to suggest structural ideas for the layout and navigation.

Prompt:

• Role — You are a UX designer with experience creating conversion-focused websites
• Context — I’m designing a website for a new language learning platform targeted at working professionals. The brand tone is modern, clear, and trustworthy
• Task — Suggest a full website structure from navbar to footer, including main navigation items, homepage sections, and footer links
• Tone — Clear and professional

Result:

This provides a clear blueprint to work from. I can always adapt it later, but it’s better than starting from a blank canvas.

6. Communicating with other AI

Sometimes, the smartest way to write a great AI prompt… is to let AI help you write it. I use ChatGPT to help me structure better prompts for other AI tools, because ChatGPT understands how to “speak AI” — perhaps even better than most of us.

Let’s say I need a landing page illustration in MidJourney for a budgeting app aimed at Gen Z. I want it to feel soft, modern, and expressive.

Prompt:

• Role — You are an AI prompt engineer with experience crafting image generation prompts for MidJourney
• Context — I want to create a landing page hero image for a budgeting app for Gen Z users. The vibe should be soft, modern, and friendly. I need a lady holding a smartphone with the app
• Task — Write a detailed MidJourney prompt that captures this vibe, including subject, style, mood, and lighting
• Tone — Descriptive and visual
  

Now, ChatGPT will give a very detailed prompt to paste in Midjourney; the new prompt goes like this:

Then Midjourney will output high-quality images based on the new prompt:

This way, ChatGPT doesn’t just assist me in writing; it becomes a creative partner that helps me collaborate better with other AI tools.

Bad prompts give bad results

One thing I’ve learned is that if you give AI vague input, you’ll get vague output. Before I started using the RCTT (Role, Context, Task, Tone) structure, most of my prompts sounded like this:

“I need help with a landing page structure for a new app.”

And here’s the kind of response I got back:

It’s not wrong, but it’s too generic to be helpful. There’s no real understanding of the audience, the product, or the goal.

Now, compare that to what I shared in Prompt #5 under “How to apply the framework,” I used a structured prompt to describe what the landing page was about in a structured detail. The result was a much clearer layout tailored to my actual use case.

Best practices for writing AI prompts

AI doesn’t control itself — it reflects your input. So to get better results, follow these best practices:

• Be clear and specific — Avoid vague instructions like “Write something good.” Instead, clearly describe what you want. The more details you give, the better the AI can understand your intent and deliver relevant, high-quality content that meets your expectations
• Define your audience — Let the AI know who the writing is for. Whether it’s tech professionals, college students, or small business owners, defining the audience helps tailor the tone, vocabulary, and examples to resonate better with the intended readers and create more effective communication
• Set a tone or style — Tone shapes how your writing feels. Ask for a tone that suits your goals, which could be casual, confident, playful, professional, or empathetic. If you don’t specify a tone, the AI might default to something generic or overly formal, which could miss the emotional mark of your message
• Add context or use case — Tell the AI where or how the content will be used, like in a landing page, blog post, tweet, or pitch deck. This helps it structure the writing appropriately and focus on what matters most for that particular setting, making the output more useful and targeted
• Offer examples or comparisons — Guide the AI by referencing styles you like. For example, “Write like Apple’s product page” or “Make it sound like a casual blog post on Medium.” Giving comparisons makes it easier for the AI to match tone, pacing, and style more accurately from the start
• Refine through feedback loops — Don’t expect perfection in one shot. AI thrives on iteration. If the result is close but not quite right, respond with feedback like “Make it more concise” or “Add a sense of urgency.” Each round of feedback improves the result and helps the AI learn your preferences

Other AI tools I use (and when I reach for them instead)

While ChatGPT is my go-to for most of my design-related prompts, I’ve also gotten my hands on some other AI tools that serve different needs and, in some cases, do things better:

Tool	What I use it for	Strengths	Limitations
Claude	Summarizing long documents, synthesizing research notes	Thoughtful, great with nuance and large context	Slower, less creative with open-ended tasks
Gemini	Quick design research, referencing Google/YouTube content	Fast, well-integrated with Google ecosystem	Shallow with complex ideation or structured writing
Jasper AI	Writing marketing copy, landing pages, product descriptions	Optimized for conversion-focused content	Less flexible for exploratory or UX writing
Notion AI	Cleaning up notes, outlining ideas inside Notion docs	Seamless with Notion workflow, good for admin tasks	Limited depth, not great for original thinking

Each tool has its strengths. But for design thinking, prompt crafting, and creativity, I stick with ChatGPT.

Conclusion

AI isn’t here to replace designers — it’s here to empower us. And that power doesn’t just come from using AI tools, but knowing how to talk to them.

The framework I shared (Role, Context, Task, Tone) is how I get consistent, useful results across different tools, whether I’m writing UX briefs, crafting microcopy, or building out landing pages.

• Role — Who the AI is acting as
• Context — What you’re working on and why
• Task — What you want it to do
• Tone — How it should sound

The best part? You don’t need to be a “ChatGPT prompt engineer” to do this. Just a clear thinker who knows what they want — and how to guide AI to deliver it.