The best AI prompts for UX/UI design, 18 to copy (2026)

AI prompts for UX/UI are instructions that tell an AI tool which design task to solve, for whom and within what limits. This library has 18 prompts to copy into ChatGPT, Claude, Copilot, Gemini and UX Pilot, from the user interview to the screen.

Add Founderz as a preferred source

What you’ll take from this

This guide sums up what you will find in the library and how to use it in your daily design work.

• 18 copy-and-paste prompts: UX research, problem statements, personas, user flows, screen briefs and interface generation.
• Prompts built around what each tool does best: ChatGPT, Claude with long documents, Copilot in Word, Excel and Teams, and Gemini in Docs and Sheets.
• The three-layer method (intent, context and constraints) with a fields table and an example of a weak prompt made strong.
• A tool comparison table, a phase-by-phase workflow and the limits of AI in design.

If you want to apply these prompts to real projects, the AI course for creatives at Founderz works through each phase with practical cases. To generate a custom prompt in a minute, use the free generator.

ChatGPT prompts for UX/UI

ChatGPT is the general-purpose tool for the text side of the design process: scripts, synthesis, problem statements and drafts. Each prompt is written to copy and paste; replace what is in brackets with your case and keep the rest. Library reviewed in November 2026.

Prompts for UX research and user interviews

UX research is where AI saves the most visible time: a script that used to take an hour is now ten minutes of prompting and review.

Interview script with open questions

You are a senior UX designer experienced in discovery interviews.
Context: I am researching why new users drop out of the onboarding of [a B2B expense management app]. Audience: [finance managers at small businesses].
Task: write a script of [8] open questions for a [30]-minute interview, with an opening line and a closing line.
Constraints: avoid closed yes/no questions and questions that hint at the answer. Ask about real past behavior, not opinions about the future.
Format: numbered list and, under each question, one italic line with what I want to learn.

Synthesize transcripts into patterns

Act as a UX researcher. I will paste [8] interview transcripts with [field sales reps who manage their schedule from their phone].
Task: identify [4 to 6] recurring behavior patterns.
For each pattern give: a short name, a one-sentence description, how many of the [8] users mention it and a verbatim quote as evidence.
Constraints: do not interpret beyond what the transcripts say, and mark any pattern mentioned by fewer than [3] users as "weak".
Format: table with the columns Pattern, Description, No. of users and Quote.

[paste the transcripts here]

Prompts for defining the problem and creating user personas

Once the research is done, the prompt helps define the problem and draft personas. Knowing how to create a solid persona is the difference between designing for a real person or for an assumption.

Problem statement from research patterns

You are a product designer. Turn these research patterns into [3] problem statements.
Patterns: [paste the 5 patterns here].
Format for each: "As a [user], I need [action] to [goal], but [friction]". Add one line below with the pattern that supports it.
Constraints: do not propose solutions or features, only the problem. Order them from the one that affects the most users to the one that affects the fewest.

Data-based user persona

Act as a UX researcher. Create a draft user persona from this real data: [role, work context, goals, frustrations and tools they use].
Include: fictional name, role, goals, frustrations, a real usage scenario of [a workday with our product] and one representative quote.
Constraints: do not invent any data that is not in the research; if information is missing, write "no data" in that field.
Format: a card of at most [200] words, with one heading per field.

Prompts for user flows and screen briefs

Before opening the canvas, AI helps you structure the flow and prepare the brief for each screen. That way you make your first design choices before moving a single pixel.

Onboarding user flow

You are a senior UX designer. Generate the user flow for [a freelancer signing up for the first time to an invoicing tool] to complete their first invoice on day one.
Business goal: [activate the user on day one].
Constraints: at most [6] steps, mark the potential friction points and the decision points, and do not include email verification.
Format: numbered list of screens; for each, the user action, the system response and the drop-off risk (low, medium or high).

Screen brief with states and accessibility

Act as a product designer. Write the brief for the [sign-up] screen of [an invoicing app for freelancers].
Include: required interface elements, visual hierarchy, content and helper text for each field, and the states (empty, error, loading and success).
Constraints: meet WCAG AA for contrast and keyboard navigation, and respect the components of our design system (primary button, text field and inline error message).
Format: table with the columns Element, Content, States and Accessibility note.

Claude prompts for UX/UI

Claude stands out with long documents and careful drafting. Use it when you have a lot of research material to read in full or a delicate text to write with nuance.

Research report from many documents

You are a senior UX researcher. I am attaching [12] interview transcripts and the notes from [2] usability test sessions of [our booking app].
Task: write a findings report for the product team, with a 5-line executive summary, the [5] main findings ordered by impact and [3] recommendations.
Constraints: each finding has at least two verbatim quotes with the name of the source document; do not claim anything that is not in the documents and point out contradictions between users.
Format: H2 headings per finding, and at the end a list of "questions the research does not answer".

Critical review of a flow with heuristics

Act as a usability evaluator. I am giving you the description of a [an online store purchase flow] with its [5] screens.
Task: evaluate the flow against Nielsen's 10 heuristics and find specific problems, screen by screen.
Constraints: for each problem give the heuristic, the severity (0 to 4) and a proposed fix; do not invent screens I have not described and say when you need more information.
Format: table with Screen, Problem, Heuristic, Severity and Fix.

[describe the flow here]

Interface microcopy and tone of voice

You are a UX writer. Write the texts for [the payment error screen] of [a subscription app]: title, message, text for the primary button and the secondary one.
Context: user [in a hurry, on mobile]. Tone: [friendly, clear and never blaming the user].
Constraints: title of at most [6] words, message of at most [20], no technical jargon or exclamation marks.
Format: 3 numbered versions, and a final line explaining how each one differs.

Copilot prompts for UX/UI

Copilot adds value when it works inside Word, Excel and Teams with your own files and meetings. These prompts assume your version has access to that file or meeting.

Research report in Word from your notes

Use the document [onboarding-research-notes.docx] as the only source.
Write a findings report for the product team: a 5-line summary, [4] findings with one quote from the notes each and [3] next steps.
Constraints: professional tone, at most [2] pages, no information that is not in the notes.
Format: H2 headings and a final table of next steps with owner and date.

Usability survey analysis in Excel

Analyze the table [SUS-responses.xlsx] with [40] responses to a usability questionnaire (SUS scale) for [our app].
Calculate the average SUS score overall and by [user role] segment, find the [3] lowest-rated items and suggest a chart to present them.
Constraints: explain each formula you use and warn me if the sample size of any segment is below [10].
Format: summary table and three conclusions in plain language for someone who is not a designer.

Design review notes from Teams

Using the transcript of the Teams meeting [Design review of the purchase flow], write the meeting notes.
Include: decisions made, comments from the development team per screen, open questions and actions with owner.
Constraints: do not attribute decisions to anyone who did not say them and mark as "pending" whatever was left unagreed.
Format: at most [250] words, with an actions table (action, owner, date).

Gemini prompts for UX/UI

Gemini works best if your team lives in Google Workspace: it can lean on Docs and Sheets to leave the design material in the same place where you share it.

Usability test plan in Docs

You are a UX researcher. Write the plan for a moderated usability test of [the new booking flow of our app] with [6] participants.
Include: goals, participant profile, [4] tasks with a success criterion, closing questions and metrics (success rate, time and errors).
Constraints: [30]-minute sessions, tasks worded without hints about the interface and without technical language.
Format: document with H2 headings and a table for the tasks.

Prioritize usability problems in Sheets

Help me prioritize these usability problems: [paste the list here with description and number of users affected].
Score each from 1 to 5 on impact and on effort to fix, and calculate a priority (impact divided by effort).
Constraints: explain in one sentence why you give each score and point out the [3] quick wins.
Format: table ready to paste into Google Sheets, with the columns Problem, Impact, Effort, Priority and Reason.

Summarize user feedback by theme

Group these [60] user comments about [the booking app] into themes: [paste the comments here].
For each theme give the name, how many comments it includes, the sentiment (positive, neutral or negative) and two verbatim quotes.
Constraints: at most [6] themes, no generic labels like "other" and order from most to least frequent.
Format: Markdown table.

More prompts to generate interfaces with UX Pilot and Figma

Here the type of prompt changes: you describe a screen or a flow and the tool produces a layout, a wireframe or a prototype. These are drafts to iterate on, not deliverables.

Three-step onboarding wireframe

Create a mobile wireframe of the onboarding screen of [a productivity app] with [3] steps.
Include a progress bar, a title and a supporting sentence per step, a primary button "Continue" and a secondary button to skip.
Style: [minimalist, lots of white space, a single accent color]. Clear visual hierarchy and buttons with a tap target of at least 44 px.
Deliver: the 3 screens in high fidelity, with real text and no filler copy.

Dashboard with variants to compare

Design a web dashboard for [sales managers at a small business] showing [monthly revenue, pipeline and pending tasks].
Generate [3] variants with different hierarchy: the first prioritizes the numbers, the second the tasks and the third a plain-language summary.
Constraints: 12-column grid, AA contrast and at most [5] blocks per screen.
Deliver: the 3 variants on a desktop screen, with realistic sample data.

Checkout screen consistent with your design system

Design the checkout screen of [an online fashion store] using this design system: [Inter font, primary color #5045C8, buttons with 8 px rounded corners, 8 px base spacing].
Include the order summary, address form, payment method selection and a final confirmation button.
Constraints: single-column form, inline errors and a secure payment indicator.
Deliver: mobile and desktop versions, with the form error states in one variant.

What AI prompts for UX are and who they are for

An AI prompt for UX/UI is a natural-language instruction that tells an AI model which design task you want to solve and under what conditions. You do not need to code: you need to know how to ask. It helps to separate two types of AI prompts. The first is the text prompt (scripts, research synthesis, problem statements, personas), where ChatGPT and these ChatGPT prompts are the base of almost any UX research workflow. The second is the interface generation prompt: you describe a screen or a flow and the AI produces a layout, a wireframe or a prototype that connects the user and the interface.

The library is built for three profiles. The UX designer who wants to speed up the research phase without losing rigor. The UI designer who wants first screen drafts to iterate faster and improve design quality. And the product designer who coordinates both disciplines, works on digital product design and needs criteria for when to trust what the AI returns. In every case AI does not replace your judgment: it prepares raw material that you review, discard or refine.

The logic is the same for any of the ChatGPT prompts to use for UX design: each AI prompt starts from the task and the context. To use AI to design better means making better design choices, not delegating them. It works just as well for a five-day design sprint, for the whole UX process, for documenting design systems or for preparing UI designs of digital products with AI, always aiming to enhance the user experience with real evidence.

AI prompts by design process phase: from research to UI

UX/UI prompts work best when you assign them to a specific phase of the design process. The design process benefits from AI in its repetitive parts: drafting, synthesizing and structuring. The part that needs judgment, such as which problem to prioritize, which solution fits and which design choices to discard, is still yours. Prompts prepare the material and you decide what to do with it in your UI design.

Design thinking, lean UX and service design with AI prompts

Prompts fit the frameworks you already use. In design thinking they help you empathize and define, in lean UX they take you quickly from hypothesis to test and in service design they map the whole service. UX designers who work this way use AI to prepare each cycle, and evidence from users decides the next step.

How I built these prompts for UX design and how to create your own

I have been writing prompts since 2020, before ChatGPT existed, first for ads and guides and then for the processes of entire teams. Here I apply what I learned to UX/UI: AI does not guess what you do not give it. An effective UX design prompt works when it has three layers: intent, context and constraints. Without them, the AI fills the gaps on its own and returns something generic. This pattern fits well with methods like design thinking and lean UX, where every decision starts from evidence instead of assumptions.

These are the fields I use in every prompt in the library, in this order:

Five tips for writing your own UX/UI prompts

1. Assign each prompt to one phase of the design process (research, definition, flow or screen), never to “design” in general.
2. Paste your real data (transcripts, notes, comments) and say the source is only that material. Always add “do not invent data that is not in the research”.
3. Ask for the format you will use next: a table to paste into Sheets, a numbered list for a flow, a card of at most 200 words.
4. Ask for evidence: verbatim quotes, number of users, a “weak” or “no data” flag. That way you review the output in minutes.
5. Iterate with specific feedback. “Shorten the flow to 4 steps and prioritize the payment screen” works better than starting over.

Example: from a weak prompt to a strong one

Free-form prompts vs. template prompts for UX design

How the prompt changes in ChatGPT, Gemini, Copilot and Claude

• ChatGPT: the best general option for text, attached files and drafts. It works with the prompts in this library as they are and does not generate visual interfaces.
• Gemini: lean on Docs and Sheets so the result lands already in an editable table or document. State the output format clearly.
• Copilot: refer to the specific file or meeting (a .docx, an .xlsx, a Teams meeting) and limit the source: “use only this document”.
• Claude: paste long documents in full (transcripts, reports) and ask for verbatim quotes. It is very useful for careful drafting and for reviewing your own prompt.

Which AI tools should you use for UX/UI design?

There is no single tool for the whole process. ChatGPT handles text and synthesis, UX Pilot generates screens and flows and Figma brings AI into the design environment you already use. The key is knowing what to ask each one. You will find a broader look in our guide to AI tools for UX/UI.

Note: the terms of each plan change often. Check the free version and its limits on the date you use it.

A practical rule: use ChatGPT for everything textual, an interface generation tool for the first pixels and Figma to turn those drafts into real design. Do not force one tool to do all three. This combination also fits neighboring disciplines like service design, where the whole flow, not just the screen, is what matters.

How to bring AI prompts into your design workflow

AI adds the most when you insert it at specific points of the design process, not when you try to make it do everything. This order works for UX designers, UI designers and product designers at any level.

1. define what you want to solve before touching the AI. If you are not clear on the intent, the prompt will not be either.
2. write the prompt with its layers (intent, context, constraints). Start from the research or the problem, never from the final screen.
3. let the AI produce the draft, whether a script, a persona, a user flow or a wireframe.
4. validate the output against your real research and discard what does not fit. This is where you decide.
5. take the approved material to your design tool and adjust it to your design system.
6. go back to the AI with specific feedback to design better on every round.

Limits of AI in UX/UI design: where your judgment still rules

AI speeds up tasks, but there are decisions where your judgment cannot be replaced. Keep them in mind before delegating too much.

• UI generators produce drafts, not deliverables. Always review at pixel-perfect level before taking anything to production.
• Accessibility: decisions about contrast, keyboard navigation or screen readers require specific knowledge that AI does not guarantee.
• AI does not know your commercial, legal or brand constraints unless you tell it.
• a generic output rarely respects your components, tokens and patterns. Adapting it is still manual work.

Frequently asked questions about AI prompts for UX/UI

These are the most common questions from UX designers, UI designers and product designers who start using AI prompts.

What are free-form prompts in UX/UI design?

They are open, conversational instructions with no fixed structure. They work for exploring a poorly defined problem or generating ideas when you do not yet know what you are looking for. When one works well, turn it into a reusable template.

How can ChatGPT help with UX design?

It helps with the text tasks: interview scripts, research synthesis, problem statements and draft personas. It turns hours of repetitive work into minutes of prompting and review. It does not generate visual interfaces; for that use UX Pilot or Figma.

How do you write AI prompts that produce useful designs for real projects?

Use three layers: intent (the task), context (for whom and in what situation) and constraints (format, limits and what to avoid). Avoid “design an app”. The more real context you give about the user and the business goal, the more usable the result.

If AI generates ideas, layouts and content, what is left for the designer?

Judgment is left: which problem to prioritize, which solution fits the business, how to guarantee accessibility and consistency with the design system. AI speeds up execution; validation with real users and final responsibility stay with the professional.

Which AI tools for UX/UI have a free version?

ChatGPT offers a freemium plan, UX Pilot a free plan with generation limits and Figma a free plan with AI features. Terms change often, so check the limits on the date you plan to use them.

Is it reliable to use AI UI generators for real projects?

As a starting point yes; as a final deliverable no. You need to review the result at pixel-perfect level, check accessibility, adapt it to your design system and confirm it fits your business constraints.

Do I need to know how to code to use AI prompts in UX/UI design?

No. Prompts work in natural language: you write what you want the way you would explain it to a colleague. What is worth learning is how to structure instructions and judge what the AI returns, which you learn by practicing on real cases.

Can I use Claude for UX/UI design?

Yes, especially for long documents: synthesizing many transcripts, writing research reports with quotes and reviewing a flow against heuristics. Claude does not generate visual interfaces, but it helps you write the prompt you then paste into UX Pilot or Figma.

Is Copilot useful for UX/UI design?

Yes, if you already work in Microsoft 365: it can draft research reports in Word from your notes, analyze usability surveys in Excel and summarize design reviews in Teams. Its value is working with your own files and meetings.

Your next step with AI prompts for UX/UI

Pick one prompt from the library, change what is in brackets and run your first test with real data from your project. Then change a single field, such as the constraints or the output format, and compare. Working this way, one variable at a time, teaches you what each part of the prompt decides much faster than reading theory.

If you want to apply these prompts to concrete projects from the first modules, the AI program at Founderz is the natural next step. The question is not whether AI will enter your design workflow. It is whether you want to be the one directing it.