Copy-paste prompts for the thinking layer of design — research plans and interview scripts, personas and user flows, wireframe structure, microcopy that doesn't sound like a robot, and the design rationale you need before a stakeholder review. Works with ChatGPT, Claude and Gemini. No signup — just tap Copy.
Good design is 20% pushing pixels and 80% deciding what to build, why, and how to explain it. That other 80% — research plans, personas, flows, copy, and defending a decision to a PM — is mostly reading and writing, which is exactly where an AI assistant earns its keep. It won't replace your taste or your user, but it will get you past the blank canvas fast. Each prompt below hands the model a role, your real context, and the exact artifact you need. Swap the [brackets] and go.
Understand the user before you open Figma — plans, scripts and synthesis.
Best for: getting real insight instead of leading the witness.
Act as a senior UX researcher. Write a 30-minute user interview script for the study below. Structure it: warm-up, background on their current workflow, open questions about the problem (past behavior, not hypotheticals), and a graceful wrap-up. Every question must be open and non-leading — flag and rewrite any that hint at the answer I want. Add 2 follow-up probes per key question and note where to shut up and listen. WHAT I'M TRYING TO LEARN: [the research goal / decision this informs] WHO I'M TALKING TO: [user type, their context]
Best for: turning messy research into one usable persona.
Turn my raw research notes below into a single, evidence-based persona — not a demographic cliché. Include: goals, the job they're hiring the product to do, their current workaround, top frustrations, and the decisive moment where they'd switch or churn. Ground every trait in something from my notes; if a section isn't supported by the data, say "not enough evidence" instead of inventing it. End with the one insight that should most change my design. MY RESEARCH NOTES: [paste interview notes, support tickets, survey quotes]
Best for: the pile of transcripts you're supposed to "find patterns" in.
You are helping me synthesize qualitative research. From the notes below, cluster what users said into 4–6 themes. For each theme give: a plain-language name, how many participants it came from, 1–2 representative quotes, and the design implication. Separate what users literally said from your interpretation. At the end, list the 3 findings strong enough to act on now and the ones that need more research. RESEARCH DATA: [paste your notes/quotes across participants]
Best for: framing a feature around motivation, not features.
Based on the context below, write 5 Jobs-To-Be-Done statements in the form "When [situation], I want to [motivation], so I can [expected outcome]." Focus on the underlying progress the user is trying to make, not the UI. Then, for the strongest JTBD, list 3 ways today's solution fails them — those are my design opportunities. PRODUCT + USER CONTEXT: [what the product does, who uses it, the problem area]
These are the free taster. 25 AI Prompts for UI/UX Design gives you the complete, categorized set — research scripts, persona and flow builders, wireframe planners, microcopy generators, accessibility checks and critique frameworks — each with fill-in-the-blank variables and usage notes. One download, yours forever.
Get the UI/UX Design Pack — $5 → Instant download · 25 prompts across research, flows, copy & systems · 100% money-back guaranteeInformation architecture, user flows and wireframe skeletons — before the visuals.
Best for: seeing every step and dead-end before you design screens.
Act as a product designer. Map the end-to-end user flow for the task below as a numbered step list. For each step note: the user's goal at that moment, the decision or input required, the happy path, and at least one edge case or failure state (empty, error, no-permission, offline). Flag the single step most likely to cause drop-off and suggest how to de-risk it. Keep it structural — no visual design yet. THE TASK / FLOW: [what the user is trying to accomplish] CONTEXT: [platform, constraints, entry point]
Best for: deciding what goes on a screen and in what priority.
Give me a low-fidelity content blueprint for the screen below. List the sections top to bottom in priority order, and for each: its single purpose, the content/components it holds, and the one action it should drive. Rank by what the user needs first, not what the business wants to show. Note anything that's competing for the same attention and should be cut or demoted. No colors or styling — structure and hierarchy only. THE SCREEN: [what screen, for whom, the primary user goal]
Best for: a nav or menu that feels off but you can't say why.
Review the navigation / information architecture below as a UX architect. Point out: labels a first-time user might misread, categories that overlap or don't match a real mental model, anything buried too deep, and where a card-sort might reshuffle things. Suggest a cleaner grouping and clearer labels, and explain the reasoning in plain language a PM would accept. CURRENT IA / NAV STRUCTURE: [paste your menu, sitemap, or category list] WHO USES IT + THEIR TOP TASKS: [context]
The microcopy, labels and messages that carry more weight than the layout.
Best for: buttons, errors and empty states at a moment of doubt.
Act as a UX writer. For the moment below, write microcopy that meets the user where their head actually is. Give me: the primary button label, a one-line supporting sentence, and — if relevant — an error and an empty-state message. Write in the user's voice, name the thing they're worried about, and tell them exactly what happens next. No "Oops, something went wrong," no cutesy filler, no jargon. Give 2 options per item with a note on the tradeoff. THE MOMENT: [what screen/action, what the user fears or is unsure about] PRODUCT VOICE: [tone — plain, warm, formal, playful]
Best for: a technical error that blames or confuses the user.
Rewrite the error message below so it's helpful, not scary. It should: say what happened in human terms, avoid blaming the user, and give a concrete next step to recover. Drop error codes into a "details" aside, not the headline. Keep it short and calm. Give me the headline, the body line, and the button label. CURRENT ERROR: [paste the message and when it appears] WHAT ACTUALLY WENT WRONG + HOW TO RECOVER: [the real cause and fix]
Best for: turning a blank first screen into a first success.
Write the copy for a first-run empty state for the feature below. Its job is to get the user to their first small win, not to explain everything. Give me: a one-line headline that frames the value, a single sentence of what to do first, and one primary action label. Optionally a "why this matters" line under 12 words. Encourage, don't overwhelm — one clear next step only. THE FEATURE + FIRST WIN: [what it does, the smallest valuable action a new user can take]
Type, color, components and accessibility decisions you can defend.
Best for: a consistent system instead of random font sizes.
Act as a design systems engineer. Propose a typographic scale and spacing system for the product below. Give me: a modular type scale (name, px/rem, weight, line-height, and where to use each), a base spacing unit with a step scale, and 2 sentences on the ratio you chose and why. Keep it small enough to actually enforce — no 14-size scale nobody follows. Output as a clean table. PRODUCT + PLATFORM: [web/app, density, brand feel — editorial, dense-dashboard, playful, etc.]
Best for: defining a reusable component's variants and states.
Help me spec the component below for a design system. Define: its single responsibility, the props/variants it needs (and the ones it should NOT absorb to avoid becoming a god-component), all interactive states (default, hover, focus, active, disabled, loading, error), and content guidelines. Suggest a clear, boring, searchable name. Flag where this overlaps an existing pattern and might be the same component. THE COMPONENT + WHERE IT'S USED: [describe it and its contexts]
Best for: catching a11y issues before dev, not after.
Act as an accessibility specialist reviewing a design (WCAG 2.2 AA). From my description below, list likely issues by severity: color-contrast risks, touch-target sizes, focus order and keyboard access, missing labels/alt text, reliance on color alone, and motion concerns. For each, give the specific fix and the criterion it maps to. End with the top 3 to fix before shipping. THE DESIGN: [describe the screen — colors, text sizes, interactions, components]
Pressure-test the work and explain it so stakeholders say yes.
Best for: finding the holes before your reviewer does.
Be a tough but fair design critic. From my description below, challenge the design: where will users get confused or drop off, which edge cases (empty, error, long content, slow network, first vs. returning user) break the layout, what assumption is it quietly making about the user, and where does business goal fight user goal? For each issue give severity and a concrete fix. Be specific — no generic "add more whitespace." THE DESIGN + GOAL: [what it is, the user goal, the business goal]
Best for: the review where "it looks nicer" won't cut it.
Help me present this design decision to stakeholders. Write a short rationale that: states the user problem, the options I considered, why I chose this one, and how it ties to a metric or user need they care about. Frame tradeoffs honestly and pre-empt the 2 objections a skeptical PM or engineer will raise. Confident, evidence-based, no design jargon. Under 200 words. THE DECISION + CONTEXT: [what I designed, why, what I traded off, the goal it serves]
Best for: "make it pop" and other feedback you can't use.
A stakeholder gave me vague design feedback. Act as a design lead and translate it into specific, testable requests. For each comment: guess the underlying concern, propose 1–2 concrete changes that would address it, and write a clarifying question to confirm before I redo work. Separate real usability issues from personal taste, and flag anything I should push back on with a reason. THE FEEDBACK: [paste what they said] THE CONTEXT: [the design, the goal, who gave the feedback]
This page is the free sampler. 25 AI Prompts for UI/UX Design is the complete kit — research scripts, persona and flow builders, wireframe planners, microcopy generators, accessibility passes, design-system specs and critique frameworks — each with variables and notes. Or grab The Ultimate AI Prompt Vault (68 prompts across business, writing, design & research) if you want everything in one download.
Get the UI/UX Design Pack — $5 → Or the full 68-prompt Vault for $7 with code JULY25 · see the Vault →The designers getting real leverage from AI don't ask it to "design a landing page." They use it for the reading-and-writing 80% of the job — the research plan, the persona draft, the flow you sketch before Figma, the empty-state copy, the rationale doc. These are tasks where a strong first draft in ten seconds beats an hour of staring, and where your judgment still makes the final call. The model handles structure and language; you handle the user, the craft and the decision.
Yes. Nothing here is ChatGPT-specific — they're plain-language instructions, so they work in ChatGPT, Claude, Gemini, Copilot and local models. Paste your context into the brackets and run. They pair especially well with a design tool: generate the flow or content blueprint here, then build it in Figma.
Use it before the artboard, not on it. Let the model draft research questions, cluster your notes, and write the first pass of microcopy — then bring your own eye to the pixels. Never ship AI copy or a persona unedited: read it against a real user you've actually talked to, and cut anything that sounds generic. The prompt gets you 90% of the way; the last 10% of specificity is your design judgment, and that's the part nobody can prompt for you.
Microcopy is high-frequency, high-impact, and easy to get wrong under deadline. A good prompt turns "Oops, something went wrong" into an error message that tells the user what happened and how to recover — in their voice, at the moment they're anxious. Do that across a whole product and you've measurably lowered support tickets and drop-off, which is a much easier ROI story to tell a PM than "it looks cleaner."
These 16 cover the moments most product designers hit every sprint. If you'd rather not rewrite them each time — and want the complete set with fill-in-the-blank variables — the 25 AI Prompts for UI/UX Design pack is five dollars and pays for itself the first time it saves you an afternoon of blank-canvas research or copy. Want everything, across every part of your work? The Ultimate AI Prompt Vault bundles 68 prompts into one organized download.
← 50 more free ChatGPT prompts for everyday work, writing & code · Prompts for developers → · Prompts for content creators → · Compare all prompt packs →