Prompt Library

33 ChatGPT Prompts That Handle the Non-Design Parts of Design Work

33 copy-paste prompts

Copy-paste prompts for UX research, UI microcopy, design briefs, client presentations, accessibility audits, and documentation. More designing, less writing about designing.

In short: This page contains 33 copy-paste ready prompts, organized into 6 categories with a description and pro tip for each. The first 5 prompts are free instantly, no signup needed. Hand-curated and tested by the AI Academy team.

Louis Corneloup
By Louis Corneloup · Founder, Techpresso
Last updated ·Hand-curated & tested by the AI Academy team

UX Research & User Insights

6 prompts

Create User Personas from Research Data

1/33

✨ What it does

ChatGPT turns research for [product/project] into 3 behavior-based personas with name, role, goals, and frustrations, not demographic cartoons. Paste the interview quotes and survey notes, then design the next screen for the persona you will ship first.

I've conducted user research for [product/project]. Here's what I know about our users: [Paste research findings: interview quotes, survey data, behavioral observations, demographics] Create 3 distinct user personas that: are based on behavioral patterns (not just demographics), include: name, role, goals, frustrations, behaviors, tools they use, quote that captures their mindset, highlight the key differences between personas (why we need 3, not 1), identify which persona is primary (largest or most valuable segment), and include a "day in the life" scenario showing how they'd encounter our product. Make them feel like real people, not marketing stereotypes.

💡

Pro tip: A persona is only useful if it changes your design decisions. After creating personas, test by asking: "Would this design choice be different for Persona A vs Persona B?" If not, the personas aren't distinct enough.

Design a User Interview Script

2/33

✨ What it does

ChatGPT writes a user-interview script for [product/feature] aimed at [what you want to learn], with warm-ups and questions about what people do, not what they say. Book [remote/in-person, duration] sessions, and keep asking for the last time they did the task.

I'm conducting user interviews for [product/feature]. Research goal: [what you want to learn]. Participants: [who — existing users, potential users, specific segment] Interview format: [remote/in-person, duration] Create an interview script with: warm-up questions (build rapport, understand context), behavioral questions (what they DO, not what they SAY they'd do), task-based questions (walk me through how you currently...), emotional probes (what frustrates you about..., what would make you feel...), specific question types to avoid (leading, hypothetical, compound), and closing questions that surface unspoken needs. Include follow-up probes for shallow answers. Mark which questions are essential vs optional (for time management). Add notes on what to observe beyond verbal responses.

💡

Pro tip: Never ask "Would you use this feature?" People say yes to hypotheticals. Instead ask: "Tell me about the last time you faced this problem. What did you do?" Past behavior predicts future behavior.

Synthesize Usability Test Results

3/33

✨ What it does

ChatGPT synthesizes usability notes from [product/prototype] tests with [number] participants into severity-ranked findings, task success rates, and what to fix first. Share the critical list with the team, then schedule the high-severity, low-effort fixes.

I ran usability tests on [product/prototype] with [number] participants. Here are my raw notes: [Paste observations, quotes, success/failure on tasks, completion times, confusion points] Synthesize these results into: top findings ranked by severity (critical/major/minor), task success rates with analysis, common pain points with specific evidence (not just "users were confused"), positive findings (what worked well — don't just focus on problems), and actionable recommendations for each finding. For each recommendation: effort level (quick fix / medium / redesign needed), expected impact on user experience, and how to measure if the fix worked. Present this in a format ready for stakeholder presentation.

💡

Pro tip: Include video clips or screenshots with your findings whenever possible. A 10-second clip of a user struggling is more convincing than any written description.

Map a User Journey for a Key Flow

4/33

✨ What it does

ChatGPT maps the user journey for [specific flow: onboarding, checkout, first-time use, support request, etc.] in [product] with doing, thinking, feeling, and pain points per stage. Walk the live product against that map, then pick one stage to improve.

Map the user journey for [specific flow: onboarding, checkout, first-time use, support request, etc.] in [product]. User persona: [describe the user] Current state: [describe the existing experience, if any] Create a journey map with stages: For each stage: what the user is doing, what they're thinking, what they're feeling (emotional state), touchpoints (where they interact with us), pain points and friction, moments of delight, and opportunities for improvement. Include: before and after the direct product interaction (what triggers them to start, what happens after they finish), the "backstage" processes that affect each stage (systems, teams, policies), and measurable metrics for each stage (conversion rate, time spent, support tickets).

💡

Pro tip: Validate your journey map with real users. Show them the map and ask "Does this match your experience?" They'll correct assumptions you made and reveal stages you missed.

Write a UX Research Plan

5/33

✨ What it does

ChatGPT writes a UX research plan for [project/product decision] tied to [describe what the research should inform], with methods, a minimum version, and time, [amount], and team constraints. Run the minimum plan if the timeline is tight, then bring answers to the next review.

I need to conduct UX research for [project/product decision]. The decision we need to make: [describe what the research should inform]. Constraints: timeline [days/weeks], budget [amount], team [who's available] Create a research plan that includes: research objectives (specific questions to answer), recommended methodology and why (usability test, survey, diary study, A/B test, analytics review, card sort, etc.), participant criteria (who to recruit and how many), recruitment strategy, research guide/protocol, analysis approach, deliverables and timeline, and how findings will be used (connect to specific design decisions). If multiple methods would help, prioritize by impact and feasibility. Include a "minimum viable research" version if time is tight.

💡

Pro tip: Five users uncover 85% of usability issues. Don't wait for a 30-person study — a quick test with 5 people is better than no test with perfect methodology.

Create a Competitive UX Analysis

6/33

✨ What it does

ChatGPT compares the UX of [List 3-5 competitors] on [product/feature], covering information architecture, key flows, visual design, microcopy, and mobile. Steal one pattern that works, and note the shared weakness your design can fill.

Analyze the UX of competitors for [product/feature]. Competitors to analyze: [List 3-5 competitors] For each competitor, evaluate: information architecture (how content is organized), key user flows (onboarding, core action, conversion), visual design and brand expression, microcopy and tone of voice, mobile responsiveness and performance, accessibility, and unique UX patterns worth noting. Create: a comparison matrix rating each competitor on key UX dimensions, patterns that all competitors share (table stakes), differentiators (what each does uniquely well), common weaknesses across all competitors (opportunity for us), and specific recommendations for our product based on the analysis.

💡

Pro tip: Sign up for each competitor's product and complete their key flow as a real user. Screenshots and marketing pages don't show you the actual UX — only using the product does.

Prompts get you started. Tutorials level you up.

A growing library of 300+ hands-on AI tutorials. New tutorials added every week.

Start 7-Day Free Trial

UI Microcopy & Content Design

6 prompts

Write UI Microcopy for a Feature

7/33

✨ What it does

ChatGPT writes microcopy for [describe the feature/screen/flow] in your [product name and type] voice, covering buttons, errors, empty states, and alternatives. Pick one option per element, then paste the winners into the design file.

Write the microcopy for [describe the feature/screen/flow]. Product: [product name and type]. Brand voice: [describe — friendly, professional, playful, etc.]. Elements needing copy: [list all UI elements: buttons, labels, tooltips, empty states, success messages, error messages, onboarding steps, etc.] For each element: write the copy (concise, clear, action-oriented), explain why you chose these words, provide 2 alternatives, and flag any copy that needs A/B testing. Follow these principles: use the user's language (not internal jargon), be specific (not "something went wrong" — say what went wrong), lead with the verb for buttons ("Save changes" not "Changes saved"), and keep it scannable (users don't read, they scan).

💡

Pro tip: Read the microcopy out loud as if you're telling a friend how to use the feature. If it sounds awkward spoken, it'll feel awkward on screen.

Write Error Messages That Actually Help

8/33

✨ What it does

ChatGPT writes error messages for [product/feature] from [List each error: what went wrong, when it happens, what the user was trying to do] that say what happened and the next step, without blame. Drop them into the product, then test each error path once.

Write error messages for [product/feature]. Here are the error states: [List each error: what went wrong, when it happens, what the user was trying to do] For each error, write a message that: explains what happened (in human language, not error codes), tells the user what to do next (specific action), doesn't blame the user ("we couldn't process" not "you entered incorrectly"), maintains brand voice even when things break, and includes a recovery path (button, link, or instruction). Create three tiers: inline validation (instant, brief), form-level errors (after submission), and system errors (server issues, outages). For system errors, include: what information to show, whether to suggest retrying, and when to show a support contact.

💡

Pro tip: The best error messages prevent the error from happening again. "Password must be 8+ characters with a number" at the time of entry is better than "invalid password" after submission.

Design Empty States That Drive Engagement

9/33

✨ What it does

ChatGPT designs empty-state copy and a visual note for screens in [product] that have no data yet, each with a headline, body, and a clear next action. Add the call to action to the blank screen, then make sure it starts the real task.

Design empty state content for these screens in [product]: [List screens that have empty states: dashboard with no data, inbox with no messages, search with no results, etc.] For each empty state: write the headline and body copy, suggest an illustration style or visual, include a clear call-to-action (what should the user do?), make it feel encouraging not empty, and explain the strategy (education, motivation, or guidance). Create two versions: first-time user (they've never used the feature) and returning user with no current content (they know the feature but there's nothing to show). Empty states should feel like opportunities, not dead ends.

💡

Pro tip: Include an example or sample data in empty states when possible. Showing users what the screen looks like with data helps them understand what to create and motivates them to fill it.

Write Onboarding Flow Copy

10/33

✨ What it does

ChatGPT writes onboarding copy for [product] across [number] steps, with titles, under-30-word body, buttons, and why each step matters for [key activation metric]. Put the copy on the screens, then cut any step that does not move that metric.

Write the copy for an onboarding flow for [product]. The flow has [number] steps: [Describe each step: what the user needs to do and why] Target user: [describe — technical level, motivation, context] Goal: get users to [key activation metric] For each step: write the screen title, body copy (under 30 words), button label, help text where needed, and progress indicator copy. Overall principles: keep it progressive (don't overwhelm with everything at once), connect each step to a benefit ("Add your team so you can collaborate in real-time"), allow skipping where possible (with a subtle note about why they shouldn't), and celebrate completion (the final screen should feel rewarding). Include a "skip-all" path for power users who don't need onboarding.

💡

Pro tip: Measure completion rate for each step. The step with the biggest drop-off is the one to fix first. Usually it's because the benefit isn't clear or the effort feels too high.

Create a Writing Style Guide for UI Copy

11/33

✨ What it does

ChatGPT builds a UI copy style guide for [product name] from your brand personality and audience: voice, capitalization, punctuation, and terms. Share it with whoever writes UI, then check the next screen against those rules.

Create a UI copy style guide for [product name]. Brand personality: [describe]. Target audience: [describe]. The style guide should cover: voice and tone guidelines (with examples), capitalization rules (sentence case, title case, when to use each), punctuation rules (periods in buttons? Oxford comma?), terminology preferences (sign in vs log in, remove vs delete, etc.), number formatting (numerals vs words, dates, currencies), do's and don'ts with before/after examples, microcopy patterns for common elements (buttons, tooltips, errors, confirmations), and accessibility writing guidelines (alt text, aria labels, screen reader copy). Include 10+ before/after examples showing the style guide in action.

💡

Pro tip: Start with the 10 most common UI patterns in your product and write definitive copy for each. Consistency on those 10 patterns covers 80% of the user experience.

Write Notification and Email Copy for Transactional Messages

12/33

✨ What it does

ChatGPT writes transactional email, push, and in-app copy for [product] from [List each notification: trigger event, channel (email/push/in-app), purpose], with short titles and a specific next tap. Kill any notice people would mute, then ship the rest with the right channel.

Write copy for these transactional notifications in [product]: [List each notification: trigger event, channel (email/push/in-app), purpose] For each notification: subject line or push title (under 50 characters), body copy (concise, actionable), call-to-action (specific button/link), tone appropriate to the context (urgent for security, friendly for milestones, etc.), and timing recommendation (when to send relative to the trigger). Follow these rules: only notify if the user needs to do something or would want to know, never be annoying (respect notification fatigue), include enough context that the user doesn't need to open the app to understand, and make the CTA button do the obvious next action. Create a notification priority framework: which notifications are critical vs nice-to-know.

💡

Pro tip: If a notification doesn't require user action and doesn't provide immediate value, it probably shouldn't be a notification. Convert nice-to-know updates into a digest or activity feed instead.

Design Briefs & Documentation

6 prompts

Write a Design Brief for a New Feature

13/33

✨ What it does

ChatGPT writes a design brief for [feature/project] covering the product, the business goal, the user problem, scope, timeline, and constraints, plus open questions. Hand it to the designer, and do not add a solution they must draw.

Write a design brief for [feature/project]. Context: Product: [product name and description] Business goal: [why we're building this] User problem: [what user need this addresses] Scope: [what's included and excluded] Timeline: [design and development timeline] Constraints: [technical, brand, accessibility, etc.] The brief should include: problem statement (specific, measurable), user stories with acceptance criteria, success metrics (how we'll know the design works), competitive context (how others solve this), technical constraints the designer needs to know, brand and design system requirements, accessibility requirements (WCAG level), and open questions that need design exploration. Write it as a collaboration document — opinionated enough to guide the designer but open enough for creative solutions.

💡

Pro tip: A good brief answers "what" and "why" but leaves "how" to the designer. If you find yourself describing the solution, you're designing by committee instead of writing a brief.

Document a Design System Component

14/33

✨ What it does

ChatGPT documents a design-system component from [name: e.g., Button, Modal, Card, Toast], its variants, and where it appears, including when not to use it and edge cases. Add the page to the system, then send it to the next person who will implement it.

Document this design system component for my team: Component: [name — e.g., Button, Modal, Card, Toast] Variants: [list variants — primary, secondary, destructive, etc.] Used in: [list contexts/pages where it appears] Create comprehensive documentation: component description and purpose, when to use this component vs alternatives, all variants with use cases for each, props/parameters with types and defaults, states (default, hover, active, disabled, loading, error), accessibility requirements (ARIA, keyboard, screen reader behavior), spacing and sizing specifications, responsive behavior, do's and don'ts with visual examples described, and code usage examples. Include edge cases: what happens with very long text, missing data, or error states.

💡

Pro tip: Document the "when NOT to use" section as carefully as the "when to use" section. Misuse of components is the #1 cause of design system inconsistency.

Write Design Handoff Notes

15/33

✨ What it does

ChatGPT writes handoff notes for [describe feature] from files in [tool: Figma/Sketch/XD] and design system [name], covering flow, motion, breakpoints, and intentional deviations. Walk engineering through the notes, and flag every place you left the system on purpose.

I'm handing off a design to developers. The feature is [describe feature]. Design files: [tool — Figma/Sketch/XD] Design system used: [name] Create handoff notes that: walk through the user flow step by step, specify interactions and animations (timing, easing, triggers), call out responsive breakpoints with behavior at each, list all states for interactive elements, specify loading states and skeleton screens, describe error handling and edge cases, note accessibility requirements (focus order, aria labels, color contrast), highlight where the design deviates from the design system (and why), and list any assets that need to be exported. Organize by screen/component for easy developer reference.

💡

Pro tip: Sit with the developer for 15 minutes after handing off. A brief walkthrough catches more questions than any document, and saves hours of Slack messages later.

Create a Design Decision Log

16/33

✨ What it does

ChatGPT logs design decisions from [project/feature] with the choice, alternatives, rationale, and evidence for each call you list. File that log where the team can find it, then point to it the next time the same fight returns.

I need to document the design decisions made during [project/feature] design. Key decisions: [List each decision: what you decided and the alternatives considered] For each decision, document: the decision made, alternatives considered (with screenshots or descriptions), rationale (why this option won), evidence supporting the decision (user research, data, best practices), trade-offs acknowledged, stakeholder who approved, and conditions under which we'd revisit this decision. Create a format that's easy to reference months later when someone asks "why did we do it this way?" Include a summary table for quick scanning.

💡

Pro tip: Log decisions in real time, not retroactively. The rationale for decisions made 3 months ago is fuzzy. Decisions logged the day they're made capture the full context.

Write an Accessibility Audit Report

17/33

✨ What it does

ChatGPT runs a conceptual WCAG 2.1 [AA/AAA] audit on [product/feature] from your UI description, covering perceivable, operable, and the rest. Fix the critical issues in the file before development, then recheck contrast yourself.

Conduct a conceptual accessibility audit for [product/feature]. I'll describe the current design: [Describe the UI: layout, interactive elements, colors, content structure, forms, navigation] Audit against WCAG 2.1 [AA/AAA] standards: perceivable (color contrast, text alternatives, captions, adaptable content), operable (keyboard accessibility, focus management, timing, navigation), understandable (readable, predictable, input assistance), and robust (compatible with assistive technology). For each issue found: severity (critical/major/minor), WCAG success criterion violated, specific location in the design, recommended fix, and implementation effort. Include a summary score and prioritized fix list. Note: this is based on design description — a full audit requires testing with assistive technology.

💡

Pro tip: Run a real accessibility test with a screen reader after development. Design audits catch structural issues but miss implementation details like missing aria labels or broken focus traps.

Write a Design Case Study

18/33

✨ What it does

ChatGPT writes a portfolio case study from project [name], [who], [what I did], [timeline], and [results, metrics, impact], focused on how you thought, not only the screens. Cut the fluff, then add one metric you can stand behind.

Help me write a design case study for my portfolio. Project: Project name: [name] Company/client: [who] My role: [what I did] Duration: [timeline] Outcome: [results, metrics, impact] Structure the case study: hook (compelling opening that summarizes the impact), challenge (the problem in business and user terms), process (my approach — research, ideation, testing, iteration), key design decisions (the interesting choices, not every screen), results (measurable outcomes), and reflection (what I learned, what I'd do differently). For each section: suggest what visuals to include, what to emphasize for hiring managers, and the ideal word count. Write it in first person, showing my thinking process — not just the final output. This should demonstrate how I think, not just what I made.

💡

Pro tip: Include one decision where you were wrong and how you corrected course. It shows maturity and real design thinking — not just a polished highlight reel.

Like these prompts? There are full tutorials behind them.

Learn the workflows, not just the prompts. 300+ easy-to-follow tutorials inside AI Academy — and growing every week.

Try AI Academy Free

Design Thinking & Problem Solving

5 prompts

Run a Design Sprint Remotely

19/33

✨ What it does

ChatGPT plans a [3/5]-day design sprint on [problem/opportunity] for a [remote/in-person/hybrid] team, with a timed agenda and facilitation notes for [Figma/Miro/FigJam/etc.]. Book the days, then run the activities as timed, not as an open meeting.

Help me plan a [3/5]-day design sprint for [problem/opportunity]. Team: [who's participating, roles] Format: [remote/in-person/hybrid] Tools: [Figma/Miro/FigJam/etc.] The challenge: [describe the problem we're solving] Create a sprint plan with: day-by-day agenda with timing for each activity, facilitation instructions for each exercise, templates and frameworks to prepare in advance, async pre-work to maximize synchronous time, decision-making process for each phase, specific adaptations for remote facilitation (remote sprints need different timing and tools), and a testing plan for the final prototype. Include: icebreakers, energy management, and how to handle dominant voices and quiet participants.

💡

Pro tip: Cut every activity by 20% from the standard sprint timing. Remote attention spans are shorter. Better to finish early and have energy for discussion than to run overtime.

Generate Design Concepts from Constraints

20/33

✨ What it does

ChatGPT generates 5 design approaches for [feature/problem] inside your technical, brand, business, user, and [when it needs to ship] constraints, including why each might fail. Pick one direction you can ship, and drop the pretty option that breaks a constraint.

I need design concepts for [feature/problem]. Constraints: Technical: [what's technically possible/impossible] Brand: [design system, visual guidelines] Business: [revenue model, stakeholder requirements] User: [research insights, accessibility needs] Timeline: [when it needs to ship] Generate 5 distinct design approaches that all work within these constraints. For each concept: describe the approach in 2-3 sentences, explain the core interaction pattern, identify which constraint it optimizes for (and which it sacrifices), user experience trade-offs, implementation complexity (low/medium/high), and a "why it might fail" honest assessment. Then recommend which 2 concepts to prototype and test, and why.

💡

Pro tip: Present multiple concepts to stakeholders, not just your favorite. It shifts the conversation from "approve/reject" to "which direction" — which is more productive and gives you design authority.

Evaluate a Design Against Heuristics

21/33

✨ What it does

ChatGPT scores the design in [Describe the design in detail: screens, flows, interactions, copy] against Nielsen's 10 heuristics with examples, severity, and a 1-5. Fix major and catastrophic issues first, then ignore cosmetic nits until those are gone.

Evaluate this design against Nielsen's 10 usability heuristics (or specify another framework): [Describe the design in detail: screens, flows, interactions, copy] For each heuristic: score (1-5) with justification, specific examples of compliance and violation, severity of any violations (cosmetic/minor/major/catastrophic), and recommended improvements. Heuristics: visibility of system status, match between system and real world, user control and freedom, consistency and standards, error prevention, recognition rather than recall, flexibility and efficiency, aesthetic and minimalist design, help users recognize/diagnose/recover from errors, and help and documentation. Provide an overall score and the top 3 improvements that would have the biggest impact.

💡

Pro tip: Heuristic evaluation catches about 50% of usability issues. Use it as a quick filter before investing in user testing, which catches the other 50% that requires real users.

Design for Edge Cases and Error States

22/33

✨ What it does

ChatGPT lists empty, loading, error, and permission states for [feature] from [Describe the ideal user flow], plus how each should look. Design those states in the file, then walk them before you call the feature done.

I'm designing [feature] and need to think through edge cases. The happy path is: [Describe the ideal user flow] Help me identify and design for: empty states (no data, first use), loading states (slow connection, large data), error states (validation, system, network), permission states (unauthorized, expired session), extreme content (very long names, many items, no items), interrupted flows (back button, refresh, abandoned process), multi-device scenarios (started on mobile, continued on desktop), and accessibility edge cases (keyboard-only, screen reader, reduced motion). For each: describe the scenario, explain the user's mental state, and suggest the design approach. Prioritize by likelihood and impact.

💡

Pro tip: Design the error state before the happy path. If you know how to handle failure gracefully, the happy path design becomes easier because you've already thought through the complete system.

Conduct a Design Critique

23/33

✨ What it does

ChatGPT critiques [feature/screen] against [what this design is trying to achieve], using what works, what to improve, and what to explore. Share the three lists with the designer, then agree on one change to make this week.

Help me structure a design critique for [feature/screen]. The design: [Describe the design or paste a detailed description] Design goal: [what this design is trying to achieve] Critique the design on: does it solve the stated problem? Is the visual hierarchy guiding attention correctly? Is the information architecture intuitive? Are the interactions clear and predictable? Does the copy support the user's task? Is there unnecessary complexity that could be removed? How does it perform at different viewport sizes? Are there accessibility concerns? Provide feedback in the format: "What's working well," "What could be improved," and "Questions to explore." Be specific — "the CTA button gets lost" is useful; "improve the visual hierarchy" is not.

💡

Pro tip: Always lead with what's working. Critique sessions that start with problems put designers on the defensive and shut down creative thinking.

Client & Stakeholder Communication

5 prompts

Present a Design Concept to Stakeholders

24/33

✨ What it does

ChatGPT builds a stakeholder presentation for [feature/project] aimed at [audience: executives/product managers/engineers/clients], with the concept, decisions, research, and likely pushback. Rehearse the business outcome first, then show the screens after the why.

I need to present my design concept for [feature/project] to [audience: executives/product managers/engineers/clients]. The design: [describe the concept] Key decisions: [the important choices you made] Research backing: [data supporting the design] Concerns I anticipate: [what stakeholders might push back on] Create a presentation outline: opening that connects design to business goals, problem statement with user evidence, design walkthrough organized by user story (not by screen), rationale for key decisions (connect each to data), prototype demo talking points, tradeoffs and alternatives considered, metrics we'll track to measure success, and next steps and timeline. Include speaker notes and anticipated questions with responses. The presentation should work for a non-designer audience.

💡

Pro tip: Never present a design without context. Show the problem first, the user pain second, then the solution. A design presented without context looks like a subjective choice.

Write a Design Proposal for a Client

25/33

✨ What it does

ChatGPT writes a design proposal for [client/project] covering the client, [what they want designed], [their deadline], and [range or "to be proposed"], starting with proof you understood the problem. Send the proposal after you rewrite the challenge in their words.

Write a design proposal for [client/project]. Client: [describe the client and their business] Project scope: [what they want designed] Timeline: [their deadline] Budget: [range or "to be proposed"] Create a proposal with: executive summary (1 paragraph), understanding of the client's challenge (prove you listened), proposed design approach and methodology, project phases with deliverables at each, timeline with milestones and review points, team and roles, pricing structure (retainer/project/hourly with justification), terms and conditions highlights, and portfolio examples relevant to this project. Write in a tone that builds confidence: we understand your problem and we've solved similar ones before. Include a "why us" section that doesn't sound like a generic pitch.

💡

Pro tip: Mirror the client's language in the proposal. If they said "customers" don't say "users." If they said "engagement" don't say "retention." Matching their vocabulary builds trust.

Handle Design Feedback Without Losing Your Mind

26/33

✨ What it does

ChatGPT sorts feedback from [stakeholder role] into insight, taste, business, misunderstanding, and out of scope, then drafts replies. Answer the business concerns first, and park taste notes for a later pass.

I received feedback on my design from [stakeholder role]. Their feedback: [Paste the feedback — the good, the bad, and the "make the logo bigger"] Help me: categorize each piece of feedback (actionable design insight / subjective preference / valid business concern / misunderstanding / out of scope), craft professional responses for each, identify which feedback to accept, which to push back on (with reasoning), and which to explore further, write a response email that: thanks them for engagement, clarifies misunderstandings, proposes solutions for valid concerns, diplomatically addresses subjective preferences, and sets expectations for the next review. Tone: collaborative, not defensive. The goal is alignment, not winning an argument.

💡

Pro tip: When someone says "I don't like it," ask "can you tell me more about what's not working for you?" The real concern is usually something specific and fixable. "I don't like it" is never the actual feedback.

Justify a Design Decision with Data

27/33

✨ What it does

ChatGPT builds a case that [your recommendation] beats [their preference] for [stakeholder], using [the feature, the users, the business goal] and a way to validate. Send the case, then run the small test it names before anyone redraws everything.

I need to convince [stakeholder] that my design decision is the right one. They want [their preference] but I believe [your recommendation] is better. My recommendation: [describe your design choice] Their preference: [describe what they want] Context: [the feature, the users, the business goal] Build a case that: starts with shared ground (what we agree on), presents evidence for my recommendation (research data, best practices, competitive analysis), honestly acknowledges the merits of their preference, shows the risks of their preference (with specific examples or data), proposes a way to validate (A/B test, user test, phased rollout), and offers a compromise if one exists. Don't frame it as "I'm right, you're wrong." Frame it as "here's what the data suggests, and here's how we can verify."

💡

Pro tip: If you can't articulate WHY your design is better beyond gut feeling, maybe the stakeholder's suggestion deserves more consideration. Data-backed conviction is persuasive. "I just know" is not.

Write a Design Review Summary for Remote Teams

28/33

✨ What it does

ChatGPT writes an async design-review summary for [feature name] at [concept/wireframe/high-fidelity/pre-dev], with decisions, rationale, open questions, and a respond-by date. Post it the same day, then close comments when that date hits.

Summarize a design review session for team members who couldn't attend. Feature reviewed: [feature name] Design stage: [concept/wireframe/high-fidelity/pre-dev] Attendees: [who was there] Key discussion points: [Summarize what was discussed] Write a review summary that: captures every decision made with rationale, documents feedback received and how it'll be addressed, lists open questions and who's responsible for answering them, identifies changes to be made before the next review, includes before/after descriptions for any agreed-upon changes, and sets the date and scope of the next review. Format for async consumption (scannable, linked to design files, clear action items). Include a "respond by" date for anyone with concerns.

💡

Pro tip: Record important design reviews (with permission). A 30-minute recording is worth more than a 3-page summary for team members who want the full context.

Go from copy-pasting to actually mastering AI.

AI Academy: 300+ hands-on tutorials on ChatGPT, Claude, Midjourney, and 50+ other tools. New tutorials added every week.

Start Your Free Trial

Design Systems & Processes

5 prompts

Audit and Improve Your Design Process

29/33

✨ What it does

ChatGPT audits your design process from how briefs arrive, how you design, test, and hand off, plus the pain points, then names what to change. Change one bottleneck this month, and leave the rest of the workflow alone.

Help me audit and improve my design process. Current workflow: [Describe your current process: how briefs come in, how you design, how you test, how you hand off, tools used, team size] Pain points: [what's slow, broken, or frustrating] Team feedback: [what colleagues say about working with design] Analyze: where are the bottlenecks? Where does quality suffer due to process (not skill)? What's duplicated or unnecessary? Where does communication break down between design and [engineering/product/stakeholders]? Suggest improvements: streamlined process with fewer steps, better templates and reusable assets, handoff improvements, feedback loop optimization, and tools that could help. Compare my current process to industry best practices. Be honest about what I'm overcomplicating and what I'm skipping that I shouldn't be.

💡

Pro tip: Ask your engineering and product partners what frustrates them about the design process. The handoff is usually where the most time is wasted, and they see problems you don't.

Plan a Design System from Scratch

30/33

✨ What it does

ChatGPT plans a design system for [product/company] from team size, existing patterns, [React/Vue/etc.], and [Figma/Sketch/etc.], starting with a small V1 your team will actually use. Build V1 only, then add tokens after people are already using the first set.

I need to create a design system for [product/company]. Current state: Product(s): [what exists] Team: [number of designers and developers] Existing patterns: [any consistency that exists] Technology: [React/Vue/etc.] Design tool: [Figma/Sketch/etc.] Create a plan for building the design system: what to include in V1 (minimum viable design system), component inventory (audit what exists and what's needed), token system (colors, typography, spacing, shadows), component priority (which to build first based on usage frequency), documentation strategy (what to document and where), governance model (who maintains it, how changes are proposed), adoption plan (how to get the team actually using it), and metrics to track (adoption rate, consistency, dev time savings). Include a phased roadmap: what to build in month 1, 3, 6, and 12. Don't try to build everything at once — that's how design systems die.

💡

Pro tip: Start with the 10 most-used components, not the edge cases. A design system that covers 80% of use cases and ships in 2 months beats one that covers 100% and ships never.

Create Design Tokens Documentation

31/33

✨ What it does

ChatGPT documents design tokens for [product/design system] from [List: colors, typography, spacing, shadows, border radius, breakpoints, animation, z-index], with names, values, and when to use each. Publish the list where engineers look, then refuse new one-off values.

Document the design tokens for [product/design system]. Tokens to document: [List: colors, typography, spacing, shadows, border radius, breakpoints, animation, z-index] For each token category: list all tokens with their values, explain the naming convention, show usage guidelines (when to use which token), document the semantic meaning (why this color is called "danger" not "red"), specify responsive behavior where applicable, and include do's and don'ts with examples. Create a quick reference card that developers can keep open while coding. Include the implementation format (CSS variables, Tailwind config, SCSS, or design-tool-specific format).

💡

Pro tip: Name tokens by purpose, not value. "color-danger" survives a rebrand. "color-red-500" doesn't. Semantic tokens create a system that can evolve without breaking everything.

Write Design Principles for Your Team

32/33

✨ What it does

ChatGPT writes 5-7 design principles for [product/team/company] from [what we build], [who we serve], and the tensions you named, each specific enough that the opposite would also be a choice. Print the five you will actually enforce, then use them in the next critique.

Create design principles for [product/team/company]. Product: [what we build] Users: [who we serve] Brand values: [company values] Design challenges: [recurring tensions — speed vs quality, simplicity vs power, etc.] Create 5-7 design principles that: are specific to our product (not generic "make it simple"), resolve real design tensions (when two good things conflict, which wins?), are memorable (short enough to remember, interesting enough to reference), include an example of each principle in action, and rank-order them (when principles conflict, which takes priority?). For each principle: the principle statement, a one-line explanation, an example of a design decision it would resolve, and the counter-principle it beats. Test: if the opposite of your principle is something no reasonable person would argue for, it's too vague.

💡

Pro tip: "Make it easy to use" is not a principle — no one argues for "make it hard to use." A real principle is: "Simplicity for new users beats efficiency for power users." Now there's a genuine trade-off being resolved.

Set Up a Design QA Process

33/33

✨ What it does

ChatGPT designs a design QA process from team, [how often you ship], current QA, and [what goes wrong between design and production], including a developer self-check. Run the self-check on the next release, then do the formal review only after that.

Help me create a design QA process for my team. Current situation: Team: [designers and developers] Deployment cadence: [how often you ship] Current QA: [what exists — or nothing] Common issues: [what goes wrong between design and production] Create a QA process that: specifies when to QA (which environments, at what stage), provides a checklist for visual review (spacing, typography, colors, alignment), covers interaction review (animations, transitions, states), includes responsive review (key breakpoints, touch targets), addresses accessibility review (contrast, keyboard nav, screen reader), establishes a bug filing process (severity, screenshots, expected vs actual), and sets SLA for design bug fixes (which get fixed before release vs post-release). Include a "QA-ready" checklist for developers to self-check before requesting design QA. Keep the process lightweight enough to actually follow.

💡

Pro tip: Focus design QA on user-facing impact, not pixel perfection. A 2px spacing difference rarely matters. A missing loading state or broken mobile layout does. Prioritize what users actually notice.

Free tool

Prompt Optimizer

Turn a rough idea into a structured, professional AI prompt.

Try it free →

Frequently Asked Questions

No. ChatGPT handles the text-based parts of design work: research synthesis, documentation, microcopy, design briefs, and stakeholder communication. It cannot create visual designs, prototype interactions, understand visual balance, or apply the craft judgment that comes from design experience. These prompts free you from the administrative overhead (which can consume 40-60% of a designer's time) so you can spend more time actually designing. Think of it as a design ops assistant, not a design replacement.
ChatGPT works with text, not visuals, so it can't review your designs directly. What it CAN help with: evaluating design decisions against heuristics (describe your design, it evaluates), generating design system documentation, writing the rationale for design choices, creating accessibility audit checklists, and structuring design feedback. For visual AI tools, look at Midjourney for concepts, Figma AI features, or Galileo AI — but these generate visuals, while these prompts handle the thinking and writing around design work.
ChatGPT complements rather than integrates with design tools. The best workflow: use ChatGPT for research, writing, and documentation (text output), then apply those outputs in Figma/Sketch for visual design, Miro/FigJam for workshop planning, and Notion/Confluence for documentation. Some designers paste ChatGPT-generated microcopy directly into Figma prototypes for faster iteration. The prompts in this library produce text that can be directly used in any design tool — copy, specifications, component documentation, and handoff notes.
Both. The categories cover the full design spectrum: UX research and user insights (UX-focused), UI microcopy and content design (UI-focused), design briefs and documentation (both), design thinking and problem solving (UX-focused), client and stakeholder communication (both), and design systems and processes (both). If you're a pure UX researcher, the research and journey mapping prompts are most valuable. If you're a UI designer, the microcopy and design system prompts save the most time. Generalists will use everything.
Yes, with important caveats. ChatGPT excels at: writing interview scripts, structuring research plans, synthesizing findings you've collected, creating personas from your data, and planning usability tests. It should NOT be used to: replace actual user research (it doesn't know your users), generate fake research data, or substitute for real user testing. The research prompts in this library help you plan and analyze research more efficiently — they don't replace the research itself. Always validate ChatGPT's suggestions against real user data.

Prompts are the starting line. Tutorials are the finish.

A growing library of 300+ hands-on tutorials on ChatGPT, Claude, Midjourney, and 50+ AI tools. New tutorials added every week.

7-day free trial. Cancel anytime.