Media Kit Contact Submit an AI
UX/UI designer ›

✨ UX Microcopy

Produce in a few minutes dozens of microcopy variants adapted to brand tone and usage context.

Microcopy (buttons, error messages, tooltips, onboarding, empty states) is one of the most powerful — and time-consuming — UX levers. A good "Continue" instead of "Next", an empathetic error message instead of a technical one, a clear tooltip instead of a confusing one can double a flow's conversion. AI lets you produce in minutes dozens of brand-adapted variants, where it used to take hours of team debate. This guide presents workflows that multiply the UX writer or designer's output without diluting brand voice.

Step-by-step Workflow
1
Frame brand tone

Before anything: 3-5 adjectives defining tone (warm/professional, technical/accessible, formal/casual). Plus 5-10 existing microcopy examples that work. Without this frame, AI produces generic output.

2
Define precise context

CTA button in checkout vs onboarding vs confirmation: different tone. Specify: flow stage, user's probable emotional state, expected action.

3
Generate 8-12 variants

Explicitly request multiple variants (short vs descriptive, factual vs friendly, neutral vs marketing). Compare to choose; don't settle for the first suggestion.

4
Test in real context

Paste the chosen variant in the mockup or product. Read in its ecosystem, it might feel too long, too short, or misaligned. Human eye in context beats AI in abstraction.

5
Document in design system

Best formulations join the voice & tone guidelines: rule for buttons, for errors, for confirmations. Long-term consistency to protect.

Copyable Prompts
10 button CTA variants 
You are an experienced UX writer. For this context:nn**Action**: [WHAT USER DOES — e.g., complete order, create account, add to cart]n**Flow stage**: [WHERE IN THE FLOW]n**Probable user state**: [EMOTION / ENGAGEMENT LEVEL]n**Brand tone**: [3 ADJECTIVES]n**Existing microcopy examples from the brand**: [3-5 EXAMPLES]nnGenerate 10 button label variants:n- 3 __short__ versions (1-2 words): factualen- 3 __medium__ versions (3-4 words): with clear valuen- 2 __action-oriented__ starting with action verbn- 2 __benefit-oriented__ centered on resultnFor each variant: (a) the label, (b) the tone it evokes, (c) which persona it would work best for.nnRespect brand tone. No AI tics ('let's dive in', 'let's explore').
Empathetic error messages 
For these error cases:nn[LIST OF ERRORS — e.g., email already used, password too short, payment declined, file too large]nnProduce for each error:n1. **Short title** (1-5 words, factual without dramatizing)n2. **Description** (1-2 sentences): explains without jargon, suggests an actionn3. **Main CTA**: what user can do nown4. **Secondary CTA** if relevant: alternative (contact support, other method)nnPrinciple: acknowledge frustration without creating it, blame system not user, always offer a way out. Brand tone: [DESCRIBE].
Multi-step onboarding 
For this onboarding:nn**Product**: [SHORT DESCRIPTION]n**Audience**: [PERSONA]n**First jobs-to-be-done to enable**: [WHICH OUTCOME IN 5 MIN]n**Planned steps**: [LIST OF STEPS]nnFor each step, generate:n1. **Title** (5 words max, motivating without hype)n2. **Subtitle** (10-15 words, explains benefit in user terms)n3. **Main CTA** (3-5 words, clear action)n4. **'Later' CTA** if skippablenProgression: from 'why' to 'how'. First message hooks, last message celebrates completion. No jargon, brand tone [DESCRIBE].
Engaging empty states 
For this empty screen:nn**Page/section**: [DESCRIPTION]n**Why empty**: [FIRST USE / EMPTY RESULT / TRANSIENT STATE]n**Expected action**: [WHAT USER CAN DO TO FILL IT]nnGenerate 5 empty state variants:n1. **Title** (5 words max, neither sad nor false-cheerful)n2. **Description** (1-2 sentences, explains state + invites action)n3. **CTA** (clear action to exit empty state)n4. **Suggested illustration** (visual description, optional)nnVary the angles: factual, encouraging, light humor (if brand tone allows), educational, solution-focused. Brand tone: [DESCRIBE].
Recommended tools
Claude AI
★ 4.9 (55) · Gratuit

Assistant conversationnel d’Anthropic axé sécurité et contexte long. Excellent pour rédaction, analyse, résumés, code et agents. Interface claire, bons résultats en français.

Why : Excellence sur les nuances de ton et le respect d'une voice & tone guideline détaillée. Suit mieux les briefs riches que les concurrents.

ChatGPT
★ 4.9 (528) · 20 USD/mois

Assistant conversationnel polyvalent d’OpenAI. Rédige, résume, code, traduit et répond à tout type de question.

Why : Punchy pour générer rapidement de nombreuses variantes. Idéal pour A/B testing en série.

🤖
Grammarly
★ 4.7 (152) · 12 USD/mois

Assistant d'écriture alimenté par l'IA qui corrige grammaire, orthographe et style en temps réel sur toutes vos plateformes préférées.

Why : Pour la passe finale : harmonisation, ton, élimination des tics IA. Particulièrement utile en équipe avec plusieurs writers.

Estimated ROI
Time Saved
70-80% on microcopy production (10-20 min vs 1-2h)
Quality Gain
Multiple A/B testable variants, brand tone consistency
Cost
20-30€/month for the stack
Frequently asked questions
Can AI truly respect brand tone?

70-85% if you provide concrete examples in few-shot (3-5 representative paragraphs). Beyond that, some fine-tuning remains. The more codified and exemplified your voice & tone, the more precise AI becomes.

Do you need a UX writer or can AI do it all?

For small teams: AI + attentive designer is enough. For complex products (SaaS B2B with 500 screens, multi-country multi-language app): a UX writer remains valuable for overall vision, cross-flow consistency, design system governance.

How to localize AI-generated microcopy?

Phase 1: generate source version (FR or EN) with clear brand tone. Phase 2: translate in each target language WITH context ('this button appears at checkout end, tone is reassuring'). Phase 3: native speaker review per language. Never literal translation.

Can AI write security-critical error microcopy?

Yes for the frame, but always review and validate: a misleading error message on auth or payment can have serious implications (legal, security, support). Critical microcopy needs human review before deploy.

← Back to guide UX/UI designer