AI writing guidelines

This writing guide is to accompany the existing UX Content and Writing guidelines.

Voice and tone

Genesys Agentic AI is professional yet conversational. Instructions may be technical, but when speaking to the user, they should feel human and helpful. 

Agentic AI speaks in first person, using I/We when necessary, but is not overly personified. An objective companion, not a best friend. We use the present tense, short sentences, and everyday language.


---

Naming AI features

Most features shouldn’t have AI in the name; it waters it down, and as AI becomes more prominent, it will look silly.

Examples of AI-product naming within Genesys Cloud are:

• Virtual Agent
• Virtual Supervisor
• Agent Copilot
• Anomaly Detection

---

UX Writing for AI: Quick Tips

These quick tips are part of a larger effort to expand the AI Writing Guide within the design system’s AI components and documentation, which are still in progress. In addition to these tips, the standards from the Genesys UX Writing & Content Guide continue to apply.

Objective

Helping teams design AI conversations that are clear, consistent, and human-centred.

Voice and tone

Genesys Agentic AI is professional yet conversational. Instructions may be technical, but when speaking to the user, they should feel human and helpful. 

Agentic AI speaks in first person, using I/We when necessary, but is not overly personified. An objective companion, not a best friend.

We use the present tense, short sentences, and everyday language.

Core principles

• Clear – Say exactly what the user needs to do
• Concise – Keep it short and direct
• Consistent – Use the same terms for the same things
• Conversational Tone – Friendly, human, and natural
• Contextual Guidance – Tailor copy to user task and channel

Inclusive and accessible copy

• Use plain language
• Avoid jargon, idioms, or acronyms
• Write with screen readers in mind
• Focus on key information without extra details
• Be mindful of cultural sensitivity in tone and phrasing

Microcopy and prompts

• Error/fallback messages: Be specific and solution-focused—“Unable to add user - Invalid email address”
• Bot choices: Offer clear options. Avoid “Yes/No”: instead, “Send receipt” or “No thanks.”

Error handling and edge cases

• Write clear, empathetic error messages that don’t blame the user.
  E.g., “The email for April May is incorrect or missing. Do you have another email I can try?”
• Anticipate and address user frustration
• Plan explicit fallback steps
• Prepare for “I don’t know” by redirecting to predefined sub‑steps, offering help, or asking for more information.

AI writing dos and don’ts