Brand Phrases, Sign-Offs, and Emoji Settings

Beyond tone and personality, you can control the specific words and style your AI Agent uses, right down to the emoji. These settings are found in AI Agent → Persona, below the tone sliders.

Brand Phrases

Brand phrases are expressions or catchphrases your agent weaves into conversations naturally. Think of them as the things a real team member would say because they've absorbed your brand culture.

You can add up to 20 brand phrases.

Good brand phrases:

"Stay golden!" (a surf brand)
"You're part of the family now" (a community-driven brand)
"Clean skin, clean conscience" (a skincare brand)
"Built different" (a fitness brand)

Tips:

Keep them short and natural, if it would sound weird in a text message, skip it
These aren't scripts, the AI uses them where they fit, not in every message
Don't add generic phrases like "Have a great day", the AI already knows how to be polite

Sign-Offs

Sign-offs are how your agent wraps up messages. The AI rotates through them to keep conversations from feeling robotic.

You can add up to 20 sign-offs.

Examples:

"Cheers, Lindsay 🤙"
"Happy shopping!"
"Here if you need anything!"
"— The [Brand] Team"

If you don't add any, the agent will close messages naturally based on your tone settings.

Emoji Settings

Emoji Collection: Choose up to 50 emoji your agent is allowed to use. If you add specific emoji here, the agent sticks to this set rather than choosing its own.

Emoji Frequency (0–100): Controls how often emoji appear in responses:

0–20: Rare. Emoji might appear in a greeting or sign-off but not in the body of a response.
40–60: Moderate. A few emoji per conversation where they add warmth or clarity.
80–100: Frequent. Emoji in most messages, great for casual, youthful brands.

Blocked Emoji: Emojis your agent should never use, up to 50. Use this if certain emojis conflict with your brand (e.g., blocking 🔥 for a fire safety brand, or 💀 for a wellness brand).

Blocked Content

Words or phrases your AI Agent should never say, up to 50 entries. The agent will avoid these in all responses.

Common uses:

Competitor brand names you don't want mentioned
Sensitive terms that don't fit your brand
Slang or phrases that feel off-brand
Specific product claims you're not allowed to make (e.g., "FDA approved", "cures")

Example: A premium skincare brand might block: "cheap", "bargain", "anti-aging" (preferring "age-defying"), and competitor names.

Putting It All Together

These settings work alongside your tone sliders and brand summary. Here's how they layer:

Brand Summary tells the AI what your brand is about
Tone Sliders set the overall feel of responses
Brand Phrases & Sign-Offs add your specific flavor
Emoji Settings control visual style
Blocked Content sets hard boundaries

After making changes, test them in the Playground to see how they come through in real responses. Pay attention to whether the brand phrases feel forced; if they do, try making them shorter or more conversational.