white-label AI customer service

White-label here means two things: visual branding (what customers see) and voice and tone (how it sounds). Both are configurable.

Visual branding

The web widget is fully themable, with a live preview as you change settings:

• Eight preset color themes, or custom pickers for primary, background, gradient, in-color text, and hover colors.
• A custom launcher icon you upload (an icon set directly in the embed code takes priority over the panel-configured one).
• Up to four chat starters and teaser text bubbles to invite the first message, plus an optional pulsing-glow effect.
• A configurable persona and tone so the Guru sounds like your brand.

Visitors can rate each answer turn by turn, add a note, undo a rating, and download or delete their own transcript.

Voice and tone

The brand voice is defined in the dashboard in a single free-text field. You describe the tone you want, give examples of how a good reply reads, and set whatever directions matter to you, including language-specific guidance. There is no rigid form to fill out: you write the character in your own words and the Guru speaks in it.

The same voice applies on web chat, email, voice, SMS, and WhatsApp, so a customer who emails after chatting hears the same character.

A worked example

A boutique law firm writes its voice as: "You are Eleanor from Whitman Law. Be formal and precise, never casual. Never offer legal advice, only clarify firm policy. Avoid filler like 'no worries' or 'totally.'" On the visual side they choose a serif heading font, a dark green primary with a gold accent, and the firm crest as the logo.

A visitor opens the chat. The launcher shows the crest, the welcome line reads "Good afternoon. How may I assist you?", and the reply about consultation availability is signed "Eleanor, Whitman Law." If they email later, the reply carries the same voice.

A pet store writes the opposite: a friendly first-name persona, a playful and casual tone, a pet-icon avatar, and a bright orange primary. Same product, completely different feel.

Brand voice across languages

The voice you write applies in any supported language, and you can include language-specific guidance in the same free-text field. The Guru keeps the character you described while replying in the customer's language.

Why the voice stays consistent

Because the brand voice is applied on every reply rather than baked into a model, it stays the same across channels and over time. The Guru answers from your indexed content in the voice you defined, with no fine-tuning that can drift.

Trade-offs

• Heavily stylized voices (e.g., entirely in lowercase, all sentences as fragments) are possible but reduce comprehension on long answers. Keep the persona and tone settings clear so long answers stay readable.

Why this matters

Customers do not distinguish between your brand and the tools you use to serve them. A generic chatbot reply ("As an AI assistant, I can help with...") undoes brand work everywhere else. Treating the Guru as an extension of your team, with your name and voice, keeps the brand intact through every interaction.