White-Label AI Chatbot for Agencies: Website Chat and WhatsApp Under Your Brand

Short answer: a white-label AI chatbot is an AI agent you set up on a client's website chat and WhatsApp and sell as your own service. It answers from the client's website and documents, captures leads, and hands conversations to the client's staff in a shared inbox, while the console, emails and sign-in page carry your brand and the client pays your price. On ThinnestAI the same agent can also answer phone calls and Telegram, so a chat client can grow into a voice client without switching platforms.

What does a client actually get?

• Website chat: a widget added with a short script tag, which picks up a returning visitor's conversation.
• WhatsApp: the same agent on the client's own WhatsApp number through the official API.
• A shared inbox: every conversation in one place, with human takeover, assignment, saved replies and internal notes the customer never sees.
• Leads: the agent records a name, phone, email and what the person wants, even partial details, linked back to the conversation.
• One memory: a customer who chatted on the website and later messages on WhatsApp is recognised, with records linked by phone number rather than merged.

How do you build one for a client?

1. Load the knowledge base. Crawl the client's website or import its sitemap, then add pasted text and files such as price lists and policies.
2. Write the instructions. Tone, what to say about prices, what never to promise, and when to hand over. Hand-over triggers such as "can't answer" and "customer is upset" send the conversation to a person.
3. Connect the tools it needs. Shopify for order status and products, Cal.com for bookings (Google Calendar through Cal.com), or the client's own API through tool calling.
4. Install the widget. The snippet the console gives your client points at your domain, not ours.
5. Connect WhatsApp. The client connects its number through Meta's sign-up flow inside your console.
6. Test it on real questions. Ask the ten questions the client's staff hear most, then the awkward ones, before it goes live.

How is this different from a flow-builder chatbot?

Many chatbot builders ask you to draw a decision tree: a button for each question and a box for each answer. This agent works differently: it reads the question, looks in the client's knowledge base and answers in its own words, and you shape it with written instructions in a settings console. There is no drag-and-drop flow canvas.

If a client needs a rigid scripted menu, such as a regulated disclosure word for word, a flow builder may suit them better.

What will the client see of your brand?

The console runs on your domain with your logo, invitations and notifications come from your email sender, and the sign-in page is yours, with your own privacy policy and terms. One thing to know: the ready-made store plugins carry the platform's own name, so on your domain clients are shown the script-tag install instead.

Which languages does it reply in?

Chat and WhatsApp replies cover 37 languages, including English, Spanish, French, German, Hindi and all 22 scheduled Indian languages. Arabic is not one of them for chat: it is available on voice calls, and WhatsApp template messages can be written in Arabic. If you sell to Gulf clients, set that expectation early.

How do you sell and price it?

You set your own prices: a markup on usage such as AI replies and WhatsApp messages, and an amount on plans. Clients get their own workspace and a free trial you size, and pay you through your own payment account, with automatic top-up so the agent keeps answering. We charge you our rates for their usage and the difference is yours. Many agencies start a client on chat, show the transcripts after a month, and then add phone calls to the same agent.

Next step

The agencies page covers white label. For what the agent does, see chat AI agent, WhatsApp AI agent, knowledge base and shared inbox. Selling in the Gulf? Start with the UAE page.