# White-label AI agent: how agencies sell AI under their own brand

> What a white-label AI agent is, what the end client sees, the minimum platform requirements and how an agency sells AI on WhatsApp under its own brand and domain.

- **HTML version:** https://tolky.to/en/blog/white-label-ai-agent-for-agencies
- **Author:** [Marlos Carmo](https://tolky.to/en/author/marlos), Co-founder of Tolky
- **Published:** November 20, 2026
- **Reviewed:** November 20, 2026
- **Language:** en

## Summary

A **white-label AI agent** is a customer service agent that an agency delivers to clients under its own brand and domain, on top of another vendor's platform. The end client sees only the agency's brand, from login to system emails. To work without risk, the platform must use the official WhatsApp API, run on the agency's domain and give access to every client with one login.

A client asks for "an AI on WhatsApp," and the agency takes the fastest route: it recommends a vendor, sets up the agent and hands over the login. Three months later, the client talks straight to the vendor's support, gets emails with the vendor's brand and starts asking what the agency is there for.

At that point, the relationship belongs to the vendor. Anyone who runs an agency knows how this plays out: the service that should protect the account becomes the reason to renegotiate the retainer. A white-label AI agent flips the order. The technology comes from another company, but the brand, the domain and the relationship stay with your agency.

*Updated November 2026.*

## What is a white-label AI agent?

**A white-label AI agent is a customer service and sales agent that runs on a vendor's platform, but that the end client uses under the agency's brand, domain and product name.** The vendor handles infrastructure, AI models and channel connections; the agency handles the brand, the implementation and the client relationship.

The term comes from consumer goods: a product with no visible brand that the reseller labels as its own. In practice, your agency's client opens `support.youragency.com`, signs in through the agency's login screen and gets system emails with the agency's name. The vendor's name never shows up.

The agent itself is the same as in any [AI agent](https://tolky.to/en/glossary/ai-agent) project: it understands the message, answers with what the business taught it and hands off to a person when the case calls for it. Only the brand changes.

## What does the end client see, and what stays behind the scenes?

**The end client sees the agency's brand on every screen and system message; the infrastructure, the AI models and the WhatsApp API connection stay behind the scenes, run by the vendor.** The split below helps explain the model to the partner who will sell it and to the client who will buy it.

| Layer | What the end client sees | Who handles it |
| :--- | :--- | :--- |
| Address and access | Agency domain with a security certificate, agency login screen | Agency sets it up; vendor hosts |
| Brand | Agency logo, colors, favicon and product name | Agency |
| System emails | Invitations, alerts and password resets with the agency's brand | Agency sets them up; vendor sends |
| AI agent | Replies on WhatsApp and on the website in the client's brand voice | Agency implements; vendor runs the models |
| Channels | The client's official WhatsApp number, webchat on the website | Vendor connects through the official API |
| Support | First-line support from the agency | Agency at tier 1; vendor at tier 2 |

One detail often gets missed: the agent answers under the brand of **the agency's client**. Someone messaging a clinic on WhatsApp talks to the clinic's assistant; the agency's brand appears to the people who manage the operation.

## What are the minimum requirements for a white-label AI agent platform?

**A white-label AI agent platform needs at least five things: the official WhatsApp API, your own domain with branding from login to emails, access to every client with one login, control over the AI before and after publishing, and a clear data processing agreement.** Without any one of them, the agency carries a risk that never shows up in the proposal.

1. **Official WhatsApp API.** A number connected through an unofficial API can be banned, and the client will blame the agency, not the vendor. The risks are covered in [unofficial WhatsApp API](https://tolky.to/en/blog/unofficial-whatsapp-api-risks).
2. **Full domain and branding.** Swapping the logo in the dashboard is not enough. Login, public links, webchat and system emails must come from the agency's domain, or the vendor's brand leaks on the first "forgot my password."
3. **Every client in one login.** With ten clients, one password per account eats up team hours. The agency should enter each account with the same login.
4. **AI with oversight.** Simulate conversations, keep versions and investigate each answer, so the agency can stand behind what the AI says. See [trustworthy AI](https://tolky.to/en/trustworthy-ai).
5. **Data and privacy law.** The end client owns the data, and the agency usually acts as a [processor](https://tolky.to/en/glossary/data-processor) under Brazil's LGPD or equivalent laws. The vendor contract has to reflect that.

There is also a channel rule that applies to every agent on WhatsApp. Since January 15, 2026, the WhatsApp Business Platform terms bar general-purpose AI assistants, where the AI is the product itself; agents that serve a company's customers are still allowed. Selling "a ChatGPT with the agency's brand" open to any question falls in the banned zone. What applies, what Brazil's antitrust authority decided and how to frame the project are in [does WhatsApp ban general-purpose AI chatbots?](https://tolky.to/en/blog/whatsapp-bans-general-purpose-ai-chatbots).

## White label or reselling under the vendor's brand?

**When reselling under the vendor's brand, the agency sells and implements, but the client sees another company's product; with white label, the client sees only the agency, and the agency also owns the final price, the billing and first-line support.** Both models work. The choice depends on how much of the relationship the agency wants to carry.

Reselling under the vendor's brand is the shortest path: the agency uses ready-made materials and charges for its own implementation and operation services. It suits agencies with their first AI clients.

White label takes more structure: the agency sets the price, sends the invoice and handles first-line support. In return, the client buys "the agency's customer service platform," and switching vendors stops being a conversation that happens without the agency.

There is also a third path, referral, where the agency only introduces the vendor and gets paid for it. The full comparison of the three models, with who bills and who supports in each one, is in [white-label software reselling](https://tolky.to/en/blog/white-label-software-reselling).

## What does this change for your agency?

**A white-label AI agent turns a one-off setup job into a recurring product under the agency's brand, one that gets harder to replace with every month of use.** In practice, the change shows up in five places.

- **One product for the whole client base.** The same agent, adapted by industry, becomes a product line. The clinic, the real estate agency and the school get the same platform with different knowledge bases.
- **Pricing in three parts.** Implementation (project), platform subscription (a license under the agency's brand) and operation (conversation reviews, knowledge base updates, reports). How to build the price is in [how much to charge for a WhatsApp AI agent](https://tolky.to/en/blog/how-much-to-charge-for-whatsapp-ai-agent).
- **Retention through use.** The agent works every day and keeps every contact's history. The more the client uses it, the more switching costs them, and they see the agency as the owner of that operation.
- **Reports that talk about sales.** Beyond reach and clicks, the agency shows conversations handled, qualified leads and handoffs to the sales team.
- **A different team profile.** Writing the knowledge base, reviewing conversations and answering first-line support is operations work, not media work.

## How do you evaluate a white-label AI agent platform?

**To evaluate a white-label AI agent platform, check item by item whether the brand, domain, emails, multi-client access, official API, data protection and support work the way the agency will sell them.** Ask to see each item on a real screen.

| Criterion | What to ask | Red flag |
| :--- | :--- | :--- |
| Brand | Can I change the logo, colors, favicon, product name and login screen? | Vendor brand in the footer or the browser tab title |
| Domain | Do login, dashboard, public links and webchat open on my domain with SSL? | Only the dashboard changes; webchat stays on the vendor's domain |
| System emails | Do invitations and password resets go out with my brand? | Email with the vendor's sender or logo |
| Multi-client | Can I access every client account with one login? | One password per client, no consolidated view |
| Official WhatsApp API | Does the number connect through Meta's official API? | Connection via WhatsApp Web QR code |
| AI with oversight | Can I simulate, version and investigate answers? | Agent published directly, with no testing or history |
| Privacy | Is there a data processing agreement and access profiles? | Contract with no processor role and no activity log |
| Support | Who handles tier 2, and with which history? | Vendor support talking directly to the end client |

To evaluate the agent itself (knowledge base, human handoff, integrations), use the [checklist for choosing a WhatsApp AI agent](https://tolky.to/en/blog/how-to-choose-ai-agent-for-whatsapp-checklist).

## Mistakes when choosing a white-label AI agent

**The costliest mistake is choosing by the branding screen and finding out later that the channel, the AI or the support can't handle the client base.**

- **Building everything by hand.** An agent built in an automation tool ships fast for the first client; by the tenth, the agency has become server support. Compare in [n8n or a platform for agencies](https://tolky.to/en/blog/n8n-or-platform-for-agencies).
- **Promising full automation.** A well-implemented AI resolves part of the conversations and passes the rest to the client's team with context, through a well-configured [handoff](https://tolky.to/en/glossary/handoff). Promising 100% is the fastest way to lose the account.
- **Ignoring the CRM.** An agent that answers but doesn't record the lead or the pipeline stage leaves the agency with nothing to show in the report. See what matters in [white-label CRM with WhatsApp](https://tolky.to/en/blog/white-label-crm-with-whatsapp).
- **Forgetting who pays Meta.** Per-message fees on the official API usually sit outside the platform subscription. Agree in the proposal on who pays.
- **Skipping the partnership contract.** Partner pricing, deal registration and what happens if the vendor sells directly to your client must be in writing.

## Where Tolky fits in a white-label AI agent

**Tolky is a conversational AI CRM with AI agents, CRM and campaigns on the official WhatsApp API, and white label builds on the "Your brand and domain" feature of the Enterprise plan.** With it, the platform opens on the agency's domain with a security certificate, and the end client sees only the agency's brand: logo, colors, favicon, product name, login screen and system emails. Login, dashboard, public links and webchat share the same domain, and the agency enters each client account with the same login.

The agent works on WhatsApp, webchat, email and Microsoft Teams; voice is under activation. Each agent is configured by mission, with an identity and a behavior profile, and the agency simulates conversations before publishing. See [Tolky's WhatsApp chatbot](https://tolky.to/en/features/whatsapp-chatbot).

![Real screenshot of Tolky's Identity screen, in Portuguese, showing an AI agent's avatar, the company name, the opening message, the agent's identity and role text, the Sales and Business Relationship mission and the behavior profile chart](https://tolky.to/screenshots/identidade.png)

The platform handles more than 5 million AI messages a month and resolves more than 67% of interactions with AI; these are platform figures, not results from a specific client. Access controls and data protection are on the [security](https://tolky.to/en/security) page, and plans on [pricing](https://tolky.to/en/pricing).

**What Tolky doesn't do:** it doesn't support Instagram or SMS, which are coming soon, doesn't use unofficial APIs and doesn't pay Meta's fees, which sit outside the plan.

## Frequently asked questions

### What is a white-label WhatsApp chatbot?

It is a chatbot or AI agent that the agency delivers to clients under its own brand, on top of another vendor's platform. The end client sees the agency's domain, logo and product name. The WhatsApp connection should go through Meta's official API.

### Is a white-label AI agent allowed under WhatsApp's rules?

Yes, when the agent serves a company's customers. Since January 15, 2026, the WhatsApp Business Platform terms bar general-purpose AI assistants, where the AI is the product itself. A customer service agent under the agency's brand, working for a clinic or a store, stays on the allowed side.

### Can the end client tell the platform belongs to another vendor?

On a complete white-label platform, no. Domain, login screen, logo, colors, favicon, product name and system emails all carry the agency's brand. Check these items in a demo before signing.

### How much can an agency charge for a white-label AI agent?

It depends on the industry, the volume and the operation included. The usual approach separates implementation, subscription and operation, with Meta's fees on top. The guide to [WhatsApp AI agent pricing](https://tolky.to/en/blog/whatsapp-ai-agent-pricing) shows the cost from the other side of the table.

### Does Tolky offer white label?

Yes. White label builds on the "Your brand and domain" feature of the Enterprise plan and is the core of the White-label track in the Partner Program. The agency runs the platform under its own brand and domain, sets the final price and bills its clients.

### Who is responsible for client data in a white-label agent?

The end client owns the data, and the agency acts as a processor under Brazil's LGPD or equivalent laws, with a data processing agreement. Define these roles in writing before the first client.

---

## Conclusion

A white-label AI agent leaves the technology with the vendor and the brand, the price and the relationship with the agency. Before choosing, check the official WhatsApp API, the full domain, system emails, access to every client with one login and AI oversight, and only then look at the branding screen.

## Your agency can deliver this with Tolky

**The Tolky Partner Program is for agencies that want to bring AI agents, CRM and WhatsApp campaigns to their own clients, with recurring earnings.** On the White-label track, your agency runs the platform under its own brand and domain, sets the final price and bills its clients; the other tracks are referring Tolky or selling and implementing under the Tolky brand. Training and certification are free for program members, and the Tolky team supports every initial project in 1:1 sessions.

[Explore the Partner Program →](https://tolky.to/en/partners) Let's talk?

---

This is the Markdown version of the page. The full version, with images and navigation, is at https://tolky.to/en/blog/white-label-ai-agent-for-agencies

Tolky: https://tolky.to
