If you run a digital agency, you already know the pitch: clients want AI chatbots on their websites. Every competitor is offering them. And the obvious move is to resell one of the big SaaS platforms — white-label the interface, add a markup, done. But once you run the actual numbers, that model falls apart fast. At $300–$5,000 per client per month in platform fees, your margins evaporate before you've invoiced anyone. There is a better way, and it starts with owning the infrastructure. AI Chat Agent is one of a new generation of self-hosted chatbot tools that lets agencies build a profitable white-label chatbot business on a one-time license — no per-client fees, no revenue share, no monthly surprises.
This guide covers everything: what white-label AI chatbots actually are, how the economics compare, what the top platforms offer, and a step-by-step look at deploying and branding your own solution. Whether you have 3 clients or 50, the math here is the same — and it strongly favors owning your stack.
What Is a White-Label AI Chatbot?
A white-label AI chatbot is a chatbot product you rebrand and resell as your own. Your agency's logo. Your client's brand colors. No mention of the underlying vendor. From the end user's perspective, it's your product — built by you, powered by whatever you choose to call it.
The market for these tools is growing fast. According to Grand View Research, the global chatbot market is projected to reach $3.62 billion by 2030, with enterprise and agency adoption driving much of that growth. Businesses want AI-powered customer support, lead capture, and sales automation — but most don't want to build it themselves. That's the gap agencies fill.
White-label chatbots come in two flavors:
- SaaS white-label platforms — You pay a monthly fee, get a branded dashboard, and resell seats or usage. The vendor hosts everything. You never own the infrastructure.
- Self-hosted white-label solutions — You deploy the software on your own server. You control the data, the branding, the pricing, and the infrastructure. You own the product.
Most agency owners default to SaaS because it feels easier. But "easier" has a price — and over 12 months, that price tends to exceed what your clients are even paying you.
SaaS vs Self-Hosted: The Economics
Let's put real numbers on this. Suppose you sign 10 clients and want to offer each of them a branded AI chatbot. Here's what that looks like across common platforms:
| Platform | Model | Cost per Client/Month | 10 Clients/Year |
|---|---|---|---|
| BotPenguin Agency | SaaS (reseller) | $49–$149 | $5,880–$17,880 |
| CustomGPT | SaaS (per bot) | $89–$449 | $10,680–$53,880 |
| Botpress Cloud | SaaS (usage-based) | $89–$495+ | $10,680–$59,400+ |
| Voiceflow | SaaS (per editor seat) | $50–$125/seat | $6,000–$15,000 |
| AI Chat Agent (self-hosted) | One-time license | ~$0.60 (amortized) | <$80 total |
The self-hosted number assumes a €79 one-time license plus roughly $6–$10/month in VPS hosting (a Hetzner CAX11 handles 5–10 bots comfortably). Split across 10 clients over 12 months, your infrastructure cost is under $1.50 per client per month — before you've charged them a cent.
Beyond cost, there's a compliance dimension. SaaS platforms mean your client data flows through a third-party vendor's infrastructure. For clients in healthcare, finance, or any EU jurisdiction, that creates sub-processor liability, DPA requirements, and audit headaches. Self-hosting eliminates that problem entirely — the data never leaves your server.
We break this down in detail in our self-hosted vs SaaS chatbot cost comparison.
Top White-Label Chatbot Options Compared
Here's an honest look at the major players and where they fit:
BotPenguin
BotPenguin offers a white-label reseller program with branded dashboards and client sub-accounts. It's genuinely easy to set up. The catch: you're still paying per bot per month, and the branding controls are limited compared to what a self-hosted tool gives you. Good for agencies that want to start fast and don't mind thinner margins.
CustomGPT
CustomGPT is strong on RAG (retrieval-augmented generation) — you upload documents, it builds a knowledge base, done. The white-label tier is expensive and caps the number of bots. If you have more than 5 clients, the economics break down quickly. No self-hosting option.
Botpress
Botpress is a capable open-source platform with cloud hosting. The free tier is restrictive; the paid cloud plans get expensive at scale. The self-hosted open-source version exists but requires significant DevOps effort. If you're comparing, see our Botpress comparison for the full breakdown.
Voiceflow
Voiceflow is excellent for complex conversational design and multi-channel flows. It's a builder's tool — great if you have a developer on your team who wants to craft sophisticated bot logic. Less suited to a reseller model where you need fast per-client deployment. We also have a Voiceflow comparison that covers the tradeoffs.
AI Chat Agent (Self-Hosted)
AI Chat Agent is a self-hosted chatbot widget with a one-time €79 license. You deploy it on your own server via Docker Compose, and it runs unlimited bots — one per client, each with its own knowledge base, AI provider, branding, and chat history. There are no per-seat fees, no usage caps, and no vendor lock-in. It's the tool this guide focuses on for good reason: the unit economics are fundamentally different from every SaaS option above. For a deeper comparison of all self-hosted options, see our best self-hosted chatbot solutions roundup.
Profit Math: What Agencies Actually Earn
Let's get specific about what a white-label chatbot service actually earns you. Agencies typically price chatbot services in one of three tiers:
- Starter tier — $297/month: One bot, standard knowledge base, basic branding. Suitable for small businesses.
- Growth tier — $497/month: Multiple bots, custom branding, lead capture, monthly reporting.
- Agency/Enterprise tier — $997/month: Full white-label, custom AI configuration, dedicated support, multi-bot setup.
With a self-hosted tool and a $6/month server, your cost basis is near zero after the one-time license. Here's what that means at scale:
| Clients | Avg. Monthly Revenue | Monthly Infrastructure | Gross Margin |
|---|---|---|---|
| 1 | $297 | ~$10 | 96% |
| 5 | $1,485 | ~$15 | 99% |
| 10 | $2,970 | ~$20 | 99% |
| 25 | $7,425 | ~$40 | 99% |
Your break-even point is one client. The €79 license pays for itself the first time you invoice. Everything after that is margin. Compare that to a SaaS reseller model where you might be paying $150/client/month and charging $297 — a 49% gross margin that shrinks further with churn, support time, and platform price increases.
Owning your infrastructure also means owning your pricing. You can run promotions, offer annual discounts, or bundle chatbots into broader retainers without worrying about what the upstream vendor is charging.
Deploy Your White-Label Chatbot in 5 Minutes
AI Chat Agent ships as a Docker Compose stack. You need a VPS (Hetzner, DigitalOcean, Vultr — anything with 2 vCPUs and 4GB RAM works for 5–10 bots), Docker installed, and about five minutes.
The full stack includes: Node.js backend, React admin panel, PostgreSQL 16 with pgvector for embeddings, Redis for session management, and Nginx as a reverse proxy. Everything runs in containers — no manual dependency management, no PHP nightmares.
# 1. Extract your licensed AI Chat Agent package
unzip ai-chat-agent-v1.2.0.zip
cd ai-chat-agent
# 2. Copy and configure environment variables
cp .env.example .env
# Set your domain, SMTP, and AI provider keys in .env
# 3. Launch the stack
docker compose up -d
# 4. Visit your admin panel
# https://yourdomain.com/adminAfter that, you create a bot in the admin panel, configure its AI provider and knowledge base, grab the widget embed code, and paste it into the client's site. The whole process — from fresh VPS to live chatbot on a client's website — takes under 30 minutes the first time, and under 5 minutes for every client after that.
For a deeper look at production hardening, SSL setup, and server sizing, read our Docker deployment guide.
Customize Everything: Branding and Multi-Tenancy
White-labeling isn't just removing a logo. Clients expect the chatbot to feel native to their brand — their colors, their voice, their name on the widget. AI Chat Agent's widget settings cover all of this:
- showBranding toggle — Flip this off and the "Powered by" footer disappears entirely. Your client's customers never see the underlying vendor.
- Custom bot name and avatar — Name the bot whatever fits the client's brand. Upload a custom avatar image. It looks like a bespoke product, not a resold widget.
- Color theming — Set primary color, button color, and background color via HEX values. Match the client's brand guidelines exactly.
- Dark mode toggle — One setting switches the widget to a dark theme, useful for tech or gaming clients with dark-themed sites.
On the multi-tenancy side: each bot in AI Chat Agent is an independent entity. It has its own knowledge base, its own AI provider configuration, its own system prompt, its own chat history, and its own widget settings. You can run 50 bots on a single server — each completely isolated from the others — with no cross-contamination of data or configuration.
This matters operationally. When client A wants to switch from OpenAI to Claude, you change one bot's AI provider setting. It doesn't affect client B or client C. When client D uploads a new product catalog PDF, that data only enters client D's knowledge base. Per-client isolation is built into the architecture, not bolted on.
Multi-LLM Support: No Vendor Lock-In
One of the most common objections to self-hosted chatbots is AI model dependency. What if OpenAI raises prices? What if a client wants Claude? What if a regulated client needs a private endpoint?
AI Chat Agent supports all major providers out of the box:
- OpenAI — GPT-4o, GPT-4 Turbo, GPT-4o mini
- Anthropic Claude — Claude Opus 4, Claude Sonnet 4, Claude Haiku
- Google Gemini — Gemini 2.5 Pro, Gemini 2.5 Flash
- Any OpenAI-compatible endpoint — Ollama, LM Studio, Together AI, Groq, Azure OpenAI, and more
Per-bot model selection means each client gets exactly the model that fits their use case and budget. A budget-conscious client uses GPT-4o mini. A legal firm that needs strong reasoning uses Claude Opus 4. A client who wants to run everything on-premises connects a local Ollama endpoint. You configure it in the admin panel — no code changes required.
API keys are stored with AES-256 encryption. If a client provides their own API key, you store it securely and it never appears in plaintext in your database. For more on the vendor lock-in problem and how multi-LLM architecture solves it, see our multi-LLM chatbot guide.
GDPR and Data Ownership
For agencies working with EU clients — or any client handling personal data — data residency isn't optional. It's a legal requirement. And it's where SaaS chatbots create real liability.
When you use a SaaS chatbot platform, that vendor becomes a data sub-processor. You need a Data Processing Agreement (DPA) with them. You need to disclose them in your client's privacy policy. If the vendor's servers are outside the EU, you need to verify Standard Contractual Clauses are in place. If they get breached, you're implicated.
Self-hosting eliminates most of this. Your server, your data. You choose the data center jurisdiction. You control retention policies. You decide what gets logged and for how long. AI Chat Agent includes configurable data retention — you can set how long chat histories are stored and purge them on a schedule.
For clients in healthcare (HIPAA), finance, or government sectors, self-hosting isn't just a cost advantage — it's often the only option that passes a compliance review. Our GDPR-compliant AI chatbot guide covers the full compliance checklist.
Scale: From 1 Client to 50
The architecture scales linearly. One server handles 5–10 bots without breaking a sweat on a mid-range VPS. As you add clients, you have a few scaling paths:
Vertical scaling (simplest)
Upgrade your VPS. Going from 2 vCPUs/4GB to 4 vCPUs/8GB roughly doubles your capacity. On Hetzner, that's the difference between ~$6/month and ~$15/month. At 15 clients paying $297/month each, spending an extra $9/month on compute is a rounding error.
Horizontal scaling (for larger operations)
Separate bots across multiple servers — by client segment, geography, or load. Run a EU-hosted server for GDPR-sensitive clients and a US-hosted server for North American clients. Because each bot's configuration is independent, moving a bot to a different server is a configuration export/import operation.
Resource planning guidelines
- 1–5 bots: Hetzner CAX11 ($6/month, 2 vCPU, 4GB RAM)
- 5–15 bots: Hetzner CAX21 ($12/month, 4 vCPU, 8GB RAM)
- 15–50 bots: Hetzner CAX31 ($24/month, 8 vCPU, 16GB RAM) or two smaller servers
The pgvector database is the most memory-intensive component — it holds all RAG embeddings for every client's knowledge base. If clients upload large document sets, plan for proportionally more RAM. Redis and Nginx are lightweight. The Node.js backend scales with request volume, not bot count.
Even at 50 clients, your total infrastructure spend is unlikely to exceed $50–$80/month. Against $14,000+/month in client revenue at the starter tier, that's a 99%+ gross margin business.
Frequently Asked Questions
What is a white label AI chatbot?
A white label AI chatbot is a chatbot product built by one company and rebranded by another — typically a digital agency — to sell as their own. Your clients see your brand, your colors, and your domain. The underlying vendor is invisible. White label chatbot software lets agencies offer AI-powered customer support, lead capture, and sales automation without building the technology from scratch.
How much does white label chatbot software cost?
SaaS white label chatbot platforms charge $49 to $495+ per client per month, which adds up to $6,000–$60,000 annually for 10 clients. Self-hosted alternatives like AI Chat Agent cost a one-time fee (e.g., €79) plus $6–$20/month for VPS hosting regardless of client count — reducing per-client cost to under $2/month.
Can I use a white label AI chatbot with my own domain and branding?
Yes. With self-hosted white label chatbot solutions, you deploy on your own server under your own domain. You control the bot name, avatar, color theme, and can disable all vendor branding. Each client gets a fully customized widget that looks like a bespoke product built by your agency.
Do I need coding skills to build a white label chatbot?
No. Tools like AI Chat Agent require only basic command-line comfort — running docker compose up -d and editing a .env file. After deployment, all bot configuration, client onboarding, knowledge base management, and branding happen through a visual admin panel with no coding required.
Is a self-hosted white label chatbot GDPR compliant?
Self-hosting gives you full control over data residency, which is critical for GDPR compliance. You choose the data center jurisdiction, control retention policies, and eliminate third-party sub-processor liability. For agencies serving EU clients or those in healthcare and finance, self-hosted white label AI chatbots are often the only option that passes compliance review.
How many clients can a white label chatbot server handle?
A single VPS can handle 5–50+ client bots depending on resources. A $6/month server (2 vCPU, 4GB RAM) supports 5–10 bots comfortably. Scaling to 50 clients typically requires only $24–$80/month in infrastructure — maintaining 99%+ gross margins at standard agency pricing of $297–$997/month per client.
What AI models does a white label chatbot support?
Modern white label chatbot platforms support multiple LLM providers. AI Chat Agent works with OpenAI (GPT-4o, GPT-4 Turbo), Anthropic Claude (Opus 4, Sonnet 4), Google Gemini (2.5 Pro, 2.5 Flash), and any OpenAI-compatible endpoint including Ollama for on-premises deployment. Each client bot can use a different model.
The white-label chatbot opportunity is real, and the demand from clients is only growing. The question is whether you build that business on someone else's infrastructure — paying them a cut of every dollar you earn — or whether you own the stack and keep the margin. For most agencies, the math isn't close.
AI Chat Agent is available for a one-time €79 license. It includes everything covered in this guide: multi-bot support, RAG knowledge base, white-label widget, multi-LLM configuration, lead capture, and operator live reply. No monthly fees. No per-client charges. Deploy it, brand it, and start billing clients this week. Try the live demo to see the admin panel in action, or get the license and have your first client bot live by the end of the day. More posts on the blog cover specific use cases — from ecommerce to compliance — if you want to go deeper.