WhatsApp API with n8n: 2026 Guide & Ban Risks

WhatsApp has become the primary communication channel for businesses across Europe. But most agencies and companies that automate WhatsApp do it through intermediaries (BSPs) like WATI, ManyChat, or Twilio — and that comes with a significant hidden cost.

When you use a BSP, you pay three times:

1. Meta, per conversation (unavoidable).
2. The BSP, a monthly subscription (€50 - €200).
3. The BSP, a markup on every message sent.

In this guide, you’ll learn how to eliminate points 2 and 3 by setting up your own WhatsApp Gateway with Evolution API v2, Meta’s Official Cloud API, and n8n. You still pay Meta per conversation, but you save the BSP subscription and their per-message markup.

How Much BSPs Charge vs. Meta Directly

To size up the savings, it’s worth comparing what Meta charges directly (official 2026 pricing for Europe) versus what you end up paying through a BSP:

Conversation Type	Meta Cost (Europe Avg)	Typical BSP Cost	Your Savings
Marketing	~0.05 €	~0.07 €	37%
Utility (Orders)	~0.02 €	~0.03 €	40%
Authentication (OTP)	~0.03 €	~0.05 €	40%
Service (User-initiated)	FREE	~0.01 €	100%

What is Evolution API v2?

Evolution API is an Open Source project that acts as a WhatsApp Gateway. It runs on your own server (Docker) and exposes WhatsApp as a REST API you can control via HTTP requests.

It has two operating modes:

Mode 1: Official API (Meta Cloud API) ✅ Recommended

This is the professional route. It requires registering a business on Meta (Business Manager) and verifying your identity.

• Cost: You pay per 24-hour conversation (from €0.02 to €0.08 depending on message type and country). Customer replies are cheaper than business-initiated messages (marketing).
• Advantage: Official, 100% stable, scalable, and zero ban risk if you follow the policies. Supports rich templates (interactive buttons, catalogues).
• Ideal for: Production environments, marketing campaigns, automated customer support, and agencies that need to sleep well while offering SLAs to their clients.

Mode 2: Baileys (QR Code Connection) ⚠️ Testing Only

How does it work? Just like connecting WhatsApp Web in your browser by scanning a QR code, Evolution API creates a hidden web session and exposes it as a REST API. You scan the code and that number can be controlled via HTTP requests in n8n.

The Good:

• Completely free: No per-message cost, regardless of who initiates the conversation.
• No Meta approval needed: You can connect any regular number (even personal) in seconds, without going through Business Manager.
• Group support: Unlike the official API, you can interact with and read messages in WhatsApp groups.

The Bad:

• Permanent number ban: Meta can revoke your number with no chance of recovery.
• Requires phone online: The physical phone with the account must stay connected to the internet.
• Disconnections: If the phone loses signal, Meta updates the protocol, or the session expires, the API goes down.
• No rich templates: You cannot send interactive buttons or pre-approved templates officially.

Can You Use WhatsApp API for Free?

Technically yes, but not without consequences. Evolution API’s Baileys mode lets you send and receive WhatsApp messages without paying Meta a cent. Sounds ideal, but there’s a catch: Meta doesn’t allow it and actively pursues these connections.

For prototyping and testing flows, Baileys is a fantastic tool. It lets you validate your entire n8n automation without spending on official conversations. But for real production with clients, the only safe option is Meta’s Official API (Cloud API), where you pay per conversation at the real price (from €0.016) with no intermediaries.

	Baileys (QR)	Official API (Cloud API)
Cost per message	Free	From €0.016/conversation
Ban risk	⛔ High (permanent)	✅ None
Recommended use	Testing & prototypes	Production & clients
Requires Business Manager	No	Yes
Rich templates	No	Yes

In short: Use Baileys to develop and test. Use the Official API to bill clients and sleep well.

Technical Tutorial: Your Own WhatsApp Stack (Official API)

To build this, you need a server (VPS) running Docker, Redis, and Postgres. (Note: At AIBuildr, our servers come with Evolution API v2 pre-installed and configured with Redis & Postgres, optimised for Meta’s Official API mode).

Prerequisite: You need a Meta Business Manager account with a verified phone number for the WhatsApp Business API.

If you build it yourself, the docker-compose.yml looks like this (simplified):

services:
  evolution-api:
    image: attdevelopers/evolution-api:v2.1.0
    environment:
      - SERVER_URL=https://api.yourdomain.com
      - DATABASE_PROVIDER=postgresql
      - REDIS_ENABLED=true

Step 1: Connect n8n to WhatsApp

Once Evolution is running, go to n8n. You don’t need complex nodes, just the HTTP Request node or a Webhook.

Receive Messages (Webhook):

1. In Evolution, go to your instance and enable “Webhooks”.
2. Paste your n8n Production Webhook URL.
3. Enable MESSAGES_UPSERT events.

Now, every time someone writes to your number, n8n will receive a JSON like this:

{
  "data": {
    "key": {
      "remoteJid": "34666555444@s.whatsapp.net"
    },
    "message": {
      "conversation": "Hi, I'd like info on your pricing."
    }
  }
}

Step 2: The Brain (AI Agent)

This is where automation makes the real difference. We’ll connect that message to an AI Agent that responds automatically. And remember: if the user initiates the conversation, it’s a service conversation — Meta doesn’t charge for it.

Agent Configuration in n8n

1. Use the AI Agent node.
2. Connect it to a Vector Store (Qdrant) if you want it to know about your business (RAG).
3. In the System Prompt, give clear sales instructions and tone.

Step 3: Send the Response

To reply, use an HTTP Request node in n8n pointing to Evolution API:

• Method: POST
• URL: https://your-evolution-api.com/message/sendText/instance
• Body:

  {
    "number": "34666555444",
    "text": "Hi! Sure, our pricing starts at..."
  }

Advanced Use Cases with AI Agents

Sending and receiving messages is just the beginning. The real power of running your own gateway connected to n8n lies in what you can build behind it. At AIBuildr, we design the internal network so orchestration is instant and secure.

Case 1: Automated RAG with Qdrant

If you sell complex services, your WhatsApp agent can read all your PDF manuals or your entire website to answer questions like an expert.

• Connect the AI Agent node in n8n directly to your Qdrant cluster.
• AIBuildr advantage: Your vector cluster is already interconnected on the same private network and secured — no need to configure IPs or ports exposed to the internet, just use your auto-generated keys.

Case 2: High-Speed Persistent Memory (Redis)

A basic bot forgets what the customer said 3 messages ago. To create a fluid experience, you need short-term memory.

• Instead of overloading a SQL database, add a Redis node to your n8n flow.
• Internal Network Magic: Just use the environment variable {{ $env.REDIS_URL }} (something like redis://redis-your-slug:6379). It’s shared cache isolated to your instance, delivering reads/writes in milliseconds.

Case 3: Complex Delegation & Orchestration (Flowise)

n8n is brilliant for workflows, but sometimes you need the multi-agent reasoning that Flowise offers (one agent evaluates intent, another extracts data, another drafts the response).

• Configure n8n to handle the initial WhatsApp message routing.
• If the message needs intelligence, use an HTTP Request node in n8n pointing directly to your Flowise agent’s internal API: {{ $env.FLOWISE_URL }}/api/v1/prediction/...
• Zero Latency: The request travels through Docker’s isolated internal network. It’s an instant Zero-Trust network, and best of all: this communication doesn’t count towards your public bandwidth usage.

Testing with Baileys: Prototype Before Production

If you don’t have your Business Manager verified yet or you want to test your n8n flow quickly, you can temporarily use Baileys mode to validate the integration.

Quick steps for testing:

1. In Evolution API, create a new instance selecting Baileys mode.
2. Scan the QR code with a prepaid number you don’t use for anything else.
3. Set up the same n8n webhook and test your AI agent flow.
4. Once you’ve validated everything works, create a new instance in Cloud API mode with your verified business number and point your webhook to it.

This process lets you iterate on your automation without spending on Official API conversations during development.

Conclusion: Cut the Middleman, Not the Security

If you run an agency or SaaS, the markup BSPs charge can eat your margin. By switching from WATI/ManyChat to Evolution API + Meta’s Official API + n8n:

1. You eliminate the BSP monthly subscription (~€100/mo savings).
2. You eliminate the per-message markup (30-50% savings).
3. You still pay Meta per conversation (from €0.016), but at the real price, with no intermediaries.
4. You gain full control of your data (it’s on your VPS, not third-party servers).

At AIBuildr, we hand you the infrastructure ready: n8n + Evolution API + DBs on a private server in Europe. Just connect your Business Manager and start automating with the Official API.

Deploy your WhatsApp Automation server today →