How to Set Up a WhatsApp AI Agent for Your UK Business in 30 Minutes
TL;DR: A WhatsApp AI agent answers your business messages 24/7, books appointments, handles FAQs, and qualifies leads — all without human involvement. Using Evolution API + n8n, a functional agent can be live within 30 minutes for testing (full production setup takes 2-5 days). No coding required if you use a visual workflow builder.
---
Why UK Businesses Are Adopting WhatsApp AI Agents
WhatsApp has 2+ billion users globally and over 30 million active UK users. For consumer-facing UK businesses — solicitors, dental practices, estate agents, beauty salons, restaurants — WhatsApp has become the preferred contact channel for a significant portion of customers.
The problem: most SMBs respond to WhatsApp messages like they responded to emails in 2005 — manually, during office hours, with inconsistent quality. A WhatsApp AI agent solves this.
What changes with an AI agent:
- Messages answered in under 3 seconds, 24/7
- Consistent, accurate responses every time
- Appointment booked without involving staff
- Lead details captured and synced to CRM automatically
- Staff freed from repetitive triage work
The Technology Stack
This guide uses three components:
1. Evolution API — The WhatsApp layer. Self-hosted, open-source software that connects your WhatsApp number to your automation system. Think of it as a bridge between WhatsApp and your workflow engine.
2. n8n — The workflow engine. A visual automation tool where you design what happens with each incoming message: classify it, send it to AI, retrieve calendar data, send a reply, log to CRM.
3. OpenAI GPT-4o (or Claude 3.5 Sonnet) — The AI brain. Processes the customer's message and generates a natural, context-aware response based on your business's knowledge base.
Step-by-Step Setup
Step 1: Deploy Evolution API (15 minutes)
You need a VPS server (DigitalOcean, Hetzner, or similar — €4-15/month for a 2GB RAM instance).
On your server, run:
# Install Docker if not present curl -fsSL https://get.docker.com | sh
Pull and run Evolution API
docker run -d \
--name evolution_api \
-p 8080:8080 \
-e AUTHENTICATION_API_KEY=your_secret_key \
atendai/evolution-api:latest
Access the Evolution API dashboard at http://your-server-ip:8080. Create a new instance, scan the QR code with your WhatsApp Business number.
Step 2: Set Up n8n (5 minutes)
Fastest option: n8n Cloud free trial — no server setup.
Alternatively, self-hosted on the same server:
docker run -d \ --name n8n \ -p 5678:5678 \ -e N8N_BASIC_AUTH_ACTIVE=true \ -e N8N_BASIC_AUTH_USER=admin \ -e N8N_BASIC_AUTH_PASSWORD=your_password \ n8nio/n8n
Step 3: Build Your First Workflow
In n8n, create a new workflow with these nodes:
1. Webhook node — Receives events from Evolution API
- URL will look like:
https://your-n8n-url/webhook/whatsapp
- Copy this URL into Evolution API's webhook settings
2. IF node — Filter: only process text messages (ignore status updates, receipts)
- Condition:
{{$json.event}} equals messages.upsert
3. OpenAI Message node — Generate AI response
- System prompt: "[Your business context and instructions]"
- User message:
{{$json.body.messages[0].message.conversation}}
4. HTTP Request node — Send reply via Evolution API
- Method: POST
- URL:
http://your-server:8080/message/sendText/your-instance-name
- Headers:
apikey: your_secret_key
- Body:
{ "number": "{{$json.body.messages[0].key.remoteJid}}", "text": "{{$ai_response}}" }
Step 4: Write Your System Prompt
This is the most important step. Your system prompt determines everything about how your agent behaves.
Template for a UK SMB:
You are a professional AI assistant for [Business Name], a [type] business based in [location].
Your role: Answer customer WhatsApp messages professionally, help them book appointments, and answer questions about our services.
Business information:
- Opening hours: Monday-Friday 9am-6pm, Saturday 10am-4pm
- Services: [list your services]
- Pricing: [key prices or "contact us for a quote"]
- Address: [address]
- Phone: [phone number]
Rules:
- Always be polite and professional
- If you cannot help, offer to connect them with a human agent
- For appointments, ask for: name, preferred date/time, service needed
- Do not make up information you don't have
- Always identify yourself as an AI assistant when directly asked
When a customer wants to book: collect their details and confirm you've noted the request. Tell them someone will confirm within [timeframe] or trigger the calendar booking tool.
Step 5: Test Before Going Live
Test with your own phone number first:
- Send "Hello" — does it respond naturally?
- Ask about your opening hours — does it give correct info?
- Request an appointment — does it collect the right details?
- Ask something off-topic — does it handle gracefully?
- Ask "Are you a bot?" — does it answer honestly?
Iterate on your system prompt until responses feel right. Plan for 2-3 hours of testing and refinement.
Adding Appointment Booking
Connect n8n to your calendar system:
For Google Calendar:
- Add Google Calendar credentials in n8n
- Use "Google Calendar: Check Availability" node
- Use "Google Calendar: Create Event" node
Workflow logic:
- Customer: "I'd like an appointment on Friday afternoon"
- n8n: Query Google Calendar for Friday 14:00-18:00 slots
- n8n: If slots available → offer them
- Customer: "3pm works"
- n8n: Create event, send confirmation
GDPR Compliance for UK Businesses
Before going live, ensure you have:
1. Transparency notice: Add to your first message: > "Hi! I'm an AI assistant for [Business]. I can help with [X, Y, Z]. For complex queries, I'll connect you with our team. Your messages are processed to provide this service. [Privacy Policy link]"
2. Data minimisation: Only collect what you need. Don't store full conversation history indefinitely — set a retention period (30-90 days is typical).
3. Data processor agreement: If your n8n or Evolution API is on a third-party server, ensure your hosting provider has a GDPR-compliant DPA.
4. Right to opt out: Include in your welcome message or FAQ: "To stop receiving messages from our AI assistant, reply STOP."
What Results to Expect
Based on typical UK SMB deployments:
| Metric | Before AI Agent | After AI Agent | |--------|----------------|----------------| | Response time | 4-8 hours average | Under 3 seconds | | Messages handled without staff | 0% | 65-80% | | After-hours enquiries converted | 10-15% | 50-60% | | Staff time on WhatsApp/week | 8-12 hours | 2-3 hours | | Appointment no-shows (with reminders) | 20-25% | 8-12% |
Getting Professional Help
Setting this up yourself takes 1-3 days if you're comfortable with servers and APIs. For most business owners, it's worth getting a specialist to handle the setup — so you can focus on your business.
MGL Digital Media sets up WhatsApp AI agents for UK businesses from £119/month (all-inclusive: setup, hosting, maintenance, updates).
Our stack: Evolution API + n8n + GPT-4o, hosted in EU for GDPR compliance.
Book a free 15-minute strategy call → /whatsapp-ai-asistan
---
Author: Mustafa Gul · MGL Digital Media, Enfield, London · info@mgldigitalmedia.com Company No: 16007414 · Updated May 2026