How to Set Up a WhatsApp AI Chatbot in 5 Minutes

WhatsApp is where your customers already are. With over 3 billion active users worldwide and open rates exceeding 90%, WhatsApp is the most powerful channel an SME can use to reach prospects and support existing customers — yet most businesses still reply manually, one message at a time.

In 2026, the gap between businesses using AI agents on WhatsApp and those still handling chats manually is not just a productivity gap — it is a revenue gap. AI-powered WhatsApp chatbots can qualify leads at 2 AM, answer product questions instantly, and hand off hot prospects to your sales team with full conversation context — all without a single extra hire.

The good news: setting one up no longer requires a developer, a Meta Business verification odyssey, or a six-figure integration budget. Tools like Botline abstract all of that complexity so you can go from zero to a live AI agent in the time it takes to drink a coffee.

Here is exactly how to do it.

Before you touch Botline, gather the following. Each item takes less than two minutes to confirm:

• A WhatsApp Business number — this can be a new SIM, a VoIP number (e.g. Twilio), or an existing WhatsApp Business number you are migrating. It must not already be registered on the WhatsApp Business App on a phone, or you will need to deregister it first.
• A Facebook / Meta Business account — a free personal Facebook account works to start. You will connect it during setup so Meta can verify ownership of the phone number.
• A Botline account — the Free plan is enough to complete this guide. Sign up at botline.cc/signup.
• Your business knowledge base — a Google Doc, PDF, or even a list of bullet points describing your products, pricing, FAQs, and policies. The better this is, the better your AI agent will perform.

That is genuinely it. You do not need a developer, a server, or a Meta Business verification (Botline handles API access for you under its own BSP agreement).

1. Create your Botline account at botline.cc/signup. Choose your plan — Free works for testing, Starter ($29/month) is recommended for a live business.
2. Go to Settings → Channels → WhatsApp and click Connect WhatsApp Number.
3. Click "Continue with Facebook". You will be redirected to Meta's embedded sign-up flow. Log in with your Facebook account and follow the prompts to create or select a Meta Business portfolio.
4. Add your phone number. Enter the number you want to use for WhatsApp. Meta will send a 6-digit verification code via SMS or a voice call — enter it to confirm ownership.
5. Wait for provisioning. Botline connects your number to the WhatsApp Cloud API. This typically takes 30–90 seconds. You will see a green Connected badge when it is ready.

Important: if your number was previously registered on the WhatsApp Business App (the smartphone app), you must open that app and go to Settings → Account → Delete Account to deregister it before the Cloud API can claim it. Your chat history is not affected on the recipient side.

Your AI agent is only as good as the information you give it. Botline's knowledge base uses RAG (Retrieval-Augmented Generation) — meaning the AI retrieves the most relevant chunks of your documents before generating a reply. This keeps answers accurate and grounded in your actual business data.

1. Go to Knowledge Base → Add Source. You can upload PDFs, paste website URLs for automatic crawling, or type content directly.
2. Upload your core documents: product catalogue, pricing sheet, FAQ document, returns/shipping policy, and any sales scripts you use manually today.
3. Add your website URL (optional but recommended). Botline will crawl your site and extract content automatically — useful for "about us" questions and feature descriptions.
4. Review the parsed content. After each upload, Botline shows you a preview of what was extracted. Edit any sections that look garbled or incomplete.
5. Click Sync. The knowledge base re-indexes in the background (typically under 60 seconds for documents under 20MB).

Pro tip: add a section called "Things my agent should never say" — for example, competitor names, pricing you do not want public, or anything legally sensitive. The AI will treat these as exclusion rules.

Botline lets you create one or more AI agents, each with its own name, tone, and role. For most SMEs, one agent covering sales and support is the right starting point.

1. Go to Agents → Create Agent. Give your agent a name customers will see — e.g. "Aria from FreshBakes" or simply "Support Bot".
2. Write a system prompt. This is the instruction set that shapes how your agent behaves. A good starting prompt looks like this:

"You are Aria, a friendly sales assistant for FreshBakes. Your job is to help customers choose products, answer questions about ingredients and delivery, and collect their order details. Always be warm and concise. If you cannot answer a question confidently, say so and offer to connect them with a human."

3. Choose your AI model. Botline supports multiple models: GPT-4o, Claude 3.5 Sonnet, Gemini 1.5 Pro, DeepSeek V3, and local Ollama models via BYOK (Bring Your Own Key). GPT-4o is the default and works well for most businesses. If you have an OpenAI API key, plug it in under Settings → BYOK to use your own quota and reduce costs.
4. Set escalation rules. Under the Smart Escalation tab, define when the agent should hand off to a human: e.g. "if the customer mentions a complaint", "if the order value exceeds $500", or "if the customer asks three times without resolution".
5. Assign the knowledge base you built in Step 2 to this agent.

Never send an untested AI agent to real customers. Botline has a built-in test console that lets you simulate conversations before going live.

1. Open the Test Console (Agents → [Your Agent] → Test). Type messages as if you were a customer and review the responses.
2. Ask edge-case questions: things outside your knowledge base, rude messages, requests for competitors, requests for a human. Verify the agent handles each gracefully.
3. Check escalation: trigger one of your escalation rules and confirm the conversation flags correctly in the inbox.
4. Send a real WhatsApp message to your connected number from a personal phone. Confirm the agent responds within 2–3 seconds.
5. Flip the agent to Live (toggle on the Agents page). From this moment, all inbound WhatsApp messages to your number are handled by the AI first.

Congratulations — your WhatsApp AI agent is live. You can monitor all conversations in real-time from the Botline inbox, intervene manually at any point, and review analytics under the Reports tab.

• Update your knowledge base regularly. Add new products, seasonal promotions, and policy changes as they happen. An outdated knowledge base is the number-one cause of AI hallucinations.
• Review the first 50 conversations manually. You will spot patterns — questions the agent struggles with, phrasing it uses that does not match your brand voice — and can refine the system prompt accordingly.
• Use lead scoring. Botline scores leads based on intent signals in the conversation. Route high-scoring leads directly to your top salesperson via the escalation rules.
• Send proactive messages with WhatsApp Stories. The Starter plan and above lets you broadcast promotional content via WhatsApp Status — a far less intrusive channel than bulk message blasts.
• Connect to your CRM via n8n. Botline's n8n integration lets you push qualified leads directly into HubSpot, Salesforce, or a Google Sheet — zero manual data entry.
• Use multiple agents for different topics. A Botline Pro plan lets you run separate agents for sales, support, and billing — each with its own knowledge base and escalation path, routed by keyword or customer tag.

• Using a number still active on the WhatsApp Business App. This breaks the Cloud API connection. Always deregister first.
• Uploading a thin knowledge base. A five-bullet FAQ is not enough. The richer your content, the fewer escalations you will see. Aim for at least 2,000 words of business content at launch.
• Writing a vague system prompt. "You are a helpful assistant" tells the AI nothing useful. Be specific: who you are, what you sell, your tone, what you will and will not do.
• Skipping the test phase. One embarrassing AI response sent to a real customer can undo trust quickly. Test thoroughly before flipping to Live.
• Never reviewing conversations. AI agents improve over time only if you review gaps and update the knowledge base. Block 15 minutes per week for this.
• Ignoring Meta's messaging policy. WhatsApp has rules about opt-in consent for proactive messages. Ensure customers have consented before sending marketing messages — Botline's opt-in tools help manage this automatically.

With those pitfalls avoided, your WhatsApp AI agent will be a genuine asset — handling the volume your team cannot, and freeing your people for the conversations that actually need a human touch.