How to Build WhatsApp AI Voice Agents with ElevenLabs (Step-by-Step Tutorial)
Imagine your business answering customer calls 24/7 with a natural-sounding AI agent that can handle inquiries, book appointments, and provide information - all through WhatsApp. This isn't future tech - it's available now with ElevenLabs' powerful voice AI. In this guide, we'll walk you through exactly how to set it up, whether you're using an existing business number or starting fresh.
Why Every Business Needs a WhatsApp Voice Agent
Customer service calls don't stop when your business closes. Missed calls mean missed opportunities - whether it's a potential client inquiring about your services or an existing customer needing support. Traditional IVR systems feel robotic and frustrating, while hiring 24/7 staff is prohibitively expensive for most businesses.
AI voice agents solve this perfectly. The salon in our example handles men's haircut inquiries automatically through WhatsApp, quoting prices based on stylist selection ($75 for basic cuts). The agent sounds completely natural, can answer in multiple languages, and never gets tired or frustrated.
85% of customers prefer messaging over phone calls for simple inquiries. WhatsApp's 2 billion+ users make it the ideal platform for AI-powered customer interactions that feel personal yet scale effortlessly.
Getting Started with ElevenLabs
ElevenLabs provides the most realistic AI voices available today. Their technology powers natural conversations that adapt to context, with emotional range that makes interactions feel genuinely human.
After logging into your ElevenLabs dashboard, navigate to the WhatsApp integration panel. If you're new to ElevenLabs, you'll need to create a business profile first - a 30-second process requiring your business name, admin email (use a real one), and basic details.
Step 1: Business Profile Creation
Click "Create Business Profile" and fill in:
- Your official business name
- Admin email address (not a throwaway)
- Basic contact information
Step 2: WhatsApp Business Account
Parallel to ElevenLabs, create your WhatsApp Business account:
- Select your business category
- Add your website and time zone
- Complete the basic profile information
Pro Tip: Even if you plan to use an existing number, create a new WhatsApp Business account first. This simplifies the porting process later.
WhatsApp Business Account Setup
WhatsApp offers two main paths for connecting your AI agent: using a display name only (quickest but limited) or connecting an actual phone number (more work but full functionality).
The display name option gives you a free 555-prefix number immediately, but with critical limitations - no voice calls, only text. This works for demos but isn't suitable for most business applications.
Key Decision Point: If your business relies on voice interactions, you'll need to verify a real number. The process takes 2 days but unlocks the full potential of AI voice agents.
Number Options: Porting vs. New
You have three options for connecting phone numbers to your WhatsApp AI agent:
1. Existing Personal Number
If you're using a personal cell number with WhatsApp currently, you'll need to delete that account first. This isn't recommended for most businesses as it mixes personal and professional communications.
2. WhatsApp Business Number
Ideal for existing businesses. Simply input your current WhatsApp Business number during setup. Meta will guide you through the migration process to connect it with ElevenLabs.
3. Landline Number
Surprisingly, landlines work well. As long as the number isn't already connected to WhatsApp, you can verify it via phone call. This is a great option for businesses wanting to keep their established landline number.
Free Number Option: Meta provides free 555 numbers that can be upgraded to support voice calls after business verification. This is the most cost-effective path for new implementations.
Business Verification Process
To enable voice calls on your WhatsApp number, you must complete Meta's business verification. This involves submitting documentation to prove your business is legitimate.
The process typically takes about 2 days. You'll need to upload documents like business registration, tax ID, and proof of address. Once verified, you'll also need to connect a payment method (credit card), though basic usage remains free.
Critical Setting: Enable Voice Calls
After verification, navigate to your WhatsApp Business Manager settings. Under phone number management, find and enable the "Allow voice calls" option. Without this, your AI agent won't be able to receive or make calls.
Verification Tip: Have all your business documents ready before starting. Incomplete submissions can delay the process by several days.
Agent Configuration & Best Settings
With your number verified and connected, it's time to configure your AI agent. In ElevenLabs, you'll see your WhatsApp number now appears in the connected accounts section.
Step 1: Select Your Agent
Choose which ElevenLabs agent to connect to this number. For our salon example, we selected the "Kasa Maria Salon" agent with a Latino voice preset.
Step 2: Configure Response Modes
Decide whether to enable:
- Outbound messages/calls: Lets your agent initiate conversations
- Messaging responses: Allows text replies to messages
Step 3: Voice Model Selection
ElevenLabs' V3 model with expressive mode enabled creates the most natural conversations. However, this presents a unique challenge for WhatsApp integration.
Voice vs. Text Tradeoff: V3's expressive mode adds realistic vocal tags (like "clears throat") to the transcript. These appear as unwanted text in WhatsApp messages. For text-heavy use, consider disabling V3 or creating a separate text-only agent.
Key Limitations & Workarounds
While powerful, the ElevenLabs-WhatsApp integration has some important limitations to understand before deployment.
1. Context Separation
Voice and text interactions maintain separate contexts. If a customer calls about an appointment, then texts about the same appointment, the agent won't connect these automatically. This can be addressed with custom development to bridge contexts.
2. V3 Text Artifacts
As mentioned, V3's expressive mode creates unwanted text tags in WhatsApp messages. The simplest solution is to disable V3 for text-heavy agents or route text interactions to a dedicated non-V3 agent.
3. Number Restrictions
Free 555 numbers don't support voice calls. Landlines work but may have limited functionality. For full features, a verified mobile business number is recommended.
Optimization Tip: Analyze your customer interactions. If 80%+ are voice, use V3. If mostly text, create a dedicated text agent with V3 disabled for cleaner conversations.
Watch the Full Tutorial
For visual learners, our video tutorial walks through every step of this process, including the actual voice agent demonstration shown at the beginning (timestamp 0:45). You'll see exactly how to navigate the ElevenLabs dashboard, configure WhatsApp settings, and test your agent.
Key Takeaways
WhatsApp AI voice agents represent a transformative technology for businesses of all sizes. By implementing the solution outlined here, you can provide 24/7 customer service that feels personal and human, without the costs of round-the-clock staff.
In summary: 1) Choose the right number strategy for your needs, 2) Complete business verification to unlock full functionality, 3) Configure separate agents for voice and text if needed, and 4) Understand the current limitations to set proper expectations. With these foundations, you're ready to deploy AI-powered customer interactions that scale with your business.
Frequently Asked Questions
Common questions about WhatsApp AI voice agents
Yes, you can port your existing WhatsApp business number to ElevenLabs. The process involves verifying your business with Meta and connecting your payment processor.
It typically takes about 2 days for verification. Once complete, you'll need to enable voice calls in your WhatsApp Business Manager settings to unlock full functionality.
- Prepare your business documentation beforehand
- Ensure your number isn't tied to a personal WhatsApp account
- The migration process preserves all your existing contacts
The main limitations include technical constraints and platform-specific behaviors you should understand before implementation.
Expressive mode in V3 models generates unwanted audio tags in text conversations. Voice and text interactions maintain separate contexts by default. Free 555 numbers from Meta don't support voice calls.
- For text-only agents, disable V3 or create a dedicated text agent
- Context bridging requires custom development
- Full features require a verified business number
The basic technical setup can be completed in about 30 minutes if you're using a new number. This includes creating your ElevenLabs agent and connecting it to WhatsApp.
Business verification with Meta typically takes 2 days. For existing numbers, the porting process can take additional time depending on your current provider and verification status.
- Basic demo setup: 30 minutes
- Full production deployment: 2-3 days
- Complex customizations may require additional time
Yes, the same agent can technically handle both voice and text interactions, but with important considerations regarding user experience.
The V3 model's expressive mode creates unwanted text artifacts in WhatsApp messages. For optimal performance, we recommend creating separate agents - one optimized for voice with V3 enabled, and another for text with V3 disabled.
- Voice-focused agent: Enable V3 for natural conversations
- Text-focused agent: Disable V3 for clean messaging
- Routing logic can determine which agent handles each interaction
Meta requires standard business documentation to verify your company and prevent misuse of the WhatsApp Business API.
Typically, you'll need to provide your business registration documents, tax identification number, and proof of business address. The exact requirements may vary slightly depending on your country and business type.
- Legal business registration documents
- Tax ID or equivalent
- Proof of business address (utility bill or bank statement)
The cost structure involves multiple components that businesses should understand before implementation.
WhatsApp offers free starter numbers (555 prefix) but these don't support voice calls. For full functionality, you'll need a verified business number which may incur costs. ElevenLabs has its own pricing tiers based on usage volume. While there are no upfront costs, you'll need to connect a valid payment method to both platforms.
- WhatsApp: Free for basic usage, paid for high volume
- ElevenLabs: Pay-as-you-go based on voice generation
- No long-term contracts required
By default, the platform maintains separate contexts for voice and text interactions, which can create disjointed customer experiences.
If a customer calls about an appointment and later texts about the same appointment, the agent won't automatically connect these interactions. However, this limitation can be overcome with custom development to create a unified context bridge between channels.
- Default behavior: Separate contexts
- Custom solution: Context bridging possible
- Implementation requires additional development
GrowwStacks specializes in implementing AI voice agents tailored to your specific business needs and customer workflows.
We handle the entire setup process including number porting, ElevenLabs configuration, and custom workflow development. Our team creates solutions that address the limitations mentioned in this article, such as context bridging between voice and text channels. We've deployed these systems for salons, medical practices, law firms, and e-commerce businesses.
- End-to-end implementation
- Custom context bridging solutions
- Free 30-minute consultation to discuss your needs
Ready to Transform Your Customer Service with AI?
Every missed call or delayed response costs your business revenue and customer trust. Our AI voice agent solutions can be deployed in as little as 3 days, handling 80% of routine inquiries automatically while providing 24/7 availability.