AI Agents That Live in
WhatsApp, SMS & Email
Your users don't live in chat widgets. They live in WhatsApp, they respond to SMS, they read emails. Deploy AI agents where your customers actually are.
Deploy in 5 minutes. No credit card required.
The Problem with AI Agents Today
LLM APIs (OpenAI, Anthropic)
Powerful AI, but no native channels. You still need to integrate WhatsApp, SMS, build session management, handle tool calls... months of work before your first customer conversation.
CPaaS (Twilio, Vonage)
Great messaging infrastructure, but no AI. You bring the LLM, the vector DB, the orchestration... and pay per message on top. Building blocks, not a solution.
Zavu: Both, Combined
Native WhatsApp, SMS, Email + Built-in AI agents + Knowledge bases + Flow builder + Tool integration. One platform. Deploy in minutes, not months.
Simple API, Powerful Agents
Configure once in the dashboard, or programmatically via API.
import Zavu from '@zavudev/sdk';
const zavu = new Zavu('YOUR_API_KEY');
// Create an agent for your sender
const agent = await zavu.agents.create({
senderId: 'sender_abc123',
name: 'Customer Support',
provider: 'openai',
model: 'gpt-4o-mini',
systemPrompt: `You are a helpful customer support agent for Acme Inc.
Be friendly, concise, and helpful. If you can't help,
offer to transfer to a human agent.`,
});
// Add a knowledge base with your FAQs
const kb = await zavu.agents.knowledgeBases.create({
agentId: agent.id,
name: 'Product FAQs',
});
await zavu.agents.knowledgeBases.documents.create({
knowledgeBaseId: kb.id,
title: 'Return Policy',
content: 'Returns accepted within 30 days with receipt...',
});
// Enable the agent - it now responds on WhatsApp, SMS & Email
await zavu.agents.update({
agentId: agent.id,
enabled: true,
});Why Zavu for AI Agents?
The only platform with native channels AND native AI agents.
| Feature | Zavu | OpenAI | Twilio |
|---|---|---|---|
| Native WhatsApp/SMS/Email | |||
| Built-in AI Agents | |||
| Knowledge Base (RAG) | |||
| Visual Flow Builder | |||
| Multi-channel (same agent) | |||
| Human Handoff | |||
| Tool/Webhook Integration | |||
| Session State Management |
Everything You Need to Build AI Agents
Multi-Provider LLM
Use OpenAI, Anthropic, Google, or Mistral. Switch providers without code changes. Or use Zavu managed at cost.
Knowledge Base (RAG)
Upload docs, FAQs, or product info. Automatic chunking, embeddings, and semantic search. Your agent stays accurate.
Flow Builder
Design guided conversations with conditions, data collection, and branching logic. No code required.
Tool Integration
Connect your APIs via webhooks. Check order status, book appointments, or query your database - all from the conversation.
Human Handoff
Seamless transfer to human agents when needed. Full context preserved. Notify your team via webhook.
Multi-Channel Native
One agent responds on WhatsApp, SMS, AND Email. No duplicate logic. Customers reach you however they prefer.
One Agent, All Channels
Your customers choose how to reach you. Your agent responds everywhere.
Same knowledge base. Same flows. Same tools. Different channels.
Use Cases
Customer Support
Answer FAQs, handle returns, check order status - 24/7 on WhatsApp.
Lead Qualification
Capture leads, ask qualifying questions, and book meetings automatically.
Appointment Reminders
Send reminders, handle rescheduling, confirm attendance with AI.
E-commerce Assistant
Product recommendations, inventory checks, shipping updates.
Frequently Asked Questions
How is this different from building agents myself?
Without Zavu, you need: LLM API + Vector DB + Channel integrations (Twilio/Meta) + Session management + Tool orchestration + Human handoff logic. That's 6+ services to manage. With Zavu, it's one platform, everything integrated, deploy in minutes.
What LLM providers do you support?
OpenAI (GPT-4, GPT-4o, GPT-4o-mini), Anthropic (Claude 3.5 Sonnet, Claude 3 Opus), Google (Gemini), and Mistral. You can bring your own API key or use Zavu managed at cost (no markup).
How does the Knowledge Base work?
Upload text, markdown, or documents. We automatically chunk the content, generate embeddings, and enable semantic search. When your agent receives a message, relevant context is retrieved and injected into the LLM prompt. Your agent stays accurate and grounded.
Can I connect my own APIs and databases?
Yes! Create tools with webhook endpoints. Define the parameters, and your agent can call your APIs during conversations. Check inventory, create tickets, query databases - whatever your business needs.
What about human handoff?
Use a transfer step in your flows or let the LLM decide when to escalate. When transfer happens, we can notify your team via webhook with full conversation context. The human agent sees everything that was discussed.
How much does it cost?
You pay for: 1) Channel costs (SMS, WhatsApp carrier fees - same as any provider), 2) LLM tokens at provider rates (no markup if using Zavu managed). No platform fee for AI agents. No per-agent pricing. Build as many agents as you need.
Ready to Deploy AI Agents?
Stop building infrastructure. Start deploying agents. Your customers are waiting on WhatsApp, SMS, and Email.