Rabt WhatsApp CRM
AI Chatbot & Agents

One WhatsApp number, a whole team of AI agents.

Create named AI agents with their own persona, model, and tools — they hand off to each other and to your human team, and answer from a knowledge base built from your own documents.

AI Agents — 3 Active

Sara — Sales

GPT-4o · friendly sales

Replying

Omar — Support

Claude 3.5 · precise support

Layan — Front Desk

Gemini 1.5 · initial routing

Conversation handed off from Sara to Omar (technical question) ✓

Runs on the official WhatsApp Business API

Multi-agent handoff
Instant replies
Shared team inbox
RAG knowledge base
Named agents

Give every agent its own name, model, and job

Set up as many AI agents as your plan allows, each with its own AI provider and model (OpenAI, Claude, Gemini, and others), its own persona, and its own tools — a sales agent, a support agent, a front-desk router, whatever your business needs.

  • Independent provider/model per agent — mix OpenAI, Claude, and Gemini in the same account
  • Custom persona and tone per agent, not one generic bot voice
  • Per-agent tool access — only give an agent the actions it actually needs
AI Agents — 3 Active

Sara — Sales

GPT-4o · friendly sales

Replying

Omar — Support

Claude 3.5 · precise support

Layan — Front Desk

Gemini 1.5 · initial routing

Conversation handed off from Sara to Omar (technical question) ✓
RAG knowledge base

Answers pulled from your own documents, not guesses

Upload PDFs or plain text and RABTCRM indexes them with pgvector so agents answer from your real policies and product info. Search is language-agnostic — a customer can ask in Arabic and still get an accurate answer from English-language source documents.

  • PDF and text upload, indexed automatically
  • Per-agent knowledge base — different agents can draw from different document sets
  • Language-agnostic query matching across your uploaded content
Knowledge Base — 3 docs indexed
Return-Policy.pdfIndexed
Product-Guide.pdfIndexed
FAQ.txtIndexed
"How long is your return window?"
Matched from Return-Policy.pdf

14 days from delivery date, with original proof of purchase.

Handoff between agents

Agents pass conversations to each other — and to your team

A front-desk agent can route a technical question straight to your support agent, carrying full conversation context with it. Any agent can also hand a chat to a human teammate at any point — it lands in the shared team inbox like any other conversation.

  • Automatic handoff rules between named agents
  • Full context carried across the handoff, nothing repeated
  • Escalate to a human agent at any time, from any AI agent
Agent Setup — Sara

Provider / Model

OpenAI · GPT-4o

Persona

Friendly, concise, uses emoji

Enabled tools

Search knowledge baseCreate WooCommerce orderHand off to Support
Changes save automatically

What's included

Everything a multi-agent WhatsApp chatbot needs

Built on the shipped Multi-AI-Agent system and RAG knowledge base — not a roadmap item.

Named agent roster

Create multiple agents, each with its own identity, persona, and job — not a single generic bot.

Agent-to-agent handoff

Agents transfer conversations between each other automatically, with full context preserved.

RAG knowledge base

pgvector-backed document search so answers come from your uploaded PDFs and text files.

Per-agent AI provider

Choose OpenAI, Claude, Gemini, or other providers independently for each agent.

Human escalation

Any agent can hand off to a human teammate — it appears in the shared team inbox instantly.

Official WhatsApp API

Runs on the same Meta Cloud API connection as the rest of your WhatsApp messaging.

Questions

AI Chatbot & Agents FAQ

QHow many AI agents can I create?

This depends on your plan — each plan sets an agent limit. Every agent gets its own name, persona, and AI provider (OpenAI, Claude, Gemini, and others).

QHow does handoff between agents work?

Any agent can hand a conversation to another agent mid-chat — for example, a front-desk agent routing a technical question to your support agent — carrying the full conversation context with it.

QWhere does the chatbot get its answers from?

From your own RAG knowledge base: upload PDFs or text files, and the agent does a vector search (pgvector) over them to answer from your actual content instead of guessing.

QDoes it work in a different language than the uploaded documents?

Yes — knowledge base search is language-agnostic. A customer can ask in Arabic and still get an accurate answer even if the source documents were uploaded in English.

QDoes AI replace my support team entirely?

No — any agent can hand the conversation to a human teammate at any point, and it shows up in the shared team inbox like any other chat.

Put a multi-agent AI chatbot on your WhatsApp number

Start free, connect your WhatsApp Business API, and configure your first agent in minutes.

No credit card required.