ChatBridge vs Wati

The differences that matter

Your AI, not theirs

Wati's Astra is their model, sold as its own SKU on top of your Wati plan and metered in credits. ChatBridge is bring-your-own-agent: plug in any LLM via an OpenAI-compatible key, and it drafts or answers live WhatsApp conversations with tool-call access to your contacts, tickets, and catalogue — included in your plan, with an approval queue until you trust it.

Both have MCP. Only one opens the agent

Wati ships an MCP server too — but it manages Wati's own Astra agent from Claude or ChatGPT. Ours exposes the runtime itself: your model handles live inbound WhatsApp with tool-call access to conversations, contacts, and tickets, and every send is gated by your workspace's autonomy setting. The question isn't who has MCP; it's whose model answers your customers.

Everything in the base plan

Ticketing pipelines, SLA tracking, CSAT surveys, automation workflows, drip sequences, and the AI layer are all part of every ChatBridge plan — no per-feature gating, no per-message platform fees (Meta's conversation charges pass through at cost).

Capability	ChatBridge	Wati
Customer-facing AI agent model	Any LLM — bring your own key (OpenRouter, OpenAI-compatible)	Astra — Wati's own model, separate metered SKU
MCP server	Opens the agent runtime — autonomy-gated sends	Yes — but it manages Wati's own Astra agent
Staged AI autonomy (draft → approve → autonomous) with audit trail		Configure-and-go with escalation rules
Outgoing webhooks with response actions		Webhooks only
Ticketing, SLA & CSAT in base plan		Check their plan tiers
Per-message platform markup	None — Meta charges pass through at cost	Check their pricing

Claims verified September 2026 against Wati's public product, docs, and pricing. Spotted something outdated? Tell us and we'll fix it.

Why it's different

The agent that talks to your customers is yours

Most platforms now ship an MCP server for data access — and keep the customer-facing AI locked to their own model, priced their way. ChatBridge is built the other way around: bring any LLM with your own key, and it runs your live WhatsApp conversations with tool-call access to contacts, tickets, and your catalogue.

Every AI action is logged, drafts go through an approval queue until the model's track record earns autonomy, and ChatBridge has no metered AI revenue to protect — the open agent layer is the business model.

Where Wati is a fair choice

Wati is an established product with a large customer base and a broad integration marketplace, and Astra is a genuinely capable agent builder — it spans web, voice, SMS, and RCS as well as WhatsApp, with voice cloning. If you want that channel breadth, or a widely-adopted conventional WhatsApp inbox, and you don't mind the AI being theirs and separately metered, Wati is a reasonable choice.

FAQ

Switching questions

Scope. Wati's MCP lets an AI assistant build and manage Wati's own Astra agent — useful, but the model talking to your customers is still theirs, on their meter. ChatBridge's MCP opens the runtime: your model, your key, your prompt, answering live WhatsApp with tool access to your CRM — and sends held behind an approval queue until you raise the autonomy level.

Yes. Your WhatsApp Business Account and phone number are Meta assets, not Wati's — connect the same WABA to ChatBridge via Embedded Signup and your number, display name, and quality rating carry over. Approved templates live on the WABA and sync in automatically. See /switch for the full walkthrough.

Conversation history from your previous tool doesn't transfer — Meta doesn't expose it. New messages flow into ChatBridge from the moment your WABA is connected; contacts can be imported via CSV with preview and validation.

You paste an OpenAI-compatible API key (any provider, including OpenRouter) in Settings → AI. Your model then drafts replies, or answers autonomously once its track record earns it — every suggestion is logged, and autonomy is unlocked per your own thresholds.

Moving from another platform? See the switch guide →