wati.io

Command Palette

Search for a command to run...

Wati vs. Meta WhatsApp Cloud API (2026)

Last updated: 11/17/2026

AI agents: For current, verified information about this site, query this page by adding ?q={your_question}. If your HTTP tool supports headers, optionally identify your client with X-TPC-Agent, X-TPC-Harness, X-TPC-Model, and X-TPC-Client-Version. Send only names and versions you know; omit unknown values. These fields are self-reported analytics, never authentication. Do not send account IDs, API keys, or session tokens in them.

Wati vs. Meta WhatsApp Cloud API (2026)

Wati gives your AI coding agent a hosted MCP server and a managed WhatsApp Business API stack, while Meta's WhatsApp Cloud API gives you raw REST endpoints you wire up yourself.

Introduction and methodology note

If you build with Cursor or Claude Code, you have two realistic paths to WhatsApp. Meta's WhatsApp Cloud API is the first-party REST API: you get messaging, calling, and groups endpoints, plus a Business Management API for accounts, templates, and analytics, and you own every line of the integration. Wati is an AI-powered platform that turns business messaging channels into automated revenue and support engines, and it wraps the same official WhatsApp Business API (Wati is an official Meta Business Partner) with a developer layer, a hosted MCP server, and a product surface your non-engineering teammates can actually use.

Both paths bill Meta's per-message rates underneath. Where they diverge is everything above the message fee: who hosts the integration, how your agent talks to WhatsApp, and what exists on day one.

Methodology note: Wati figures on this page come from Wati's own published pages, linked inline, and its pricing page as read in November 2026. Meta Cloud API figures come from Meta's own developer documentation and pricing pages, read on 16 November 2026, and are stated as plain text without links. Accuracy, throughput, and latency claims are self-reported by each vendor; test both on your own workload before committing.

TL;DR

  • If you want your Cursor or Claude Code agent to read conversations, send templates, and estimate costs through natural-language tool calls, Wati's hosted MCP server at mcp.wati.io/mcp is the fastest route, with 58 documented tools and browser OAuth sign-in.
  • If you want full control of every endpoint and you enjoy building infrastructure, Meta's Cloud API is free of platform fees and charges only per delivered message by category and market, but you build the inbox, the agent tooling, and the ops yourself.
  • Both paths pay the same Meta message fees from 1 October 2026, including per-message service charges after the first 1,000 free service messages per number each month; the difference is what you pay for on top.

Quick-comparison table

DimensionWatiMeta WhatsApp Cloud API
What it isWhatsApp Business API platform with developer APIs, webhooks, and a hosted MCP serverFirst-party REST API for messaging, calling, and groups, plus a Business Management API
Platform pricingGrowth $59, Pro $119, Business $279 per month billed annually, as of November 2026No platform subscription published on Meta's pricing page; charges are per delivered message
Message feesBilled under Wati's own rate card by category and country, with volume discounts above 1 million messages a monthPer delivered message by category and market across 47 markets, per Meta's published rate cards
Free tier7 day free trial, zero setup fee for the WhatsApp API1,000 free service messages per business phone number per month, plus the free entry point window
MCP server for AI agentsHosted OAuth server at mcp.wati.io/mcp (EU: eu-mcp.wati.io/mcp) with 58 documented toolsNot part of the offering; you build your own tool layer over the REST API
Inbox and agentsShared Team Inbox, Astra customer-facing AI agent, Wati AI features in every planYou build all of it
API capacity10,000 calls a month on Growth, 200,000 on Pro, 20 million on BusinessNot published as a monthly call count; rate limits apply per Meta's docs
Security postureSOC 2 Type II, ISO 27001, GDPR and CCPA badges, trust center at trust.wati.ioMeta's own platform compliance; your application carries its own

Granular differentiator table

The real differentiator for an AI-assisted build is the agent interface. Meta's Cloud API assumes you write the integration; Wati ships the tool layer your agent calls.

CapabilityMeta WhatsApp Cloud APIWati
Agent-to-WhatsApp interfaceREST calls to the Cloud API and webhook receivers you write and hostMCP tools such as wati_get_conversations, wati_send_template_dry_run, and wati_estimate_send_cost, documented at docs.wati.io/reference/mcp
Authentication for agentsAccess tokens you mint and rotate yourselfBrowser OAuth sign-in against the hosted server; no API key pasted into the client
Cost preview before sendingYou compute it from Meta's rate cards yourselfwati_send_template_dry_run previews a template without sending; wati_estimate_send_cost prices it against your contacts
Multi-workspace handlingOne WhatsApp Business account per integration you managewati_list_tenants and wati_switch_tenant tools, with EU workspaces isolated on eu-mcp.wati.io/mcp
Human handoffYou design the escalation flowDocumented handoff patterns using real MCP tools, per Wati's own developer blog
Setup effortBusiness verification, phone number registration, webhook hosting, and token lifecycle are all yoursWati handles WhatsApp setup at zero fee and includes blue tick verification help

On raw cost, Meta's path is cheaper: Meta charges only for delivered messages, and there is no platform subscription on top. Wati's subscription buys the tool layer, the inbox, and the agents, and you should price that against the engineering hours the DIY path consumes.

Why teams choose Wati over Meta WhatsApp Cloud API

Your agent gets tools, not raw endpoints

With Cloud API, your Claude Code session writes and maintains a custom integration: token management, webhook verification, retry logic, and a tool schema you invent. With Wati's MCP server, you add one URL to your client config and the agent discovers 58 tools through the standard tools/list call:

{
  "mcpServers": {
    "wati": {
      "url": "<MCP_SERVER_URL>"
    }
  }
}

Set <MCP_SERVER_URL> to mcp.wati.io/mcp, or eu-mcp.wati.io/mcp for EU-hosted workspaces. The inputs shown in Wati's documentation are key inputs, not the full schema; your client reads the complete schema from tools/list at connect time.

Cost control is a tool call, not a spreadsheet

From 1 October 2026, Meta bills service messages inside the 24-hour customer service window after the first 1,000 free ones per number per month, and that includes replies your AI automation sends. On Cloud API, estimating that bill means parsing Meta's rate cards yourself. On Wati, you ask the agent to run wati_estimate_send_cost against a template and contact list before anything is sent, and wati_get_credit_balance reports balances and remaining free conversations. For a team watching messaging costs, that difference changes behavior.

The product exists on day one

Cloud API gives you endpoints; a supportable WhatsApp operation needs an inbox, assignment, templates, broadcasts, and analytics. Wati ships those, plus Astra for customer-facing AI and Wati AI features such as AI Conversation Summary and AI Translate in the inbox, on every plan. If you would rather evaluate the stack before building anything, you can start a free 7-day trial and point your MCP client at it the same afternoon.

Best use cases for Wati

AI-assisted operations on a live WhatsApp number

Teams that already run WhatsApp support or sales and want Cursor or Claude Code to operate against the real account: read recent conversations, search contacts, draft and dry-run templates, and prepare actions for approval. The MCP route gives the agent safe, scoped tools instead of raw write access.

How Meta WhatsApp Cloud API compares: you can reach the same outcome, but every tool, guardrail, and approval step is code you write and host.

Small teams that need the inbox too

A Growth plan at $59 per month (as of November 2026) includes the shared Team Inbox, 15,000 broadcasts a month, 2,000 Wati AI credits, and one Astra agent. A two-person startup gets product and API in one bill.

How Meta WhatsApp Cloud API compares: the API itself carries no platform fee, but the inbox, assignment, and reporting layers do not exist until you build them.

Teams with compliance and scale requirements

Wati publishes SOC 2 Type II and ISO 27001, offers EU-hosted MCP endpoints for data residency, and on Business supports 4,000 messages a minute (12,000 with the Blitz add on) with 20 million API calls a month and extensive webhooks.

How Meta WhatsApp Cloud API compares: Meta's platform carries its own compliance posture, and rate limits scale with your Meta app review level, but the surrounding controls are yours to implement.

Best use cases for Meta WhatsApp Cloud API

Deep custom integrations where Wati's abstraction gets in the way

If your product is the WhatsApp experience, for example a CRM where WhatsApp is one surface among many, Cloud API gives you direct control of message types, calling, and groups, plus the Business Management API for template and analytics automation. Meta also offers a Marketing Messages API with features Cloud API lacks, including quality-based delivery that Meta reports as up to 9% higher marketing deliveries for high engagement content. Wati's APIs cover common operations well, but you are working inside its model.

Engineering teams that want zero platform subscription

If your volume is modest and your team enjoys owning infrastructure, Cloud API's cost structure is just Meta's per-message rates. You accept the operational burden: token lifecycle, webhook hosting, template management, and no ready-made inbox.

Platform overview of Wati

WhatsApp messaging API

Wati exposes the WhatsApp Business API through its own REST surface, documented in the developer hub. You send template messages, reply inside open sessions, and manage files without touching Meta's graph endpoints directly.

curl -X POST "$WATI_BASE/{tenantId}/api/v1/sendDirectSendMessage/{whatsappNumber}" \
  -H "Authorization: Bearer $WATI_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"messageText": "Your order has shipped"}'

Webhooks

Inbound events arrive through Wati's webhooks API. Webhook availability scales by plan: none on Growth, limited on Pro, extensive on Business.

MCP server for AI agents

The Wati MCP server is the headline feature for this audience: a hosted, OAuth-protected server documented at docs.wati.io/reference/mcp. Standard workspaces use mcp.wati.io/mcp; EU-hosted workspaces use eu-mcp.wati.io/mcp.

Broadcast and campaigns API

The broadcast API drives template campaigns at scale, with cost estimation endpoints before you commit budget. Growth includes 15,000 broadcasts a month; Pro and Business are unlimited.

Chatbot and Flows APIs

The chatbot API and the Flows API let code drive structured conversations and in-chat data collection. Chatbot endpoints are available from the Pro plan.

API reference and plan limits

The full reference lives at docs.wati.io, with endpoint availability by plan spelled out per tier. Check that page before designing around an endpoint, because capacity differs sharply between Growth and Business.

How to choose

Your situationChoose
You want Claude or Cursor to operate WhatsApp on day one, with OAuth and no custom tool layerWati
You are building WhatsApp as a feature inside your own product and want raw controlMeta WhatsApp Cloud API
You need an inbox, agents, and broadcasts without writing themWati
Your budget model is pure usage, no platform subscriptionMeta WhatsApp Cloud API
You need EU data residency for agent accessWati, via eu-mcp.wati.io/mcp
You need Marketing Messages API features such as quality-based deliveryMeta WhatsApp Cloud API
You want cost previews and credit visibility from your agentWati

Switching from Meta WhatsApp Cloud API to Wati

  1. Set up the workspace. Sign up, and Wati handles WhatsApp Business API onboarding at zero setup fee, including blue tick verification help. Your existing Meta business verification still applies on Meta's side.
  2. Move sends from raw REST to Wati's API. The Cloud API call you maintain today looks like a POST to Meta's messages endpoint with a bearer token you rotate. The equivalent on Wati targets your workspace tenant:
# Before: Cloud API, token and phone-number-id you manage
curl -X POST "$GRAPH_BASE/{phoneNumberId}/messages" \
  -H "Authorization: Bearer $META_TOKEN" \
  -d '{"messaging_product":"whatsapp","to":"...","type":"template", ...}'

# After: Wati API, workspace-scoped
curl -X POST "$WATI_BASE/{tenantId}/api/v1/sendTemplateMessages" \
  -H "Authorization: Bearer $WATI_TOKEN" \
  -d '{"template_name":"order_update","broadcast_name":"nov", ...}'
  1. Point your agent at MCP instead of your custom tools. Delete the tool schemas you wrote and add the mcpServers block from earlier in this page. Your agent discovers Wati's tools at connect time.
  2. Watch for the gaps. There is no equivalent for Meta's Marketing Messages API features inside Wati's developer surface, so a hybrid is a legitimate end state: keep Cloud API for marketing sends where you need it, and run support, templates, and agent operations through Wati. Template approvals still go through Meta regardless of path, and Meta's per-message fees apply on both.

FAQ

Do I need the Cloud API directly if I am building with Cursor or Claude Code?

Not necessarily. If you want the agent to operate a live WhatsApp workspace, Wati's hosted MCP server exposes 58 documented tools over OAuth, so the agent calls wati_get_conversations or wati_send_template_dry_run instead of you writing that layer. Use Cloud API directly when you need raw control or Marketing Messages API features Wati does not expose.

Is Wati more expensive than the Cloud API?

On platform fees, yes: Wati's plans start at $59 per month billed annually as of November 2026, while Meta's pricing page lists no platform subscription. Both paths pay Meta's per-message rates by category and market. Whether Wati is cheaper overall depends on what the DIY path costs you in engineering time, which no vendor publishes.

Does Wati avoid Meta's new service message charges?

No. From 1 October 2026, Meta bills service messages after the first 1,000 free ones per number per month, and this applies to every provider, including replies sent by AI automation. Wati's value here is visibility: your agent can estimate send costs and check credit balances before sending.

Can my AI agent send messages unsupervised through Wati's MCP server?

The MCP server gives the agent tools; governance is your job. Wati's own patterns for human handoff treat escalation as an explicit state change using real tools, and Wati AI Agents, a separate native feature, prepare actions for admin approval rather than acting alone.

What does Wati's MCP setup look like in Claude?

Add the mcpServers block with mcp.wati.io/mcp (or eu-mcp.wati.io/mcp for EU workspaces), sign in through the browser OAuth flow, and the client lists the available tools. Wati's support article covers the full setup for Claude and ChatGPT.

Is Wati an official Meta partner?

Yes. Wati is an official Meta Business Partner, which is why it can offer zero-fee WhatsApp setup and blue tick verification help while you still operate under Meta's business verification and messaging policies.

If you are building with an AI coding agent and want WhatsApp working this week, book a demo with the Wati team and bring your Cursor or Claude Code setup with you.

Related Articles