Does OpenClaw WhatsApp require a Meta developer account?

No. OpenClaw connects via the linked device model using whatsapp-web.js — the same way WhatsApp Web works. No API key or developer account needed.

OpenClaw WhatsApp Setup: Complete 2026 Guide

Q: Do I need a separate phone number for OpenClaw WhatsApp?

You don't need one to make it work, but a dedicated number is strongly recommended to keep your personal conversations separate from your AI agent's automated responses.

Q: Why does my OpenClaw WhatsApp connection keep dropping?

The most common causes are a WhatsApp app update disrupting the session, a server restart clearing session files, or hitting the 4-device linked device limit. Re-scan the QR code first, then configure session persistence.

Q: Is WhatsApp or Telegram better for OpenClaw?

Telegram is more stable and feature-rich. WhatsApp wins on adoption — it's where most people already communicate, making it the better channel for a personal AI agent you'll actually use daily.

Q: Can I run OpenClaw WhatsApp on a home server?

Yes, but you need to maintain a continuous session and configure persistent session storage so a server restart doesn't drop your WhatsApp connection and require a re-scan.

I spent a weekend running the OpenClaw WhatsApp setup on three different environments — local Node.js, a DigitalOcean droplet, and the App Platform. Two connected cleanly on the first scan. The third kept dropping after about six hours. The fix was embarrassingly simple once I found it, but it took longer to locate than the entire initial setup.

Here’s what I learned: the WhatsApp integration is the most-installed messaging skill on ClawHub, with over 5,000 active installations. That number surprised me. Telegram is technically easier to set up — no phone scanning, a proper bot token, cleaner developer experience. Yet people keep picking WhatsApp. The reason is obvious once you think about it: WhatsApp is already open on your phone dozens of times a day. Your AI agent lives where you already are, with zero behavior change required.

If you’re exploring what a personal AI assistant can actually do once it’s connected to your messaging, this setup guide is your starting point. The OpenClaw WhatsApp integration is the piece that makes it feel real — not a chatbot you visit, but an agent that’s already in your pocket.

Why WhatsApp Is the Harder (and Better) OpenClaw Channel

OpenClaw supports over 50 messaging platforms — WhatsApp, Telegram, Discord, Slack, Signal, iMessage, SMS, Microsoft Teams. Of those, Telegram is the easiest to configure. It has a proper bot API, supports files up to 2GB, handles inline buttons, and works with Markdown formatting out of the box.

WhatsApp is rated harder. The connection is less stable, the message types are more restricted, and the setup requires you to scan a QR code rather than paste a token. It’s also the platform most of your actual human contacts use. That’s the tradeoff.

OpenClaw was created by Peter Steinberger and released under the name Clawdbot in late 2025. It went through two name changes — Moltbot, then OpenClaw — before settling. It’s now crossed 150,000 GitHub stars. The core insight behind the project’s growth, according to users who adopted it early, is the channel integration model: you can start a conversation with your AI agent on WhatsApp, continue it on Telegram, and pick it up again on your desktop — same agent, same memory, same conversation. That’s the thing that made it take off.

What You Need Before the OpenClaw WhatsApp Setup

Before you scan anything, make sure you have these in place. Skipping one of them is how you lose 45 minutes to a problem that takes 30 seconds to prevent.

Node.js v22 or higher — This is not optional. The WhatsApp gateway uses whatsapp-web.js, a Node.js library that handles the linked device connection. Earlier Node.js versions may appear to work and then fail silently.
Bun runtime users: switch to Node.js for the gateway — If you’re running the rest of OpenClaw on Bun, that’s fine. But the WhatsApp (and Telegram) gateway has documented compatibility issues with Bun. Run that piece on Node.js specifically.
An existing WhatsApp account on your phone — OpenClaw connects via the linked device model, the same way WhatsApp Web works. Your phone stays the primary device. OpenClaw takes one of your 4 available linked device slots.
A dedicated phone number (strongly recommended) — Don’t use your personal number for automation. Get a separate number for your AI agent. This prevents accidental messages and keeps your personal conversations clean.
Your OpenClaw instance running — You should have the core platform deployed before attempting the WhatsApp connection. If you haven’t done that yet, start there.

That last point about linked device slots matters more than most guides mention. WhatsApp allows a maximum of 4 linked devices per account. Your phone is the primary. OpenClaw takes one slot. That leaves 3 for WhatsApp Web, WhatsApp Desktop, and anything else. If you’re already at 4 linked devices, you’ll need to remove one before the QR code scan will work.

How to Complete the OpenClaw WhatsApp Setup

The connection process itself is shorter than the prerequisites. Once your environment is correct, this is straightforward.

Navigate to your OpenClaw admin panel and go to the Channels or Connections section. The exact label depends on your version.
Select WhatsApp from the available channel options. OpenClaw will initialize the whatsapp-web.js gateway and display a QR code.
Open WhatsApp on your phone. Go to Settings → Linked Devices → Link a Device.
Scan the QR code displayed in your OpenClaw admin panel. The QR code expires after about 60 seconds, so have your phone ready before you click to generate it.
Wait for the confirmation. You’ll see a green connected status in OpenClaw and a new entry in your WhatsApp Linked Devices list on your phone.
Send a test message. Open WhatsApp on your phone, message yourself (or the number your OpenClaw is connected to), and verify you get a response from the agent.

If you’re deploying to a server rather than running locally, the QR code step is the same — but you’ll need to access it through your browser pointed at the server’s admin UI. DigitalOcean’s App Platform is the cleanest option for this if you don’t want to manage server configuration yourself. The one-click droplet gets you most of the way there with less manual setup than a raw VPS, but you’ll still handle some configuration.

Stay in the loop

Get the latest AI insights delivered to your inbox.

Join Free

Why the Linked Device Model Is Actually Good News

Here’s the part most setup guides skip. When people hear that OpenClaw uses the same linked device mechanism as WhatsApp Web, their first instinct is to treat it as a limitation — no API, no official bot registration, feels unofficial.

But the linked device model means something important: your AI agent uses your actual WhatsApp account. It receives messages from your real contacts. It can participate in group chats if you configure it to. It shows up as you — not as a bot account with a generic name that people have to add separately. There’s no new phone number to buy. No Meta developer account to apply for. No waiting for API access approval.

The tradeoff is stability. Because whatsapp-web.js mimics the WhatsApp Web browser session rather than using an official API, WhatsApp can — and occasionally does — change something that temporarily breaks the connection. The library’s maintainers usually catch up quickly, but there’s a gap window. For Telegram, there’s no such risk because the bot API is official and stable.

For a personal AI agent that handles your own messages, that tradeoff usually tilts toward WhatsApp. For a business automation handling hundreds of customer conversations, Telegram or an official WhatsApp Business API integration deserves serious consideration. Know which use case you’re solving before you commit.

When the OpenClaw WhatsApp Connection Breaks

This is the scenario that catches people off guard. Everything works Monday morning. Wednesday evening you check your phone and realize your agent hasn’t responded to anything since Tuesday night. The OpenClaw admin panel shows the WhatsApp channel as disconnected.

There are a handful of reasons this happens, in rough order of frequency:

Your phone’s WhatsApp app updated overnight — Sometimes a WhatsApp update changes the session behavior enough to drop linked devices. Open WhatsApp on your phone, check Linked Devices, and re-scan the QR code.
The server process restarted without a session persistence layer — By default, whatsapp-web.js stores the session locally. If your server restarts and the session file isn’t in a persistent volume, the connection resets. Fix this by configuring a persistent session store.
You hit the 4-device limit and WhatsApp auto-rotated your oldest linked device — If someone in your household connected a new device to your WhatsApp, it may have bumped OpenClaw. Check your Linked Devices list.
The Node.js version drifted — If you’re running this on a server that auto-updates, an unexpected Node.js version change can break compatibility. Pin your Node.js version.
whatsapp-web.js had a breaking upstream change — Check the library’s GitHub issues. If there’s a recent update, upgrading the package usually resolves it within a day of the breakage.

What WhatsApp Can’t Do That Telegram Can

Before you go all-in on the WhatsApp channel, be honest with yourself about where these limits will matter for your use case.

File size restrictions are tighter. Telegram supports files up to 2GB. WhatsApp’s limits are much lower, and the channel inherits those restrictions.
No inline buttons or rich formatting. Telegram’s bot API supports interactive buttons in messages. WhatsApp messages are plain text or media — no interactive UI elements.
Session stability is lower. The linked device model can drop. The official Telegram bot API does not have this problem.
No official API means no guarantees. WhatsApp can change their client behavior at any time. whatsapp-web.js adapts, but there’s always a lag window.
Long-polling for Telegram, session-based for WhatsApp. Telegram’s long-polling mode means you don’t need a public IP or domain to run it. WhatsApp requires the session to be maintained continuously — more infrastructure consideration for home server deployments.

None of these are dealbreakers for personal use. They matter more if you’re evaluating which channel to use for business automation at scale. For a personal AI agent that handles your messages, schedules, and follow-ups, the WhatsApp channel covers everything you need.

If you’re still deciding which AI agent platform to build on, or whether to self-host OpenClaw versus using a managed solution, the channel question is secondary to the infrastructure question. Get the platform right first.

Signs Your OpenClaw WhatsApp Setup Is Working

Don’t assume the green status light in the admin panel is the whole story. Run through these checks after initial setup and again after any server restart or WhatsApp update.

Beacon’s got your setup sorted — just follow the light. 🔦

The channel shows as connected in the OpenClaw admin panel — Green status, no error messages.
Your phone’s Linked Devices list shows the OpenClaw session — It should appear as a named device. If it’s not there, the connection didn’t complete.
A test message from your phone gets a response — Send a simple message from your WhatsApp to the connected number (or to yourself if using the same number). The agent should respond within a few seconds.
The session survives a server restart — Restart your OpenClaw server, wait 60 seconds, then send another test message. If this fails and the previous test passed, you have a session persistence issue to fix.
Messages appear in the OpenClaw conversation log — Beyond getting a response, verify the message was actually logged in your agent’s history. This confirms the full pipeline is working, not just the gateway.

Your OpenClaw WhatsApp Setup Checklist

Run through this before you call the setup done. Each item takes under 2 minutes to verify.

Verify Node.js version is 22 or higher — Run node --version in your terminal. If it’s below v22, upgrade before proceeding. Don’t skip this — compatibility failures here are silent and confusing.
If using Bun, confirm the WhatsApp gateway is running on Node.js — Bun handles most of OpenClaw fine. The gateway is the exception. Check your process configuration.
Check your Linked Devices count on your phone — You need at least one free slot out of 4. Remove an unused device if needed.
Use a dedicated phone number, not your personal number — If you’re using your personal number, this is the moment to switch. The setup cost is low; the alternative (accidental AI responses to your family) is annoying.
Configure session persistence before going live — If your server restarts, you want the WhatsApp session to survive. Set up a persistent volume or session store before you rely on this for anything important.
Send a test message and verify the full response cycle — Don’t stop at the green light. Confirm the agent actually responds, and that the exchange appears in your conversation log.
Set a reminder to check the connection in 48 hours — New setups often show their first instability point within the first two days. Catch it early rather than discovering it when something time-sensitive fails.
If you’re on a managed platform like BrainRoad — The QR scan step is the same, but session persistence and server restarts are handled for you. Budget roughly $29/month versus managing your own infrastructure.

Stay in the loop

Get the latest AI insights delivered to your inbox.

Join Free

What This Means for Your Personal AI Strategy

The OpenClaw WhatsApp setup uses the linked device model — no new phone number, no Meta developer account. Your existing WhatsApp account hosts the agent in one of 4 available linked device slots.
Node.js v22 or higher is required. The Bun runtime causes compatibility issues with the WhatsApp gateway specifically — run that piece on Node.js even if the rest of your stack uses Bun.
WhatsApp is harder to set up than Telegram and less stable, but it’s where most people already spend their time — making it the better choice for a personal AI agent you’ll actually use.
Session persistence is the most common point of failure after initial setup. Configure a persistent session store before you rely on this for anything important.
The WhatsApp channel has over 5,000 active installations on ClawHub — more than Telegram (3,200+) or Discord (2,800+). Adoption reflects where people actually communicate, not which channel is technically superior.
For personal use, the WhatsApp channel covers everything you need. For high-volume business automation, evaluate the official WhatsApp Business API or the Telegram channel instead.

Frequently Asked Questions: OpenClaw WhatsApp

Do I need a separate phone number for OpenClaw WhatsApp?

You don’t need one to make it work — OpenClaw can connect to your existing personal number. But using a dedicated number is strongly recommended. If your agent is connected to your personal number, it can respond to messages from your contacts automatically, which creates awkward situations fast. A separate number keeps your personal conversations separate from your AI agent’s activity.

Does the OpenClaw WhatsApp setup require a Meta developer account?

No. OpenClaw connects using the linked device model — the same mechanism WhatsApp Web uses. There’s no Meta developer account, no API key, and no application approval process. You scan a QR code with your existing WhatsApp app. The library handling this connection is whatsapp-web.js, an open-source Node.js project.

Why does my OpenClaw WhatsApp connection keep dropping?

The most common causes are: a WhatsApp app update on your phone that disrupted the session, a server restart that cleared the session file, or reaching the 4-device limit and having WhatsApp auto-remove your oldest linked device. Start by opening WhatsApp on your phone, removing the OpenClaw linked device, and re-scanning the QR code. If it keeps dropping, configure session persistence so the connection survives server restarts.

Can I run OpenClaw WhatsApp on a home server?

Yes, with one caveat. Unlike the Telegram channel, which uses long-polling (no public IP or domain needed), the WhatsApp connection requires maintaining a continuous session. This is manageable on a home server, but you need to ensure the process stays running and that session files are stored persistently. A server restart without persistence will drop the connection and require a re-scan.

Is WhatsApp or Telegram better for OpenClaw?

Telegram is easier to set up, more stable, and supports richer message types (files up to 2GB, inline buttons, Markdown). WhatsApp is harder and more restricted — but it’s where most people already communicate. If you want an AI agent you’ll check dozens of times a day without changing any habits, WhatsApp wins on adoption. If you’re building something for developers or need more reliability, start with Telegram.

OpenClaw WhatsApp Setup: Complete Guide for 2026