Human Track Progress3 / 6

🏁Getting Started 🔧Skills & Models 📱Channels & Messaging ⚙️Automation & Workflows 🛡️Security & Operations 🧠Philosophy & Design

Dashboard/Human Track/Channels & Messaging

📱Human Track

Channels & Messaging

Connect your AI to your phone and manage multi-channel, multi-context communication

Learning Objectives

Connect OpenClaw to at least one messaging channel (WhatsApp, Telegram, or Discord)
Operate your AI across multiple platforms simultaneously
Configure group chat behavior so your AI knows when to speak and when to stay quiet

Connect to Your Phone

OpenClaw’s most powerful feature: your AI in your pocket.

Connect it to WhatsApp, Telegram, or Discord, and you can message your AI like texting a friend. No need to open a terminal. No need to be at your computer.

Channel comparison:
- WhatsApp — Easiest setup. Scan a QR code and you’re connected. Best for personal use.
- Telegram — Most flexible. Supports bots, groups, and channels. Best for power users.
- Discord — Best for communities. Your AI can participate in server channels.

How it works: OpenClaw runs a gateway on your computer that bridges your messaging app to the AI. Messages come in through WhatsApp → gateway processes them → AI responds → response goes back to WhatsApp.

Important: Your computer (or VPS) must stay on for this to work. If your laptop goes to sleep, your AI goes offline. We cover always-on deployment in the Security & Ops category.

Privacy: Messages are processed on YOUR machine. They are not sent to Claw Academy or any third party. Your conversations stay between you and your AI provider.

Key Principle

Your AI should be as accessible as a friend on WhatsApp — not locked behind a terminal on your laptop. Connect a channel and your AI is always one message away.

Step-by-Step Instructions

Step 1: Choose a channel
We recommend WhatsApp for the easiest start. Check the OpenClaw docs for your chosen channel:

https://docs.openclaw.ai/channels

Step 2: Connect
For WhatsApp, run:

openclaw qr

Scan the QR code with your phone’s WhatsApp app.

Step 3: Send a test message
Open WhatsApp on your phone. Send your AI a message: "Hello from my phone!"
You should get a response within seconds.

Step 4: Try it anywhere
Send your AI a real question while away from your computer. Feel the magic of having an AI assistant in your pocket.

Common Mistakes

❌ QR code expired — QR codes expire after a few minutes. Generate a new one if it doesn’t work.

❌ AI doesn’t respond on WhatsApp — Make sure the OpenClaw gateway is running: openclaw gateway status

❌ Computer went to sleep — Your AI goes offline when your computer sleeps. Keep it plugged in and disable sleep, or use a VPS (covered in Security & Ops).

Multi-Channel Operations

One AI, multiple platforms. Your AI can simultaneously operate on WhatsApp (personal), Discord (community), Telegram (work), and your terminal (development).

Why multi-channel matters:
- Different contexts need different behaviors
- You want your AI accessible everywhere, not just one app
- Team members can interact with your AI on their preferred platform

Behavioral adaptation:
Your AI should behave differently based on context:
- Direct messages → Full, detailed responses
- Group chats → Brief, only speak when addressed or adding value
- Terminal → Technical, code-heavy responses

This is configured in AGENTS.md with rules like:

## Channel Rules
- WhatsApp DMs: Full responses, personal tone
- Discord server: Brief, only respond when mentioned
- Terminal: Technical, code-focused

The human rule: In a group chat, humans don’t respond to every single message. Neither should your AI. If your AI wouldn’t naturally add to the conversation, it should stay quiet.

Key Principle

Your AI should adapt its behavior to the social context — just like a human knows to behave differently in a board meeting vs a coffee chat with friends.

Step-by-Step Instructions

Step 1: Connect a second channel
If you connected WhatsApp, now connect Telegram or Discord.

Step 2: Set up behavioral rules
Add channel-specific rules to AGENTS.md (see examples above).

Step 3: Test both channels
Send a direct message on one channel and a group message on another. Verify the behavior differs appropriately.

Step 4: Fine-tune
Adjust the rules based on real usage. Too chatty in groups? Tighten the rules. Too quiet? Loosen them.

Group Chat Behavior

Group chats are the trickiest environment for an AI. Respond to everything and you’re annoying. Never respond and you’re useless. The art is knowing when to speak.

The golden rules of AI in group chats:

1. Only speak when addressed. If someone mentions your AI’s name or replies to its message, respond. Otherwise, stay quiet by default.

2. Add value or stay silent. Before responding, your AI should ask: "Does my response add something the group doesn’t already know?" If the answer is no, don’t respond.

3. One response per trigger. Never send multiple messages in a row. If the answer needs detail, put it in one well-structured message — not three fragments.

4. Match the group’s energy. A casual friend group gets casual responses. A work Slack channel gets professional ones. Read the room.

5. Never dominate. If your AI has responded to 3 messages in a row, it should go quiet and let humans talk. No one wants a conversation dominated by a bot.

Configuring group behavior in AGENTS.md:

## Group Chat Rules
- Only respond when directly mentioned (@Atlas) or replying to my message
- If not mentioned, only speak when I can add factual information not yet stated
- Maximum one response per message — never send multiple messages in sequence
- After 3 consecutive responses, go quiet until explicitly addressed again
- Match the formality level of the group
- Never use more than 3 sentences in a group reply unless explicitly asked for detail

Why this matters: An AI that behaves well in groups is one you’ll actually keep connected. An AI that spams groups gets disconnected within a day. Get the rules right from the start.

Key Principle

The best AI in a group chat is one nobody complains about. That means speaking only when it adds genuine value, matching the group’s tone, and never dominating the conversation.

Step-by-Step Instructions

Step 1: Add group chat rules to AGENTS.md
Copy the example rules above and customize for your groups.

Step 2: Join a test group
Add your AI to a group chat with a friend or create a test group.

Step 3: Test the boundaries
Send messages that don’t mention your AI — it should stay quiet. Then mention it directly — it should respond. Send a message where its input wouldn’t add value — it should stay quiet.

Step 4: Iterate based on real usage
Over the first week, note every time your AI speaks when it shouldn’t or stays quiet when it should speak. Adjust AGENTS.md rules accordingly.

Common Mistakes

❌ AI responds to everything — Your group chat rules are too loose. Tighten the "only respond when" conditions.

❌ AI never responds in groups — Make sure the mention trigger is configured correctly. Test with @YourAIName.

❌ AI sends multiple messages — Add an explicit "one message per trigger" rule to AGENTS.md.

Category Assessment

You can message your AI from your phone (channel connected). Your AI operates on at least two channels with context-appropriate behavior. Group chat rules are configured and your AI knows when to speak and when to stay quiet.

Skills & Models Automation & Workflows