Channel
Talk to your AI agent from WhatsApp
WhatsApp uses the Meta Cloud API. An inbound message goes to the same runtime, through the same tool registry, policy engine and approval gate as web chat. The agent only ever replies to someone who has just written to it.
What you get
WhatsApp capabilities, under your rules
- ✓ The same runtime, tool registry, policy engine and approval gate as web chat
- ✓ A destructive request stops for approval exactly as it would in web chat
- ✓ The allowed-senders list starts empty, meaning nobody
- ✓ Senders are identified by the platform’s own id, never a display name
- ✓ A channel run carries nobody’s permissions, so anything that needs one is refused
- ✓ A repeated delivery of the same message is recognised and dropped
What you need to connect it
- A Meta app with WhatsApp added, and a business phone number
- A permanent System User access token, the app secret, and the phone number ID
- The Callback URL and Verify token Caterfli gives you, entered in your Meta app
Human approval
Destructive work waits for a person
approval
waiting- Tool
- delete_customers
- Operation
- DELETE
- Runs against
- Production PostgreSQL
- Agent
- Support Agent
Exactly these values
{
"filters": {
"id": 100
}
}
Your decision applies to these values only. If anything about the call changes, it stops for approval again.
Asked for from WhatsApp. Connecting a channel does not create a second agent with different rules.
Connect WhatsApp to your first agent
Connect one system and read the agent Caterfli writes from it. There is a demo connection too, if you would rather look before you wire anything up.