We've all been there—messages pinging at 2 AM, or getting distracted by chats when you're in deep work mode. I decided to solve this by building a Smart WhatsApp Agent that manages my availability for me.
This isn't just a simple auto-reply bot; it's a state-aware system powered by n8n and PostgreSQL.
How It Works
The system is built on a few core concepts:
1. State Management (PostgreSQL)
A stateless bot is a dumb bot. My agent needs to know:
- Am I asleep?
- Am I in 'Deep Work' mode?
- Who is allowed to bypass these modes?
I use a simple PostgreSQL database to store this state. One table tracks my current status (status_id, status_text, updated_at), and another tracks "VIP" contacts who can reach me anytime.
2. The Workflow (n8n)
n8n orchestrates everything. The webhook listens for incoming WhatsApp messages via the WhatsApp Business API.
Logic Flow:
- Incoming Message: Webhook triggers.
- Check Status: Query PostgreSQL to get my current state.
- Decision Tree:
- If
Status == Available: Let the notification through. - If
Status == Sleeping: Send an auto-reply ("I'm sleeping, will reply tomorrow") and mute the notification. - If
Status == Busy: Check if sender is VIP. If yes, notify. If no, auto-reply ("In deep work, catch you later").
- If
3. Control Interface
I didn't want to write SQL queries to change my status. I built a simple lightweight frontend (or you could use a Telegram bot) that sends a command to n8n to update my status row in Postgres.
Why n8n?
n8n is the perfect glue for this. It handles the API webhooks, the database logic, and the conditional branching visually. It turns a complex logic problem into a flowchart that I can debug and modify in seconds.
Now, I sleep peacefully knowing my agent is guarding my notifications.