A WhatsApp Front Desk That Actually Takes Action

The problem

For many service businesses, WhatsApp becomes the front desk.

Customers ask the same questions repeatedly:

What time are you open? Do you have availability tomorrow? Can I move my appointment? Can I cancel my booking?

Individually, none of these conversations require much work.

At volume, they turn into a constant stream of interruptions.

The challenge becomes harder when customers naturally switch between English, Urdu, Roman Urdu, and Roman Hindi — and when an automated system needs to understand not just what someone said, but what they are trying to do.

A simple chatbot wasn't enough.

The system needed to distinguish between a question and an action, interact with the actual booking calendar, remember where a conversation was, and know when to stop automating.

What we built

We built an AI assistant directly into the business's WhatsApp workflow.

Every incoming message is interpreted by Claude to determine the customer's intent and the next action.

For common questions, the agent pulls information from a maintained FAQ knowledge base instead of inventing an answer.

For appointments, it connects directly to Google Calendar and can:

• Check availability
• Create bookings
• Reschedule existing appointments
• Cancel appointments

The agent also maintains conversation state, allowing a booking conversation to continue naturally across multiple messages rather than treating every message as a new request.

And because customer conversations don't always fit neatly into predefined workflows, we designed human handoff as part of the system from the beginning.

If the AI is uncertain — or the customer simply asks for a person — the conversation is escalated to staff through WhatsApp.

Unanswered handoffs are checked again on a schedule so a conversation doesn't disappear just because automation stopped.

The outcome

• Routine questions no longer require someone to manually reply every time
• Customers can book, reschedule, and cancel directly through the same WhatsApp conversation
• Google Calendar remains the source of truth rather than creating a separate booking system staff need to maintain
• English, Urdu, Roman Urdu, and Roman Hindi conversations can move through the same workflow
• Conversation context persists across multi-message interactions
• Staff are brought in when judgment is required instead of monitoring every conversation
• Human handoffs remain visible until somebody deals with them

The result wasn't a chatbot sitting beside the business.

It was an assistant connected to the systems the business already used.

Answering a message is easy. The useful automation starts when the system can actually do what the customer is asking for.