A virtual assistant on WhatsApp: what it does and what it should not do

A virtual assistant on WhatsApp should handle common questions and bookings for a specific business, say that it is a virtual assistant and hand over to a person when needed. It should not give clinical advice or answer everything.

What it does well

• Answers the most common questions outside opening hours: address, hours and services.
• Books, reschedules and confirms appointments or visits.
• Collects the contact details of people who ask.
• Hands the conversation to a team member when the topic is sensitive.

What it should not do

• Give diagnoses or clinical advice.
• Quote prices that depend on an assessment.
• Promise results.
• Invent information it does not have.
• Pretend to be a person.

WhatsApp's rules

Meta now prohibits general-purpose AI assistants, which answer on any topic, on the WhatsApp Business API. The rule applied to new users from 15 October 2025 and to everyone from 15 January 2026. Assistants with a specific function, such as customer support or bookings, remain allowed.

Transparency and data

If a customer asks whether they are talking to a person, the answer must be truthful. Because conversations contain personal data, the provider should sign a data processing agreement with the business.

Five questions to ask a provider

1. Does it use the official WhatsApp Business API?
2. What exactly is the assistant for, and what does it not answer?
3. How does it hand the conversation to a person?
4. Where is the data stored, and who are the subprocessors?
5. Is conversation data used to train AI models?

When it makes sense

• You receive many repeated messages.
• Messages arrive outside opening hours.
• Bookings are made through WhatsApp.
• The team is small and cannot reply straight away.

When it does not yet make sense

• You receive few messages a week.
• Almost every case is complex.
• There is no defined booking process yet.

How to start

1. List the ten most frequent questions and their answers.
2. Define opening hours and booking rules.
3. Decide when the conversation passes to a person.
4. Test with the team before opening to customers.
5. Review the conversations in the first week and adjust.