How to Train an AI WhatsApp Assistant on Your Business FAQs

“What time do you close?” is easy to answer when your opening hours are clear. “Can I collect my order after closing?” needs more context.

An AI WhatsApp assistant works better when your business content makes that distinction explicit. The useful preparation is organizing accurate answers, identifying exceptions and deciding when a person should take over.

Here is a practical way to prepare your knowledge base before enabling customer replies.

Start with questions your team already answers

Review common enquiries and group them by purpose:

Question type	Useful source content	Human attention may be needed when
Opening hours	Hours, holidays and location details	The customer asks for an exception
Services	Current scope and exclusions	The request needs a custom assessment
Pricing	Current published prices and conditions	A discount or negotiated quote is requested
Booking process	How to request or change a booking	Availability must be checked
After-sales help	The normal support process	A complaint or disputed outcome needs review

Use this table as a planning tool. It does not mean the assistant has live access to appointments, order status or private customer records.

Write one reliable answer for each topic

A useful FAQ entry contains the question, the answer and the boundary.

For example:

Question: What are your opening hours?
Answer: Our office is open Monday to Friday, 10 am to 6 pm.
Exception: Holiday hours may differ. Ask the team to confirm visits on public holidays.
Next action: Customers requesting an after-hours visit should speak to a staff member.

This is an illustrative content format. Replace every detail with your actual business information.

Avoid contradictory sources. If a PDF has last year's prices while the website has this year's, update or remove the old source before asking the assistant to use both.

Add knowledge sources in PingDaily

The AI WhatsApp Assistant supports website pages, supported file uploads and pasted text as knowledge sources.

Prepare a small, focused first set:

Business name, location and contact details.
Opening hours and normal response expectations.
Service descriptions and exclusions.
Current published prices, where appropriate.
Common questions and the correct escalation route.

Wait for the sources to finish indexing before testing. Keep internal passwords, confidential staff notes and unrelated client records out of this customer-facing knowledge base.

The assistant is currently in beta and is priced separately. It requires an active BYO Pro plan and your own connected WhatsApp Business number.

Set the response boundaries

PingDaily provides settings for auto-replies, quiet hours, a human-first delay, fallback behavior and reply limits.

Choose these settings around your staffing pattern. If staff normally answer during opening hours, a human-first delay gives them time to respond. If an answer needs personal judgment, the fallback should point the customer to the team.

Meta's WhatsApp Business Messaging Policy requires clear escalation paths for automated replies. Make the route to a person easy to understand.

Do not promise that a handoff means someone will reply immediately unless your team can actually provide that response.

Test answers before enabling a wider rollout

Use an internal contact and a short test set. Include both questions the assistant should answer and questions it should hand over.

Test prompt	What to look for
“What time do you open?”	The answer matches the current source
“Are you open on the next public holiday?”	The assistant handles the stated exception
“Can you give me a special discount?”	It does not invent an offer
“Please confirm my booking for tomorrow”	It does not invent a booking confirmation
“I need to speak to a person”	The escalation route is clear
A question absent from your sources	The fallback is useful and honest

Save the questions that produced confusing answers. Fix the source or configuration, re-index when appropriate and test again.

AI responses can need correction. A successful answer to one question is not enough to validate a whole knowledge base.

Make handoff useful for the team

The person taking over should be able to see what the customer asked and what the assistant already said.

Use the shared WhatsApp inbox to continue the conversation. Where Teams is enabled, organize staff access and responsibility for the handoff queue.

Review unresolved conversations during the working day. If customers repeatedly ask a question that is missing from the source material, add a reviewed answer instead of treating each case as a separate exception.

Keep the knowledge base current

Assign someone to review source content whenever your prices, opening hours, services or support process changes.

For a regular quality check, sample answers and look for:

Information that no longer matches the business.
Replies that should have gone to a person.
Unclear fallback messages.
Repeated unanswered questions.
Sources that have failed to index.

Track answer quality and completed handoffs alongside reply volume. A higher number of automatic replies is not useful if staff must correct them later.

Frequently asked questions

Will the assistant know my live appointment availability?

Do not assume that it does. A knowledge source describing your booking process is different from an integration with a live booking system.

Can I pause automatic replies?

Yes. Review the auto-reply controls and pause the assistant when source information or behavior needs correction.

What should I upload first?

Start with accurate opening hours, location, service descriptions and a short set of reviewed FAQs. Expand after testing the first answers.

Explore the AI WhatsApp Assistant, or talk to PingDaily about your support workflow.