Free guide
How to Write the Instructions for Your WhatsApp AI Agent So It Gives Correct Answers and Knows When to Ask a Human
A configuration guide for sales and admissions teams who have switched on a WhatsApp AI agent but are not sure what to put in it, how to test it, or what to fix when it gives a wrong answer.
- The 5 things every WhatsApp AI agent needs in its knowledge base before it handles a real customer enquiry
- How to write product or service descriptions for an AI agent so the answers are accurate, specific, and on-brand — not generic
- Escalation trigger writing: how to phrase the conditions that send a conversation to a human agent so the AI knows when the line has been crossed
Get your free copy
Instant download, and we'll email you the link so you never lose it.
No spam, ever. Your details stay with us.
What's inside
- The 5 things every WhatsApp AI agent needs in its knowledge base before it handles a real customer enquiry
- How to write product or service descriptions for an AI agent so the answers are accurate, specific, and on-brand — not generic
- Escalation trigger writing: how to phrase the conditions that send a conversation to a human agent so the AI knows when the line has been crossed
- The 3 most common AI agent mistakes after launch and how to fix each one (wrong pricing quoted, failing to escalate complaints, over-qualifying cold enquiries)
- A pre-launch testing checklist: 12 enquiry types to send your AI agent before going live, and what a correct answer looks like for each
- How to maintain the knowledge base over time so the agent stays accurate as your products, pricing, or processes change
- What to do when a customer complains about the AI's answer: how to review the conversation, identify the gap, and update the instructions so it doesn't happen again
Who this is for
Sales and admissions teams of 3 or more people at Kuala Lumpur businesses who have deployed or are about to deploy an AI agent on their WhatsApp inbox but have not yet built an effective knowledge base or tested the agent's responses against real enquiry types.
Made by the team behind Yalla — the chat-native CRM where the AI does the work, not just the talking.