Running an AI agent on WhatsApp Business
What running an agent on WhatsApp actually requires: the account, the 24-hour window, templates, and human handoff.
· 6 min read
The 24-hour window
Meta allows free-form replies for twenty-four hours after the customer's last message. After that, free text is rejected and an approved template is required. This reshapes the agent's design: deferred follow-ups must be planned as templates in advance, not composed in the moment.
The number cannot stay on the app
A number registered on the Cloud API cannot simultaneously be used in the WhatsApp or WhatsApp Business phone app. Many teams discover this after connecting their main number and losing phone access to it.
Human handoff on WhatsApp
On WhatsApp the customer sees no difference between agent and employee; both are bubbles in the same thread. That is an advantage when the handoff is seamless and a frustration when the customer has to re-explain. The full transcript has to reach the human.
Frequently asked
- Do I need business verification?
- Not to test, but the limits are very low. For production yes: verification raises the daily messaging limit and allows a verified display name.
- Do agent messages cost money?
- Meta bills conversations rather than individual messages, and the rate varies by country and by who initiated. Customer-initiated conversations are typically cheaper.
Read next
How to build an Arabic AI agent for customer service
A practical guide to an agent that understands Modern Standard Arabic and dialects, answers only from your knowledge, and hands off to a human when needed.
Designing the handoff from agent to human
When an agent should hand off, what must travel with it, and why a zero handoff rate is a bad sign rather than a good one.
A security checklist before launching an AI agent
What to verify before your agent talks to a real customer: prompt injection, tenant isolation, data leakage, and audit.