How to Build a WhatsApp AI Chatbot for Your Business
By Steven Van ·
Quickchat's guide builds a WhatsApp support agent with knowledge answers, order lookup and human handoff, using a fictional bike shop.
A WhatsApp support chatbot built on Quickchat can answer from a shop's knowledge, check an order's status and hand a conversation to a person. Quickchat published a walkthrough that builds one for a fictional shop, Harbor Lane Bikes. Customers can use text, voice notes or photos, and the conversation stays in WhatsApp.
How the example works
In the example, a customer asks about Saturday opening hours, checks order HL-1042, asks about servicing by voice, then sends a damage photo. The Agent answers the routine questions and transfers the damage case so a person can arrange an inspection.
- Policies and opening hours come from the Knowledge Base.
- Order status comes from an API Action.
- The human reply comes from the Quickchat AI Inbox.
The order lookup is optional. The knowledge and staffed handoff work on their own, and the real order system can be connected after testing the example. WhatsApp supplies the sender's phone number, which the order recipe uses directly in its request while asking the customer for their order number.
What you need
The guide lists a Quickchat AI account, a dedicated business phone number that can receive SMS or a voice call, and a second WhatsApp account on a different number for sending test messages. The number should not be active in WhatsApp Messenger or the WhatsApp Business app. Admin access to the Meta Business portfolio is needed, along with the business name, website, category and description.
The Free plan supports personal, non-commercial learning, and a business deployment needs a commercial plan. Preview and Simulation consume AI credits. The optional order extension also needs Python 3.9 or later and cloudflared.
Quickchat says its current connection flow does not provide a verified path for keeping an existing WhatsApp Business app number in the mobile app, and advises contacting support before changing an active account.
Setup steps covered
A downloadable kit holds the knowledge files, Main Prompt, order service and test inputs. They are editable files rather than a one-click Agent import.
- Step 1: create a separate Agent with Set up manually, then add three Knowledge Base articles: store information, workshop and bike fitting, and orders and returns. Individual order records stay out of them.
- Step 2: paste the Main Prompt into Identity and set Reply Length to Short. The prompt covers short phone-suited replies, order lookup, handoff and unsupported facts.
- Step 3: connect WhatsApp, followed by handoff and testing.
Testing in AI Preview includes checking for 10:00 to 16:00 Saturday hours and a EUR 129 full service with a two working day turnaround. International shipping is left out of the knowledge on purpose, so a question about shipping to Germany tests how the Agent handles something it does not know.