Where the manual work enters
A commerce assistant needs to separate webhook verification from real messages, keep enough context for a useful conversation, and retrieve information without mixing message handling with record updates.
Demonstration build
A webhook-based Messenger agent that handles verification and incoming messages, uses conversation memory, reads and updates product or customer records, and prepares a reply through an HTTP integration.
Conversational commerce
A commerce assistant needs to separate webhook verification from real messages, keep enough context for a useful conversation, and retrieve information without mixing message handling with record updates.
The webhook routes verification requests to an immediate response and filters message events into a Gemini-powered AI agent. The agent uses simple memory plus Google Sheets read and update tools before passing its response to the messaging API step.
The demonstration separates webhook setup traffic from customer messages and shows a context-aware response path with spreadsheet-backed lookup and logging capabilities.
Signal path
Accept webhook verification and message events.
Respond to verification or pass valid messages to the agent.
Use Gemini and conversation memory to prepare the response.
Read or update Sheets data and send the reply through the HTTP integration.
Tools in the system
Demonstrated capabilities
This project is a portfolio demonstration. It is not presented as a client deployment and makes no claim about production use, revenue, time saved, or business performance.
Next system
Lead qualification system →Discuss a process like this ↗