Hi everyone! I’m building a workflow with Make AI Agents to handle customer orders via WhatsApp and log them into Google Sheets. It’s working well so far, but I need help with these 3 things:
How can I connect an external database (like Google Sheets or Airtable) so the agent can use it to reference product info, pricing, or customer names in its responses?
How can I delay the agent’s response by a few seconds, in case the customer is still writing? Some customers send their orders across multiple messages.
How can the agent update an existing order (same customer + same date) in Google Sheets, instead of adding a new row when the client sends a correction later?
I’m using WhatsApp Cloud API, Google Sheets, and modular tools (scenarios) inside Make.
Any advice or ideas would be super appreciated. Thanks!
For the second question, I think the trick is to try and guide the customer NOT to send multiple messages but to wait for each one to be processed. It would be tricky to get the Agent to recognize multiple messages as being part of the same instruction.
Frequently, chat agents add an animated “…” emoji to the customer’s first message to show that they’re working on it. That technique is a useful indicator to the customer to wait for a response.
For the third item, ideally there needs to be some reference that the Agent can use to be sure that the customer is referring to the same order.
Alternatively, you could prompt the agent to check the Orders sheet to see if there is a potential existing order (using appropriate search terms and a search Tool) and then to ask the user if they want to amend that order.
We’ll be adding more resources soon on examples like this. Right now, you might find some good guidance simply by asking ChatGPT or Gemini for general advice on how to implement those features in agents.
Great, thank you very much. Your response was really valuable. I have another question: have you considered creating a GPT with all the information about Make, which updates automatically with your latest developments? I think it would be really useful for the community.
