Automating Price Quotes from Zalo to Google Sheets

A customer messages your Zalo OA asking for a price at 11 PM, but your sales staff has to wait until the next morning to open their computer and enter the data into a spreadsheet. That time gap is more than enough for a competitor to swoop in and steal your potential lead.

Why manual quoting processes are slowing down your business

Traditional methods are often very cumbersome. A customer messages on Zalo. A salesperson reads the message, picks out the quantity and product name, and manually pastes them into a Google Sheets spreadsheet. Once the spreadsheet calculates the final figure, the staff then copies that price and pastes it back into the Zalo chat box.

This process repeats dozens of times a day. Not only is it time-consuming, but it’s also extremely prone to errors when employees get tired. Current Artificial Intelligence can completely replace human labor in this workflow thanks to a technique called Tool Calling.

How Tool Calling works in practice

AI isn’t just for chatting and answering questions. When granted permission, it can receive commands and decide for itself when it needs to call an external tool to process data.

You can read more in the article Teaching AI Agents to Read and Write Google Sheets via Tool Calling to understand the technical mechanism. Basically, the AI acts as an orchestrator. When a customer messages “Quote me for 50 white t-shirts,” the AI automatically understands that it needs to extract the number “50” and the phrase “white t-shirts.” It then calls an API to push this information directly into the corresponding columns in your Google Sheets.

Lessons from onmee.vn’s own automation system

Before applying automation to price quotes, you need to understand the operational risks. I personally run the content automation system for onmee.vn. It uses GitHub Actions on a fixed schedule, calls the Gemini API to generate articles, runs its own quality control validators, translates them into English, commits the changes, and lets Firebase Hosting deploy them.

This entire chain runs without anyone having to press a button manually. This is a real AI automation project I built for myself. From April 9, 2026, to early July 2026, the system automatically generated over 300 posts across 89 runs.

The price of lacking quality control

Everything sounded smooth until trouble hit. Around July 3, 2026, I was rejected by Google AdSense for “Low value content.”

The root cause was that the system published too quickly at a rate of 4 posts per day, while the quality control filters weren’t tight enough. I immediately had to reduce the posting frequency and add “hard” validators to fix it. This incident taught me a big lesson about letting AI run wild without supervision.

Applying the lesson to Zalo price quotes

The lesson above applies directly to your quoting workflow. Integrating Claude AI into Zalo OA to close deals provides lightning-fast response times. However, if the AI automatically pulls information from Zalo and pushes it into Sheets to get a price without a verification step, it could misquote the customer significantly. You must set clear boundaries for the AI using validator functions.

Steps to set up an automated quoting flow

To build this system, you should follow these basic steps:

1. Connect your Zalo OA account to an intermediary automation platform like n8n.
2. Configure an AI model that supports Tool Calling, such as GPT-4o or Claude 3.5.
3. Set up a function that requires the AI to extract exactly three variables: customer name, product code, and quantity.
4. Create an API connection flow that pushes these three variables into Google Sheets and retrieves the result.
5. Set a mandatory validator rule: the system only sends the quote message if Google Sheets returns a valid number greater than 0.

Frequently Asked Questions

Can the AI write to the wrong column in Sheets?

It absolutely can if you don’t define things clearly. You must precisely describe the data format for each field when setting up Tool Calling.

How long does it take for this system to run stably?

It usually takes about 1 - 2 weeks to fine-tune the prompts. Real customers message in very messy ways, with typos or abbreviations, so the AI needs time to be instructed on how to handle these exceptions.

Do I need to buy any extra software?

You need to pay for the AI model’s API, have a verified Zalo OA business account, and maintain a server for the automation platform.

Conclusion

Automation is not a magic spell that you write once and it runs forever. My painful experience with Google AdSense is the clearest proof that machines work very fast, but mistakes are multiplied at the same speed. Using Tool Calling to automate quotes from Zalo to Google Sheets will certainly free your staff from hours of manual labor. However, start small, monitor the output data closely in the first month, and always build a safety checkpoint before the final figure is sent to the customer.