AI Reconciles 1,000 Zalo Tracking Codes in 5 Minutes

At the end of the month, your accountant has to open every Zalo chat, copy tracking codes, paste them into an Excel file, and cross-check them with the shipping partner’s reconciliation file. This process consumes several working days and is riddled with the risk of human error.

The Pain of Manual Reconciliation

Every day, sales staff send numerous tracking codes to customers via Zalo. By the end of the week or month, you receive an Excel file from the courier with thousands of data rows.

Cross-referencing tracking codes, Cash on Delivery (COD) amounts, and delivery statuses by eye is a nightmare. As a business scales and runs large sales campaigns, the surge in order volume will crash this manual operation. You need a fully automated workflow to ensure cash flow isn’t lost.

How the AI-Powered Automated Workflow Works

The system doesn’t rely on humans to sit and read every message. Instead, it connects directly to the business’s Zalo OA (Official Account) via an API protocol.

Every time a tracking code is generated or sent, the AI automatically extracts that string of characters. You can refer to Integrating Claude AI into Zalo OA for Order Closing to understand how AI reads and analyzes message contexts. The AI accurately identifies which part is the shipping code, the amount, and the customer information.

Direct Connection to Spreadsheet Systems

After retrieving the tracking code, the AI needs a place to store and reconcile it. This is where “tool calling” comes into play.

The AI system is granted secure access to your Google Sheets file. Teaching AI Agents to Read and Write Google Sheets via Tool Calling allows the system to automatically fill tracking codes into the correct columns as soon as a Zalo message is sent. When the reconciliation file from the post office is uploaded, the AI runs a command to scan all data. It compares columns and highlights mismatched rows in red within seconds.

Handling Exceptions

Reconciliation isn’t just about matching identical codes. The biggest issues often lie in mismatched COD amounts or returned orders that haven’t been recorded.

When the AI detects a tracking code where the COD amount on Zalo differs from the post office’s file, it automatically sends an alert. The business owner or chief accountant will receive a notification specifying exactly which row is erroneous. You only need to focus on handling the few problematic orders instead of searching through 1,000 normal ones.

The most valuable aspect of this system isn’t just speed. The ability to completely eliminate errors caused by manual data entry is what helps a business maintain its actual profits.

Steps to Build an Automated Reconciliation System

To implement this workflow for your business, you need to set up a basic automation sequence as follows:

1. Connect the Zalo OA API to receive message data in real-time.
2. Use a large language model (such as Claude 3.5 or GPT-4o) to extract tracking codes from unstructured text.
3. Configure the system to push filtered data to Google Sheets.
4. Set up an automated script to match data whenever a new reconciliation file is uploaded.

Frequently Asked Questions

Does this system cause Zalo messages to be lost?

No. The system only reads data through the official API provided by Zalo. It does not interfere with the right to send/receive messages or delete any chat history of the employees.

What is the monthly operating cost?

The cost mainly comes from the API call fees of the AI model when processing text. For a volume of 1,000 orders, this cost is very low, usually only a few dollars per month for the processing infrastructure.

Does AI mistake phone numbers for tracking codes?

This rarely happens if you set up precise instructional prompts. Tracking codes from shipping units usually have specific alphanumeric structures, which AI can easily distinguish from phone number formats.

Conclusion

Using humans for repetitive tasks like looking up codes and comparing spreadsheets is a major waste of resources. Current technology is stable enough to take over this tedious work with absolute accuracy. When you delegate reconciliation to machines, your staff will have more time to care for customers and focus on revenue-generating activities.