AI Studio: Hiding Sensitive Data in AI Automation
Some data in MYOB Acumatica—such as account numbers and prices—shouldn’t leave your system. If you’ve avoided using certain prompts with AI tools because of security concerns, you can put those worries aside.
Starting in MYOB Acumatica 2026.1.1, you can use data masking in AI Studio to keep sensitive data secure. Here’s how it works:
- You mask any sensitive details in your requests to a large language model (LLM) before sending them to the cloud.
- The real data never leaves MYOB Acumatica—the LLM receives only masked content.
- When the LLM returns the processed data to MYOB Acumatica, the system replaces the masked content with the actual data.
Read on to learn how AI Studio makes sure that the original data stays within MYOB Acumatica.
Masking Fields in the Instructions
Suppose that you’ve created a prompt on the LLM Prompts (ML202000) form, and this prompt generates closure notes for a support case. It includes the case’s owner and business account (shown below), and you don’t want this data to go anywhere outside of MYOB Acumatica.

To mask these fields, click (shown below) and select the needed field in the dialog box that opens.

Notice that the system adds the MASKED prefix to the masked fields (see below).

When you select an example record, the system displays the masked data on the Prompt tab (shown below) instead of the real data. This is the information that the system will send to the LLM.

When you test the prompt and the LLM sends the processed data back to MYOB Acumatica, the system replaces the masked content with the actual data (see below).

Masking Groups of Fields
You can mask lists of records—such as a case’s activities or an invoice’s detail
lines—by using the foreach loop (Item 1 below) to iterate through the
records. The system masks each field and adds a numerical suffix (starting with 0)
to
each record, as Items 2 and 3 show.

Similarly, you can add masked fields to a table. The example below shows a table that lists case activities.

For details on using the foreach loop and listing details in a table, see
Email Templates.
Masking Data in Output Format Examples
An output format example is a part of the prompt—so it’s also sent to the LLM. When the system generates the example, it uses real data from the database. If it detects a field in the example that has a masked version in the Instructions with Input Data section of the Instructions tab, the system does the following:
- Replaces this field with its masked version
- Displays a warning about this replacement (shown below)
You can replace the data in the example with any arbitrary values.

Tracking Your Data
Want to track your data throughout the masking process? Go to the Trace page to see how:
- The system has masked the specified fields (Item 1 below).
- The system has recorded the real values of those fields (Item 2).Important: These real values stay within the system and are never shared with the LLM.
- The LLM returns the data in which the fields are masked (Item 3).

Tips for Success
- Create a prompt as you usually do.
- Identify what fields have sensitive data.
- Use the new Masked Data Field command to insert each applicable field in a masked form.
- Mask only specific fields or all fields.
Masking Limitation
You can’t mask fields that are part of a larger text field (such as case closure notes or an email reply). Even if you also add masked versions of those fields somewhere else in the instructions, the system won’t mask them in this text field.
Learn More
For details on configuring LLM prompts, see Integration with LLM Providers: General Information.
