<< Click to Display Table of Contents >> Navigation:  Low-code Process Automation > Studio Cloud - Authoring environment > Bizagi Studio > AI Hub > AI Workers

Overview

AI Workers automatically populate editable fields in a form by analyzing field descriptions, applying business rules, and leveraging data already entered in previous editable fields. Unlike traditional AI agents, AI Workers do not require programming, predefined specifications, or prompts.

 

This feature significantly improves efficiency by reducing manual data entry and ensuring forms are pre-filled with accurate, contextually relevant information.

 

How AI Workers operate

When an AI Worker is triggered in your process, it follows these steps:

•Form Analysis – Examines the structure and content of the form.

•XPath Extraction & Context Creation – Identifies key data points and their relationships.

•Gen AI Invocation (GPT-4o) – Uses advanced AI to generate accurate field entries.

•Form Controls Filling – Populates fields based on extracted insights.

•XPath Persistence (Optional) – Stores data depending on the selected mode.

 

AI Workers offer you Supervision modes to control your AI Workers. The Supervised mode empowers teams to focus on higher-value tasks while maintaining oversight . For those confident in AI accuracy, the Autonomous mode, further streamlining workflows.

•Supervised: Each pre-filled field will display the value chosen by the AI along with an explanation of why that value was selected. This allows you to review and verify the AI's output, ensuring accuracy and accountability.

•Autonomous: If you trust the AI's capabilities, you can choose to let AI Workers operate autonomously, allowing tasks to be fully completed automatically.

•Autonomy Rule: this mode lets you define an Expression rule. This rule determines when the AI Worker can operate using the Autonomous mode. If the condition isn’t met, it switches to Supervised mode.

 

Implement AI Workers

To showcase how to implement AI Workers this article contains an example.

In this example, an AI Worker will be implemented to the Invoice Approval Request process. When an employee uploads an invoice file, the AI Worker will automatically extract the necessary details and fill in the required fields.

 

In order for your AI Worker to have the best performance possible you need to do the following changes to your project:

After modeling the process in which you will implement an AI Worker, add a detailed description to the activities of your process. This description need to be clear and easy to understand, this allows the AI Worker understand the process and context.

To add a description, right-click the Activity where you want to add AI Workers and select Properties and fill in the Description text box, located in the Basic tab.

 

 

When creating the Data Model, the Display name of every attribute needs to be clear and easy to understand, as the AI Worker uses the Display name as context to understand the process.

 

 

Also, add a description to the attributes of your Data Model. To do this, follow the next steps:

•Select an attribute in your Data Model.

•Click Show advanced options

•In the Help text field, add a detailed description of the attribute.

 

 

After doing this, you can start to implement AI Worker's to your process.

 

Access AI Workers configuration

In the Model Process step, right-click the task where you want to add AI Workers and select Properties. For this example, we are adding AI Workers to the Complement invoice information task.

 

 

Click the Runtime tab.

 

 

In this tab you can configure the AI Workers for the selected Activity, the following sections explain how to configure an AI Worker using each Supervision mode.

 

Supervised mode

In the Runtime tab, activate the AI Worker checkbox.

 

 

Then, select the Supervised option from Supervision Mode.

 

 

An icon will appear in the top left corner of the Activity in which you configured the AI Worker's.

 

 

This Supervision mode allows the AI Worker's to intervene the fields of the Form of the selected Activity and set it ready for user review.

 

 

Whenever a field is filled by an AI Worker a special icon will appear next to the name of the field. Hover the image to get an explanation of what the AI Worker did in the field, and why it did it.

 

 

AI Workers - Autonomous mode

In the Runtime tab, activate the AI Worker checkbox.

 

 

Then, select the Autonomous option from Supervision Mode.

 

 

After selecting the Autonomous option as Supervision Mode, a confirmation window will appear. Click the Yes button to confirm your decision and finish the configuration.

 

 

An icon will appear in the top left corner of the Activity in which you configured the AI Worker's.

 

 

With the Autonomous Supervision Mode, the AI Worker's suggestions will be applied automatically, an the task where the AI Worker is configured will continue without direct human oversight.

 

AI Workers - Autonomy Rule mode

In the Runtime tab, activate the AI Worker checkbox.

 

 

Then, select the Autonomy Rule option from Supervision Mode.

 

 

After selecting the Autonomy Rule option as Supervision Mode, a confirmation window will appear. Click the Yes button to confirm your decision.

 

 

After that, a pop-up window will appear, in thias window you need to define an Expression that will work as the AI Worker Autonomous Rule. This rule determines when the AI Worker can operate autonomously. If the condition isn’t met, it switches to Supervised mode.

To define this rule, click the New button.

 

Define the rule that determines whether the AI Worker operates in Autonomous or Supervised mode and click the Ok button. In this example, we will check if the invoice attached by the employee is considered critical for the business.

 

 

If you want to modify the rule, click the button located to the right of the AI Worker Autonomy Rule label.

 

 

An icon will appear in the top left corner of the Activity in which you configured the AI Worker's.

 

 

With the Autonomy Rule Supervision Mode, determines when the AI Worker operates in Autonomous mode when the condition is met. If it isn’t, it switches to Supervised mode.

 

Example using AI Workers

After implementing AI Workers with Supervised - Supervision mode in the Invoice Approval Request, Here are the results when the process is executed in the Work Portal.

In the first task an employee attaches a file containing an invoice.

 

 

 

When you click the Next button, in the next task the AI Worker will attempt to fill the editable fields, as the process reaches the task Complement Invoice Information.

 

 

Whenever a field is filled by an AI Worker a special icon will appear next to the name of the field.

Hover the image to get an explanation of what the AI Worker did in the field, and why it did it.

 

 

Important Considerations When Using AI Workers

AI Workers analyze the content of a form to pre-fill editable fields with accurate, contextually relevant information. However, there are certain limitations and behaviors you should be aware of:

•Form Content Scope: AI Workers only analyze visible information in the form, including fields, field descriptions, and data across all tabs.

•Nested Forms: Information from nested forms that are displayed when the form is opened is considered.

•Linked Forms: Data within form links that require user interaction (e.g., one or more clicks away) is not considered.

•Tables: AI Workers do not interpret or extract information from tables within the Edit Form or Details Form.

•Single Execution: AI Workers run once when the task is entered and do not re-execute during the same activity instance.

•Data Model Limitation: Information that exists in the data model but is not present on the form will not be considered.

•Action Events: Rules or logic executed in Action Events such as On Enter, On Save, or On Exit are ignored by AI Workers.

•Other UI Components: Data from the Overview section and Case Folder is not taken into account.

•Render Rules: Visibility rules, min/max constraints, and other rendering logic are not evaluated.

•Hidden Fields: Fields that are hidden or set to not visible in the form are excluded from AI Worker analysis, even if they are technically present.

 

 

 

 

 

Controls considerations

Does it read? This column means that the AI Worker is able to read from the field for its information to be considered in the AI processing.

Does it write? This column means that the AI Worker is able to write a value when the field is editable in the form.

 

 

Controls considerations when used in tables

Additional to the considerations for controls mentioned above, the following are considerations when used in tables.

 

Does it read? This column means that the AI Worker is able to read from the field for its information to be considered in the AI processing.

Does it write? This column means that the AI Worker is able to write a value when the field is editable in the form.

 

Last Updated 6/3/2025 4:47:33 PM