AI PROMPT – Business use cases

Applies to version: 2025 R2 and above; author: Adrian Baszak, Lily Adamowicz

Related documentation

A detailed description of the functionalities mentioned herein and their configuration can be found in the following sections of the WEBCON BPS Help:

• Business rules - Functions

Introduction

WEBCON BPS Designer Studio provides AI functionalities that can automate many activities and significantly streamline working with forms. Among them, one of the most versatile is AI PROMPT—a function that enables sending precise queries to the AI model. The text responses obtained in this way can be immediately used within workflows, enriching them with intelligently generated content.

The following sections analyze three business cases in which the use of AI PROMPT brings real value: during invoice registration, in the article review process, and in handling procurement. These examples illustrate the flexibility of this function in practical applications.

Invoice process

The "Invoice" process in WEBCON BPS is designed to automate the extraction of invoice data and eliminate the need for manually retyping information from attachments. A user attaches an invoice to the form, and the system—using predefined rules—automatically identifies and processes the data it contains.

A key element of this solution is the use of the AI PROMPT function, which enables the AI model to read the contents of the attachment and extract the relevant information. These values can then be assigned to the corresponding fields in the form. The system recognizes not only the header data of the invoice (number, date, vendor details, amount) but also individual line items, like products and services listed in the invoice table. This approach not only speeds up invoice registration but also eliminates the risk of manual data entry errors and opens doors for further automation, such as integration with accounting or procurement processes.

A graphical representation of the process is shown in the screenshot below:

After attaching the invoice to the form, the user proceeds along the "Extract invoice data" path, where automation consisting of several actions is triggered:

A key component is the "Extract invoice data" action (action type: Change value of single field). Its configuration uses the STRING REPLACE function, which replaces a specified phrase with another. In this case, the function contains AI PROMPT, where the GET ATTACHMENTS function specifies the invoice to be analyzed. The actual query content is placed in the TEXT field. This setup allows the AI model to receive the invoice content, analyze it according to the instructions, and return the result as a JSON object.

The resulting JSON is then stored in the technical field "_invoice data", serving as a data source for further processing.

The query inserted into the AI PROMPT function in the TEXT field is shown in the screenshot below:

At this stage, the data is not yet visible in the form. Mapping occurs in the subsequent actions: "Set line items" and "Set header fields", which take the JSON object from the technical field and use it to populate the corresponding header fields and the document’s Item list. Since these actions are placed on the path, the mapping occurs when the path is transitioned.

The screenshot below shows the form automatically filled with invoice data—all key information, such as the invoice number, tax IDs, dates, and list items, has been extracted from the attachment and correctly assigned to the appropriate fields:

Marketing content process

The "Marketing content" review process is designed to accelerate and automate the verification of marketing materials, while ensuring consistency and high quality of published content. This enables organizations to bring articles to market faster, minimize errors, and reduce the manual workload of marketing teams.

Within the process, a dedicated workflow guides users through the successive stages of creating and reviewing marketing articles:

When the workflow is started, the user sees the form shown in the screenshot below. They can either manually enter the article content or use the AI-powered text generation feature by clicking the "Generate content" button:

The article content is generated based on the values entered in the "Style" and "Topic" form fields. For this purpose, the AI PROMPT function is used, where a query is formulated to generate content according to the specified topic and style.

Below is the query entered in the TEXT field, which was used in the AI PROMPT rule to generate the article content:

The resulting output is, for example, as follows:

The AI PROMPT rule is also used to generate article reviews in the "Review" step. In this step, the "Set review" action (type: Change value of single field) runs on a Timer, automatically initiating the creation of the review by the AI.

In this case, the AI PROMPT function generates the review text based on the content of the "Content" form field. The review is then directly assigned to the "Review" field in the form.

Below is the query entered in the TEXT field, which was used to generate the article review:

Below is a sample review:

After the review is completed by the AI PROMPT, the form transitions along the "Complete review" path, where a task is created and assigned to the article’s author. The review result, stored in the "Review" form field, can then be verified by the author. The author subsequently decides whether to complete the review process or send the article for further analysis. This ensures the process remains fast and automated while allowing the author full control over the final version of the published content.

Purchase requests process

The "Purchase requests" process is designed to accelerate and automate the approval of equipment requests and other employee requisitions, while ensuring transparency and control at every stage of the workflow. By using the AI PROMPT function, the system can assist in request registration by automatically filling in form fields based on the description provided by the user.

The diagram of the purchase requests process is shown in the screenshot below:

When the process is initiated, the user sees the form shown in the screenshot below.

For the "Description" form field, a form rule was defined that uses the AI PROMPT function to automatically fill in the remaining form fields based on the content of the provided description.

In the TEXT argument of the rule, a prompt was created for the AI to initiate the completion of individual form fields:

An example of a form automatically completed by the system is shown below:

After the request is registered, the form moves to the "Manager approval" step, where the manager can either approve or reject the request. At subsequent stages, based on defined rules and conditions, the system routes the form for management approval or directly to the final step. This ensures that decisions are made quickly, minimizes the risk of errors, and automatically notifies the requesting person of the status of their submission at the end.

The AI PROMPT function plays a key role in this process, enabling the system to intelligently analyze the request description and support both the person submitting the request and the decision-makers throughout the workflow.

Summary

By leveraging the AI PROMPT function, all three processes in WEBCON BPS—"Invoices", "Marketing content", and "Purchase requests"—gain a new level of automation. In the "Invoices" process, AI accelerates the recognition and mapping of data from attachments; in "Marketing content", it supports the creation and review of articles; and in "Purchase requests", it intelligently analyzes submissions and assists in approval decisions. As a result, organizations can significantly reduce process completion times, minimize the risk of errors, and maintain full control over outcomes.