AI for SAP TRM: Generate your BRFplus Application from Business Rules in Natural Language

[ChristopheC]

When setting up a tax type in SAP Tax and Revenue Management (SAP TRM), creating the Business Rules Framework Plus (BRFplus) application with the business rules to apply to your forms can be a lengthy and time-consuming process.

Leveraging large language models (LLM) and machine learning, we have built a prototype that gives the user the flexibility to define the business rules in natural language and generate the corresponding BRFplus objects (application, functions, rule sets, rules, actions...), including validation, calculation checks, and billing document creation.

Prerequisites

• You have defined your form scenario in the Implementation Guide (IMG) of SAP TRM
• You have defined your form bundle type in the IMG

Process steps (it's easy as 1-2-3)

1. Provide the basic information about your tax type and decide if you want to use a sample prompt or a custom prompt.

2. Adapt the LLM prompt to your specific needs by entering the company code, the main/sub-transactions, and the system/user statuses in the example below. In the sample prompt below, we have a simple example of two fields:

• one is the Total Sales Income that cannot be empty or negative
• one is the Zero Rated Supplies which is equal to 20% of the Total Sales Income

 3. Review the LLM response, validate it, and send it for further processing.

 And voilà, the BRFplus application has been generated with all relevant objects!

Now that our BRFplus application has been generated, we can add the two newly created functions, VALIDATE and PROCESS, to the customizing of the form bundle type. The newly defined rules are then applied to the tax form.

In the BRFplus workbench, the newly generated BRFplus application can then be edited if needed.

How did we do it from a technical perspective?

This technical solution comprises several key components working in concert:

1. The Fiori Application: Serving as the central interface, this application utilizes a Wizard floorplan to guide users through the Business Rule modeling process. This step-by-step approach facilitates the collection of necessary data. The Fiori application also orchestrates the integration of all other components, including transmitting user-provided natural language instructions to the AI Core via the ISLM for encoding, and delegating the generation of BRF+ artifacts to the ABAP engine.

2. The ISLM to AI Core Interface: This component leverages the ISLM to access the AI Core's Generative AI capabilities. User instructions (provided in natural language) are augmented with meta prompts and then encoded into a JSON representation. Subsequently, this JSON payload is passed to the ABAP engine for BRF+ artifact generation.

3. The ABAP BRF+ Engine: This engine validates the incoming JSON application model, ensuring data integrity and structural correctness. Upon successful validation, it generates the corresponding BRF+ artifacts, accurately reflecting the hierarchies and attributes defined within the model.

What are the challenges solved by this prototype?

• Lengthy manual configuration of policies into business rules.
• Delays in adapting to regulatory changes and high deployment costs.

What are the business benefits?

• Accelerates tax rule implementation and compliance policy updates.
• Enhances accuracy, ensuring validation aligns with regulatory standards.
• Lowers operational costs by streamlining rule creation and updates.
  

The Intelligent Business Rules Generation is brought to you by the SAP TRM Initiative, embedding AI and BTP services to elevate the SAP TRM solution.

Are you an SAP TRM Customer? Do you want to know more about the Intelligent Business Rules Generation and other innovative solutions for SAP TRM? Reach out to your SAP Account Executive and request an SAP Business AI Discovery Workshop for SAP TRM, it's free of charge.