AI in Practice: Retrieving Project Information in SAP PS

In our „AI in Practice“ section, we’ll guide you step by step through a specific use case: from navigation and customization all the way to functional testing. This article focuses on setting up a chatbot for AI-powered retrieval of project information in the SAP Project System (PS) based on CDS views.

Transparency regarding deadlines, budgets, and progress is crucial, especially in project management. SAP PS is complex: Important information on network plans, PSP elements, or status updates often has to be searched for in different transactions. This is time-consuming for project managers and controllers and delays decision-making. Thanks to artificial intelligence, project data can now be retrieved quickly and intuitively via voice commands without in-depth SAP expertise. Through a modern chat interface, users can easily query start and end dates, project structures, or reports using natural language. The chatbot outputs the results in Markdown format, structured as a table or graph. This saves time, reduces the training required for new employees, and enables better decisions based on up-to-date data.

How Project Managers Bring Our Chatbot onto the Team

To ensure that AI can be used in SAP securely, flexibly, and in a way tailored to each company’s specific needs, Milliarum provides two key components: With the Milliarum AI Construction Kit and the General chatbot AI-powered applications can be configured in SAP in just a few steps. These applications are based on Core Data Services (CDS) views, which deliver project data with high performance. Our unique selling point is the complete flexibility we offer our customers: From OpenAI to European cloud providers to your own data center, our technology can be deployed according to your security requirements. The result is a chatbot tailored to the specific needs of project management that lightens the load on project managers and controllers by providing all relevant data in seconds.

With Governance and Data Access: Flexible AI Integration in SAP

The AI Construction Kit is the central platform for IT managers to design their own AI use cases in SAP, such as retrieving project information in SAP PS. It provides a customization environment that allows users to integrate data sources, entities, and CDS views without requiring in-depth programming knowledge. The platform is extensible, allowing multiple CDS views or tables to be combined using so-called Dependent Data Parts to enable complex project analyses. Every configuration step is logged in the system, ensuring that IT departments retain full control over data sources, authorizations, and functions. The General Chatbot serves as the actual user interface and acts as the entry point for project management and controlling. Here, users can submit queries in natural language, such as „Show me all projects with a due date next month.“ The responses are generated directly from SAP data sources. The chatbot is particularly valuable for IT departments because it provides a standardized interface that can be used for various modules—such as SD, MM, PS, or PP—and can be precisely tailored to different scenarios through Customizing. The interaction between these two tools ensures that IT departments do not have to hard-code every single function or develop their own solutions. Instead, they configure the data sources and personas once in the Milliarum AI Construction Kit and then provide the business department with a universally usable chatbot. The result is a secure and flexible AI integration into SAP that meets both IT’s needs for governance and control and the business departments’ requirements for fast, intuitive data usage.

10 Steps to Creating an AI Chatbot for Project Data in SAP PS

1. start the App Manager
In the SAP system, open the Milliarum AI Construction Kit using transaction /MILUM/5GCORB_AI_CUA or /MILUM/5GCORB_AI_CUL.
Search for the entry "Generic AI Chat" and click the Customizing icon (gear).

2. create Managed Datasource
Start the Managed Datasource Manager.
Click „Create Managed Datasource“ and enter the following:

• ID: ZPS_DEMO
• Designation: PS Expert Demo

Save your entries.

3. define managed entity
Open the tab Managed Entities and click „New Entries.“.
Create a new entity, for example:

• ID: ZPROSYS_DATA
• Designation: Overview of Project Data
• Source: CDS-View I_ProjectBasicData

4. Add the "Leading Data" section
Select your entity and open the "Leading Data Parts" section.
Click „New Entries“ and enter the following:

• Name: ZPS_PSDATA
• Core: /MILUM/GAGDS_DSRC_CDS_DYN
• CDS-View: I_ProjectBasicData

Save the configuration.

5. select attributes
Open Attribute Mapping and select the fields you want to display in the chatbot, for example:

• PROJECT (Project)
• PROJECTDESCRIPTION (Description)
• PLANNEDSTARTDATE (Planned Start)
• PLANNEDENDDATE (Planned End Date)
• PLANT (Plant)
• PROJECTPROFILECODE (Project Profile Code)

Next, generate the attributes and save them.

6. create a persona
From the menu, select Persona create.
Define:

• Persona ID: PS_DEMO
• Designation: PS Expert Project Data
• LLM: GPT-4o About the OpenAI Connector

In the system prompt, specify that the chatbot is an SAP PS expert specializing in project data.

7. configure function
Open the „Functions“ tab and click "New Entries.".
Enter:

• Functional class: /MILUM/5GAICB_F_AI_FUNC_MDSRC
• Function ID: ZPS_DEMO-ZPROSYS_DATA-R
• Description: Overview of All Project Data

Save your entries.

8. assign persona
Open the Application Customizing, navigate to the personas and click „New Entries.“.
Select your persona PS_DEMO and assign them to the chatbot.

9. perform test
Open the chat, select the "PS Expert Project Data" persona, and enter:
„Show me all project data“
Check whether the table is displayed correctly with the selected attributes.

10. Fine-Tuning with Variants & Dependent Data Parts (optional)
Open for Variants For your entity, change the column order, define sort orders, or customize field names. For dependent data parts, add additional CDS views (e.g., for project type descriptions or network activities) to further enrich the output.

You can find the complete documentation, including all screenshots and examples, at here.