SAP-samples/teched2025-IN162

IN162 - Real-time data for AI agents: Grounding with event-driven architecture

Description

This repository contains the material for the SAP TechEd 2025 Hands-on Workshop session IN162 | Real-time data for AI agents: Grounding with event-driven architecture

Requirements

There are no dedicated requirement for this session. You would be able to execute the excercises by just following the descriptions even if you do not have any experience with SAP Integration Suite and SAP Integration Suite, advanced event mesh.

However, you will be able to derive more value from this session, if you have some knowledge and understanding of event-driven architectures, namely events, queues, topics, event subscriptions along with SAP Integration Suite, SAP Joule Studio and SAP Build Process Automation Actions.

You can explore the following SAP Discovery Center missions to get started and build expertise with the SAP BTP services used in this hands-on workshop session.

• SAP Integration Suite
• SAP Integration Suite, advanced event mesh
• SAP AI Launchpad
• SAP AI Core
• Joule Studio, skill builder
• SAP Build Process Automation

Session Overview

This session introduces attendees to the power of event-driven integration pattern to deliver real-time grounding data, equipping AI system's with the latest business context to make informed and accurate decisions

Note

AI Grounding is the process of connecting an AI system's abstract knowledge and responses to specific, real-world data and context, reducing "hallucinations" and increasing the reliability and accuracy of its outputs.

Although AI Grounding is primarily used for unstructured text documents, it can also be extended to real-time structured business objects by embedding key attributes or metadata. This approach helps improve the factual accuracy and contextual relevance of AI responses when dealing with transactional business data.

In this session, we’ll explore how to apply real-time vector grounding by connecting structured data to an LLM using a vector database using SAP Integration Suite and SAP Integration Suite, advanced event mesh. You’ll see how updates to the data can instantly influence the model’s responses, keeping them current and contextually accurate.

By the end of this session, you’ll understand the core principles behind real-time vector grounding, how to design an embedding strategy for structured entities, and how to integrate this into a sample AI workflow.

The following chapter explains the scenario in more detail:

• Scenario Introduction

Pre-configured Setup

Chapters in this section provide pre-configured setups to support learning, without being part of the hands-on exercises:

• SAP S/4HANA Cloud System Configurations (for your information only)
• SAP Service Cloud Version 2 System Configurations (for your information only)

System URL and login information

To complete the exercises, the access to the following systems have been provided.
In case you are not able to access any of these systems please contact the workshop instructors.

• SAP Integration Suite, advanced event mesh
• SAP AI Launchpad
• SAP Integration Suite - Primary Tenant / SAP Integration Suite - Backup Tenant
• SAP S/4HANA Cloud
• SAP Service Cloud Version 2
• SAP Joule: Customer Success Digital Assistant

Important

• For a smooth experience, tenants have been preconfigured, and you already have all the roles and permissions needed to complete this exercise.
• User ID and password information will be provided to you by the instructors.
• When you run through the exercise steps, you need to ensure that the technical IDs of the integration artifacts that you will create are unique. Hence, add a participant number to your integration artifacts. The instructors will assign the participant number to you.
• Please adhere strictly to the instructions regarding the naming conventions for the artifacts you create. This will ensure successful completion of the tasks without conflicting with other participants.
• Do not delete, change or undeploy any artifact in the tenant other than yours.

Exercises

The complete list of exercise steps are listed below, run through them in the given order.
This section also serves as a Table of Content; use the breadcrumb navigation at the top of the pages to return here at any time.

• Exercise 1 - Explore and Configure SAP Integration Suite, advanced event mesh (AEM)
  • Exercise 1.1 - Log on to SAP Integration Suite, advanced event mesh (AEM) and explore it
  • Exercise 1.2 - Create first queue and subscribe to sales order topic in SAP Integration Suite, advanced event mesh (AEM)
  • Exercise 1.3 - Create second queue and subscribe to support case topic in SAP Integration Suite, advanced event mesh (AEM)
• Exercise 2 - Expose Embedding and Summarization Models as an API Using Gen AI Hub (AI Core)
  • Exercise 2.1 - Log on to SAP AI Launchpad and Create Configuration
  • Exercise 2.2 - Create Deployment
• Exercise 3 - Create an Integration Flow for S/4HANA Sales Order event to embedding model to SAP HANA Vector DB for AI Grounding in SAP Integration Suite
  • Exercise 3.1 (Recommended and an alternate to Exercise 3.2) - Copy an existing Integration FLow to reciever a Sales Order creation event, transform into embeddings and persist to SAP HANA Vector DB
    OR
    
  • Exercise 3.2 - Create an Integration Flow from scratch to receive a Sales Order creation event, transform into embeddings and persist to SAP HANA Vector DB
• Exercise 4 - Create an Integration Flow for Service Cloud Support Case event to embedding model to SAP HANA Vector DB for AI Grounding in SAP Integration Suite
  • Exercise 4.1 (Recommended and an alternate to Exercise 4.2) - Copy an existing Integration FLow to receive a Support Case creation event, transform into embeddings and persist to SAP HANA Vector DB
    OR
    
  • Exercise 4.2 - Create an Integration Flow from scratch to receive a Support Case creation event, transform into embeddings and persist to SAP HANA Vector DB
• Exercise 5 - Create a new Sales Order and a Support Case to trigger the respective integrations using an event-driven pattern
  • Exercise 5.1 - Create a new Sales Order in SAP S/4HANA Cloud system
  • Exercise 5.2 - Monitor Message Processing Logs in Cloud Integration after Sales Order Creation
  • Exercise 5.3 - Create a new Support Case in SAP Service Cloud Version 2 system
  • Exercise 5.4 - Monitor Message Processing Logs in Cloud Integration after Support Case Creation
• Exercise 6 - Customer Success Digital Assistant: Extending Joule with Joule Skill using Real-Time Vector Grounding
  • Exercise 6.1 - Generate summary of talking points for a customer meeting considering latest customer's sales orders and support tickets
  • Exercise 6.2 - Go through pre-built Integration Flow that summarize the current status of the customer using Generative AI Hub (Optional)
  • Exercise 6.3 - Go through pre-built Joule Skill to trigger the Integration Flow as an Action for generation of key talking points for customer meeting (Optional)

Feedback

Kindly provide your feedback on session IN162.

Additional Relevant Sessions

You can also gain some further knowledge around SAP Integration Suite and SAP Integration Suite, advanced event mesh by attending the following SAP TechEd Hands-on Workshop sessions:

• IN103 | Helping systems talk smarter with a flexible event-driven architecture | Deep Dive
• IN165 | Experience event-driven integration with advanced event mesh | Hands-on Workshop
• IN160 | Empower your business through enterprise automation | Hands-on Workshop
• IN163 | Implement exactly once in-order delivery in SAP Integration Suite | Hands-on Workshop

Contributing

Please read the CONTRIBUTING.md to understand the contribution guidelines.

Code of Conduct

Please read the SAP Open Source Code of Conduct.

How to obtain support

Support for the content in this repository is available during the actual time of the online session for which this content has been designed. Otherwise, you may request support via the Issues tab.

License

Copyright (c) 2025 SAP SE or an SAP affiliate company. All rights reserved. This project is licensed under the Apache Software License, version 2.0 except as noted otherwise in the LICENSE file.