Stop Guessing! Use Case with Conversational AI Agents in SAP Analytics Cloud

[hannahhorvath]

Introduction

Have you ever stared at your analytics dashboard, wishing it could do more—like answer your questions without having to solely search, filter, and drill into your data? Well, guess what – you can do that now!

In this blog, I’ll walk you through an exciting use case that leverages SAP AI Core with SAP’s AI Agent through the SAP Business Technology Platform (BTP) in the following order:

1. ‘Stop Guessing!’ Use Case – Business View
2. What are AI Agents?
3. Why SAP AI Agents?
4. ‘Stop Guessing!’ Use Case – Technical View
5. How can you get your hands on this?

1. ‘Stop Guessing!’ Use Case – Business View

Imagine you’re a business user at a Gas Provider Company, exploring your Finance and Billing Dashboards. As you dive into the billing data, a critical insight catches your eye: late fees are unusually high in a region. This sparks concern about potential customer dissatisfaction and retention issues. Here’s where the power of AI shines. By leveraging a Conversational AI Agent directly within your SAP Analytics Cloud (SAC) dashboard, you can uncover the underlying trends, identify contributing economic factors, gain actionable recommendations, and more – removing the need to “guess” your next steps.

Let me show you how a Gas Provider Company can use a Conversational AI Agent to accelerate decision-making and uncover deeper insights beyond the 4 walls of their Enterprise. Watch the 'Stop Guessing' Use Case Demo below!

As you saw in the video above, the business user was able to receive an output that combines their Enterprise data from the HANA database and External data from PDFs and websites. This blog will further showcase how this capability allows businesses to have insights beyond the data in their 4 walls resulting in realistic, reliable results to drive business decision making. In other words, customers are no longer limited in data and information. With AI Agents you have the possibility to leverage a multitude of tools thereby providing you with a competitive edge. In fact, Gartner has predicted that by 2028 one third of interactions with generative AI will use autonomous AI Agents to complete tasks (Source).

To further understand the value SAP AI Core Services bring with AI Agents, we must first understand what an AI Agent is.

2. What are AI Agents?

AI Agents are NOT chatbots, but rather they are autonomous applications powered by large language models (LLMs) that can integrate with applications and business processes to achieve complex goals. They can fall into two categories:

1. Task Oriented – Agents can automate workflows and multi-step tasks by leveraging integrations such as with SAP Build Process Automation.
2. Conversational – Enable natural, human-like conversations to help users gain deeper insights as seen in the use case.

 Agents achieve their goal by using multi-step planning, self-examination and reflection of work to propose improvements, tool usage, and multi-agent collaboration to split-up tasks and come up with a solution. Some of these concepts will be seen in more detail in the ‘Stop Guessing!’ Use Case -Technical View section.

For example, in the use case video, the Conversational AI Agent examined the prompt in multiple steps by calling on the data and tools. In parallel, it also performed reflection activities to continuously learn and improve. Although there was no multi-agent collaboration here, an example would be a specific scenario AI Agent such as triggering a workflow to automate PR creations for approved marketing promo campaigns or, perhaps, calling on another Conversational AI Agent that has PR information specific tools to achieve goals around answering PR related topics.

 Source: Agents is the most important BTP release this 2024

Source: AI108 TechEd Session Session Catalog - SAP TechEd 2024

3. Why SAP AI Agents?

The market for AI Agents is rapidly growing with innovations from OpenAI, Microsoft, Google and other contributors. What makes SAP AI Agents valuable?

1. SAP AI Agents integrate your semantically rich SAP data and processes, ensuring that the meaning of your business is captured. In other words, SAP knows SAP the best as it can understand, integrate, and capture your data's Business Context (eg. Hierarchies, metadata, currency conversion, languages, security configuration) - providing meaning to the AI Agent to reduce hallucinations and removing the need of manual rebuild of your data’s business context.
2. The AI Agent integrates with SAP AI Core services by collecting data and leveraging underlying large language model (LLM) within the AI Core. It is also leverages HANA's Vector Engine for Retrieval-Augmented Generation (RAG). This advanced setup minimizes hallucinations, delivering more accurate, reliable, and trusted results.
3. By leveraging SAP AI Core services and other BTP offerings, this will allow for improved efficiency and collaboration across your integrated Intelligent Enterprise Portfolio. You can create enterprise-grade AI Agents tailored to your unique needs that integrate seamlessly with tools like HANA, OData applications, SAP Build Process Automation, and even external sources like websites and PDFs, while ensuring you are the decision maker at the center of it all. This ensures secure interactions with your business data, processes, and systems while adhering to SAP’s clean core principles.

4. ‘Stop Guessing!’ Use Case – Technical View

In this section, we will dive deeper in the Use Case and explore the use case architecture and witness a Lab Preview of the Project Agent Builder – the new approach to develop agent-based projects in BTP.

Use Case Architecture

In this use case, the SAP Analytics Cloud (SAC) dashboard is leveraging data from the HANA database that has been further transformed and modeled in SAP Datasphere. With SAP Datasphere's native connectivity to SAC, the data maintains it's Business Context and can be consumed in real-time analytics. In addition, the Conversational AI Agent has been embedded into SAC as a custom application widget via Business Application Studio (and can also be embedded via SAP Build Code).

The AI Agent leverages 3 tools in this use case: Semantically rich data from HANA, Financial PDF documents, and the Gas Providers' Supplier Website. If required more tools can be added beyond the 3, allowing you to have that flexibility to achieve your goal and leverage Enterprise data with External data.

When a question is asked to the Conversational AI Agent in SAC, it sends this to the LLM and grounds it using the HANA Vector Engine & RAG while searching the tools for the appropriate data to respond back with. In addition, it performs muti-step planning to search for information using the 3 assigned tools to produce an answer while continuously reflecting for improvement.

Project Agent Builder

Project Agent Builder is where you can create your Conversational Agent and define the various tools and integrations to achieve your goal. Below you will see a lab preview screenshot of the Project Agent Builder. Upon creating a new Agent, you must identify the following:

1. Description of the Agent’s expertise and goal with instructions to illustrate the Agent’s behaviour.
2. Maximum thinking or multi-step planning steps it must take before providing a response.
3. Selection of the LLM provider, which in this case is OpenAI GPT4o.

 

 To better understand how you can leverage the Project Agent Builder and how it operates in this Use Case, check out the video below!

5. How can you get your hands on this?

AI Agents are planned to be released as a BTP Service in H1 2025.

You can additionally leverage this BTP service with Joule Studio to extend beyond the core and collaborate between multi-agents for scenarios. Further roadmap innovations will also allow you to extend your AI Agent with SAP Build Process Automation to automate tasks.

Check out the links below for our AI Roadmap in Q1 and Q2 2025.

Q1 2025 Roadmap - New Generative AI Tool Kit with SAP HANA Cloud: SAP Road Map Explorer

Q2 2025 Roadmap for AI Agents for Advanced Business Scenarios:  SAP Road Map Explorer

Source: AI108 TechEd Session Session Catalog - SAP TechEd 2024

 Further information on AI Agents with Joule Studio can be found here.