Agentic AI and SAP: #2 What is SAP Knowledge Graph?

Last week, we understood what Joule Agents are.

One term that you will keep hearing with Joule Agents is..SAP Knowledge Graph.

Today, let’s break down SAP Knowledge Graph in 2 minutes.

First the Basic…What is Knowledge Graph?

Knowledge Graph is a well-known computer science concept that has existed for many years.

It’s a way of organizing information by storing not only data, but also the relationships between data.

Let’s take a simple example to understand. Let’s say there is a customer named John placed an order OD100 for a laptop.

In traditional database, this data is stored in tables. Something like below.

In this approach the relationships between objects exist. But to understand that John placed Order OD500 for Laptop, the application must connect tables using foreign keys.

Knowledge Graph on the other hand, stores data plus the relationships between the data.

The connection is stored directly. Even without joining the tables, the Knowledge Graph knows that:

• John placed Order OD500 for Laptop

• Order O500 belongs to John

Why Knowledge Graph is so Popular nowadays?

Knowledge Graph is not a new concept. It has existed for years. But what’s new is Gen AI.

Earlier most systems only needed to store and retrieve data. But now we have AI. AI must reason, understand context, and make decisions with the data. That’s where knowledge graphs become important.

With traditional database, AI needs to join multiple tables to understand the relationships among data.

But with knowledge graph, AI can easily understand how the data is connected with each other, whats the relationship among data.

In simple words…

AI can generate answers if it has access to data.

But…AI can generate right answers only if it has easy access to data, relationships between them.

And knowledge graph helps AI access them.

This is why almost every major AI company is investing in knowledge graphs.

So…What is SAP Knowledge Graph?

SAP Knowledge Graph provides a unified semantic layer that understands business objects and their relationships across SAP applications.

This is what Joule and Joule Agents use for business context.

By connecting business data, business processes, and business relationships across SAP solutions, SAP Knowledge Graph helps Joule agents understand and solve complicated business problems more effectively.

How Customers can access SAP Knowledge Graph?

Customers typically DO NOT access SAP Knowledge Graph as a standalone application.

Instead they access it indirectly through Joule, Joule Agents (or other SAP AI experiences).

For example, when a user asks Joule…

“Which suppliers are impacting delayed orders?”

The user talks to Joule.

Behind the scene, Joule may use SAP Knowledge Graph.

There is also something called…”SAP HANA Cloud Knowledge Graph Engine”?

“SAP HANA Cloud knowledge graph engine” and “SAP Knowledge Graph” are different.

SAP HANA Cloud knowledge graph engine is a part of SAP HANA Cloud. It is an addition to the SAP HANA Cloud multi model engines.

It is designed for processing large-scale graph data and executing complex graph queries using knowledge graphs.

Developers can build and query knowledge graphs using RDF ( a way of storing data as connected facts and relationships ) and SPARQL (language used to search graph data) inside SAP HANA Cloud.

What makes SAP Knowledge Graph special?

Most AI vendors combine a LLM with enterprise data.

But SAP has a advantage. After supporting business processes for more than 50 years, SAP understands how customers, suppliers, orders, invoices, employees, and thousands of other business objects are connected across the enterprise.

So, SAP’s approach is to bring together LLM, Business Data Cloud, SAP Knowledge Graph, and decades of business process expertise.

SAP Knowledge Graph gives AI not only access to business data, but also an understanding of how business objects and processes are connected.

As a result of this, Joule and Joule Agents (and other SAP AI experiences) can reason with business context rather than relying on data alone.

What’s next?

We now know what are Joule Agents and SAP Knowledge Graph. Next natural question comes…

How does Joule know which agent to use, what data to access, and which business steps to execute?

We’ll explore that next Monday.