Business AI in ABAP Cloud – A Strategic Overview (Part 1)
Cheers ABAP folks!
If you've been anywhere near the SAP developer community recently, you've noticed that one topic keeps coming up in every conversation: AI. And with it, a question that many ABAP developers have been asking themselves — “Where does ABAP fit in an AI-driven world?”
The short answer: right at the center.
In the 2026 SAP Developer Insights Survey, ABAP remains the most widely used language in the SAP ecosystem at 82%. ABAP is not fading into the background. It is the technology backbone for SAP S/4HANA, SAP Integrated Business Planning, SAP Business AI Platform, ABAP Environment and with the ABAP Cloud development model, it is evolving to meet the demands of a cloud-native, AI-enabled future.
In this blog post, I want to give you a strategic overview of how Business AI capabilities are taking shape in ABAP Cloud: What's available today, what's on the roadmap, and how you can start building.
ABAP Cloud and AI — The Big Picture
Before we dive into the AI specifics, let's set the stage. ABAP Cloud supports a comprehensive set of development scenarios: From transactional SAP Fiori apps and embedded analytics, through process and data integration, to background processing and on-stack extensibility. The diagram below shows the full landscape, and you'll notice that Joule, SAP’s digital assistant, is explicitly part of the picture. Business AI is not a separate silo or an add-on. It is woven into the same architecture you already use for building ABAP Cloud applications with CDS entities and RAP Business Objects.
Figure 1: Joule integration in the ABAP Cloud development model
The Two Pillars: Embedded AI and Agentic AI
SAP's Business AI strategy in ABAP Cloud rests on two complementary pillars. The following diagram illustrates the overall strategy:
Figure 2: Business AI Strategy in ABAP Cloud
Embedded AI Capabilities in ABAP Cloud
Embedded AI refers to AI capabilities that run directly within your ABAP application. These are powered by the ABAP AI SDK, which provides the bare-metal functionalities, and the ABAP Cloud development model, which adds the integration, compliance, and UX layer on top.
Available and planned capabilities include:
- Data Masking: Ensure privacy and compliance when sending data to AI models. Control what data leaves your system.
- Document Extraction: Extract structured information from unstructured documents directly within your ABAP processes.
- Document Grounding: Provide domain-specific context to AI models so that responses are accurate and relevant to your business data.
- Data Prediction with RAP Recommendations: Leverage prediction capabilities integrated into the RAP programming model to suggest field values, next actions, or classifications.
- Human-in-the-Loop with RAP Notifications: Keep humans in control. AI can propose, but users confirm with notification-based workflows built into RAP.
- Background Processing Framework: Run AI workloads asynchronously using the Background Processing Framework (bgPF), so that long-running AI tasks don't block your users.
The key advantage of Embedded AI is the seamless in-app experience. Your end users don't need to learn a new tool or change their workflow — AI capabilities surface right where they already work.
Feel free to checkout the tutorial written by @jasmin_gruschke to jump start into the ABAP AI SDK.
Agentic AI Capabilities in SAP Business AI Platform
Agentic AI takes a different approach. Instead of embedding intelligence into a single application, it enables Joule and Agents to orchestrate across your business applications from the outside, through governed APIs.
The architecture supports:
- Joule (broad-scope): SAP's digital assistant that can interact with your applications via natural language, performing the same tasks a user would.
- Agents (narrow-scope): Specialised agents designed for specific, well-defined tasks with high automation potential.
- Connectivity via OData, SQL, and MCP — Multiple integration paths ensure that Joule can reach your business logic regardless of how it's exposed.
Planned investments focus on ABAP side is on achieving unparalleled accuracy, generic API coverage across the application landscape, and enabling ECC environments, extending the reach of agentic AI beyond S/4HANA Cloud.
Ok, but when should I use what?
This is the question I get most often, and the honest answer is (like often in software development): it depends on the scenario, and in some cases, you'll want both.
Figure 3: Embedding AI vs. Agentic Orchestration
Embedded AI | Agentic AI via Joule | |
Development effort | Additional development required (ABAP AI SDK) | Remote service enablement required |
Flexibility | Less flexible (model replacements, prompting bound to your code) | More flexible (orchestrated centrally) |
User experience | Seamless, in-app, low entry barrier for end users | Natural language interaction via Joule, optimized via spaces concept in Joule Work. |
Best for | Heavily used processes with high acceleration potential | Broad application coverage, high automation potential |
My personal recommendation
- If a process is used hundreds of times a day, every second counts and your active business users are used to an existing application: Embed AI directly into your application to see results fast.
- If you want to enable your application landscape for AI-driven automation at scale, enable it for Joule. Be prepared to do this in the upcoming years for the most of your business processes.
Where can I find additional information?
If you would like to see an implementation of embedded AI in ABAP Cloud or the integration in Joule by using Joule capabilities in action, feel free to check out my recent Devtoberfest session on that. Watch it here:
Stay tuned for further blog posts, where we deep dive into the implementation details!
Cheerio ABAP Folks!
