AI agents will write code and translate legacy applications for use in the SAP cloud in the future. Sonja Liénard, Head of ABAP Platform at SAP, talks about the future of the SAP programming language ABAP and SAP ABAP Platform.

What is ABAP anyway? How would you explain it to those who have heard the term before but don't know exactly what it's about? And why is ABAP so important for business software?

Sonja Liénard: ABAP has a very long history at SAP. It is SAP's first and only proprietary programming language. She turned 40 in 2023.

However, ABAP is not comparable to other programming languages ​​such as Java or C++. ABAP is specifically specialized in developing and optimizing large enterprise business software. What is special is the high level of abstraction. This makes it very easy for developers ABAP Business-Develop or expand software. The complexity is reduced because security methods, authorization checks and quality controls are already built into the language. This allows developers to concentrate entirely on the business logic, i.e. on what task the program should ultimately complete.

ABAP has been continually developed. The latest version is ABAP Cloud with a limited language scope that makes it possible to develop in a Clean Core-compliant manner. This is very important for our cloud products. If you are still in a non-cloud environment, you can use it to prepare the code on-premise or in the SAP S/4HANA Cloud Private Edition so that it can also be operated in the cloud.

ABAP is also a platform. A fun fact is that we also largely use ABAP to develop the ABAP language as well as the ABAP Platform. The ABAP Platform runs under all SAP core solutions. It is an integral part of the SAP software, whether in older versions such as ECC, in on-premise solutions, in the S/4HANA private cloud or in the S/4HANA public cloud.

ABAP remains highly relevant

Will ABAP remain central to SAP customers?

Sonja Liénard: Yes, both in terms of the language and the ABAP Platform. ABAP remains highly relevant. The language of 40 years ago is no longer the language of today. It has been continuously developed and is included in the core ERP solutions of the SAP software and in extensions. There are around five million registered and around two million active ABAP developers worldwide.

With ABAP Cloud and a dedicated ABAP AI team, ABAP is a modern development language for business solutions. I don't know of any other that covers this scope. It is used globally. Almost all of the 100 largest companies worldwide are SAP S/4HANA customers and they always run the ABAP Platform.

In the future, agents will generate code

How will artificial intelligence shape the future of ABAP development?

Sonja Liénard: For me as Head of ABAP Platform, this is one of the most exciting topics. Artificial intelligence has completely shaken up the technology market. This also impacts the SAP portfolio for developers and how you can customize and extend our solutions. We've invested in AI-powered efficiency tools, such as a chat assistant that explains code. Ghost texting involves generating code suggestions as you write.

In the medium term, agents will generate code, including in the aforementioned large enterprise software business, and even build entire solutions. In the development environment of the future, the machine will take on many tasks for us.

What is very important for SAP is: How can we use AI to translate legacy code into modern code? It's not just about pure translation, but about taking the business logic with you. There are still a lot of customers with old solutions, also based on ECC. We need to show a clear path and offer tools that enable a quick and easy path to the cloud.

That's why we're developing a service that can be used regardless of the system version. All of SAP's ABAP AI capabilities will be included here – both for increasing developer efficiency and for migrating custom code. Ideally, this would be agent-based – as “Agentic AI”.

The magic of Agentic AI

What is Agentic AI?

Sonja Liénard: At Agentic AI you work with so-called agents. Agents are specialized, can communicate with each other, pass on results from subtasks and thus solve very complex tasks. Depending on the complexity of the use case, there are different approaches.

In most solutions, a so-called “orchestrator” is integrated. The agents relevant in the respective context are made known to him. The order in which the orchestrator calls the individual agents does not have to be static, but its real strength becomes apparent when the agents are flexibly combined in a dynamic network.

So it's no longer just about increasing efficiency for developers. If they are powerful enough, agents can build entire applications and thus take on some of the developer tasks. In our case, for example, Agentic AI can support the very complex task of code transformation, significantly speeding it up and reducing complexity.

Different agents can be relevant for this: for example, an agent that specializes in explaining customer-specific code; another making code changes; and a third who estimates the effort involved in a transformation project. The magic of Agentic AI then unfolds in the combination of these specialized agents.

AI will certainly change developer roles significantly. Developers will continue to make the specifications, but in the future they will concentrate even more on the business logic than on the actual coding. You work with the generated code and review it. The thought leadership role remains with humans, who pass their input to the AI ​​using good prompting. The entire AI field is extremely dynamic and evolving at an impressive rate. There are already powerful solutions available today, so we are not talking about the future here, but about the present.

Transforming legacy applications

What are the benefits of Agentic AI for customers?

Sonja Liénard: Agentic AI will provide significant value in transforming legacy applications and custom extensions into SAP cloud solutions, and thus the latest ERP versions. In February 2026, we expanded our existing custom code management app to include AI features that help the developer understand what the code does and what changes are needed to make it future-proof. And of course the AI ​​also gives recommendations on how the code can be expanded. In the future we will supplement this with agents. However, this takes time because we do not compromise on quality and safety.

We are also investing in the developer experience of the ABAP Platform to make using it as easy as possible. Agentic AI will help reduce the complexity that has arisen over decades of development.

Also very important: quality and safety

Should you be worried about security?

Sonja Liénard: No, we consciously take enough time before releasing our solutions, precisely because quality and especially security are very important to us. Don’t worry: AI will not take control and generate or integrate solutions without being asked and uncontrolled. Humans remain the thought leader and the final authority to review the code according to our standards.

What is the status – and what happens next?

Sonja Liénard: The ABAP AI tools to increase developer efficiency have been available since February 2025. We are currently developing Agentic AI in the ABAP context. Agentic AI itself still has to prove itself. My prediction: It will change the market.

Our self-trained and specialized ABAP AI model was released on the GenAI Hub at the beginning of January 2026. It is important to note that ABAP-1 is a specialized model that was designed to explain the code of ABAP programs.

We will then provide all ABAP AI tools in an independent side-by-side service. In a second step, we convert the use cases contained therein into agents.

We also support additional development environments (IDEs), particularly Visual Studio Code, for ABAP cloud development. We will also use the AI ​​tools available there to implement Agentic AI.

The interview is the first part of a series about ABAP and the use and interaction with artificial intelligence.