AI Is Changing the Shape of SAP Work

The question is not whether AI can configure a pricing procedure. It is who decides the pricing procedure was wrong in the first place.

That distinction is the whole story of the next five years in SAP work. Most of the noise is about replacement — which roles disappear. The more useful question is about shape: which parts of the job lose value, which parts gain it, and what a good SAP professional is actually paid for once the routine is automated.

The shift is real — but not the one in the headlines

Agentic AI is good at bounded, well-described tasks with a clear definition of done: drafting configuration from a specification, proposing a mapping, generating test cases, explaining an error, writing the first version of a functional spec. That is real, and it is already happening.

It is weak exactly where SAP work is hard: deciding what the process should be, reconciling three business units that each swear their exception is mandatory, judging whether a gap is a genuine requirement or a habit, and owning the consequence when the decision turns out wrong. AI has no skin in that game. It produces confident output without accountability — and in an SAP programme, accountability is the scarce resource, not output.

So the first thing automated is not the consultant. It is the keystrokes.

Where the value moves

As configuration and documentation get cheaper, the premium moves to four capabilities that AI does not supply:

Process ownership. Not “I can build it,” but “I can decide what to build, defend the decision to the business, and live with it.” The scarce skill is judgment under disagreement.
Architecture and Clean Core discipline. When extensions are trivially easy to generate, the person who knows what not to build — what stays in the standard, what belongs on BTP, what will haunt the next upgrade — becomes more valuable, not less.
Data and integration. AI without trusted data and process context is a presentation, not a transformation. The people who make the data trustworthy and the interfaces reliable are the ones whose work AI depends on.
Change and adoption. Software has never been the hard part. Getting a real organisation to work differently is, and no agent will do that conversation for you.

Notice what these have in common: they are the parts of the job that require standing in front of a business owner and being right, or being accountable for being wrong.

The reality check

“AI will replace SAP consultants” is a lazy claim, and the opposite — “nothing changes” — is worse. Both skip the actual movement.

What changes is the mix. A role that was 70% doing and 30% deciding inverts. The professional who only configured to spec is exposed. The one who could always explain why the process looked that way was already the valuable one — AI simply removes the busywork that used to hide the difference between them.

For the people who were coasting on scarcity of hands, this is uncomfortable. For the people who were carrying the judgment, it is the best thing to happen to their market in a decade.

What to learn

If you are an SAP professional, stop competing with the agent on speed. Compete on the things it cannot do:

Learn the business process end to end, not just your module’s screens. Know why the order becomes a delivery, not only how.
Get fluent in Clean Core and extensibility decisions — the difference between a justified extension and technical debt.
Move upstream into requirements and process design: the ability to challenge a request, not just implement it.
Learn to use the tools well — prompt the agent, review its output critically, catch where it is confidently wrong. “Can operate AI” is table stakes; “can judge AI output” is the skill.

What to hire for

If you are hiring, your job descriptions are probably still written for the old mix — long lists of modules and versions. That filters for exactly the profile AI is commoditising.

Hire for judgment, process ownership and the ability to say no with a reason. Test candidates on a real decision, not a certification list: give them a messy requirement and ask what they would not build. The certifications tell you what someone was taught. The decision tells you whether they can be trusted with yours.

The bottom line

AI is not thinning the SAP profession. It is sorting it — separating the people who move keystrokes from the people who own outcomes. The first group should be worried. The second group should be learning to lead the first.

The roles are not disappearing. They are changing shape. The professionals and employers who understand the new shape first will spend the next five years hiring from a stronger position than everyone still arguing about whether the robots are coming.

— Andreas BORN

Read the Talent Brief — the SAP skills, careers and hiring decisions that matter, once a month. Subscribe →

Read The Talent Brief — the SAP skills, careers and hiring decisions that matter, once a month. Subscribe →