Systems | Development | Analytics | API | Testing

The 2027 S/4HANA deadline still looms large in the minds of SAP customers, but at this year’s SAP Sapphire event, SAP worked to move the conversation beyond cloud migration alone. Instead, they introduced a broader redefinition of what it means to be an “Autonomous Enterprise.” At the center of this new Autonomous Enterprise strategy is agentic AI. SAP envisions the future enterprise as one that can leverage its business data to power agents across its ERP applications.

Not long ago, the answer to who writes tests was simple: the quality assurance (QA) engineer does. They sat downstream of development, received a build, and translated requirements into scripts. It was a defined role with a defined output. That clarity is gone. In 2026, the person or system responsible for test creation might be a business analyst (BA) mapping out a customer journey, an AI agent expanding test coverage overnight, or a QA engineer who hasn’t written a traditional script in months.

Quick summary: AI is creating anxiety and excitement — teams can get more work done faster, but does all this automation leave the worker behind? Not necessarily, says one performance engineering leader. The AI revolution, he says, is another technological wave. To ride it, performance engineers must embrace the change.

At some of the most sophisticated engineering organizations in the world, the best developers are already writing zero percent of code manually. AI agents are generating features, spinning up test suites, and moving software through delivery pipelines faster than most governance frameworks were designed to handle. The speed is real, and so is the exposure that comes with it.

Quick summary: AI accelerates code creation, but its inherent confidence pushes structural risks downstream, where they surface as costly, release-blocking problems. As code output scales, performance validation that can’t keep pace becomes a headache and a business risk. Agentic performance testing embeds skepticism and performance awareness into the development process before risk can compound. Software development requires specialized expertise for a reason.

AI is now writing more of your code than ever. But the problem is that your test suite was built to catch errors, not to catch the difference between what an AI agent produced and what your original specification actually required. As AI tools accelerate development velocity, the volume of code moving through pipelines is outpacing traditional quality processes.

Take this scenario: Your team ships a release with 91% code coverage. Every test in the suite passes. The pipeline is green, and leadership signs off. But two days later, a critical defect surfaces in production. Upon investigation, you find that the changed code was never actually tested, and the tests that were run covered different paths entirely. That 91% was real, but it was just measuring the wrong thing. And as AI tools generate more of the code inside those pipelines, the gap widens.