Germany shows a mixed picture when it comes to the use of artificial intelligence in software development. According to OutSystems' "State of AI Development Report 2026", 56% of the companies surveyed already use generative AI for application development. This puts this approach ahead of traditional programming, which was mentioned by 40%.

At the same time, the use of so-called AI agents remains comparatively low. At 9%, the proportion of companies that do not yet use this technology is higher than in other regions. Globally, however, 96% of organizations are using AI agents in some form, and 97% are reviewing strategies for company-wide deployment.

Agent-based AI is seen as a further development of traditional applications, as it can perform tasks independently, make decisions and adapt dynamically. According to Gartner, around 40% of business applications could have specialized AI agents by the end of 2026.

However, the increasing spread of AI brings challenges with it. Worldwide, 94% of companies see increasing complexity, growing technical legacy issues and additional security risks. At the same time, there is often a lack of central control approaches; many organizations have so far used AI agents in isolated environments.

Regional differences in maturity

There are also regional differences in the degree of maturity. While countries such as the USA, the UK and the Netherlands report medium progress, France is still in the early stages. Deployment is particularly advanced in the financial sector and in the technology industry.

On average, 31% of respondents state that AI is already an integral part of their development processes, while a further 42% use it in individual phases. In addition, 52% rely on a "human-on-the-loop" model, in which systems work largely autonomously but remain under human control.

Structural factors slow down deployment

In Germany, structural factors in particular are slowing down the use of agent-based AI. Companies cite integration into existing systems (38%), legacy infrastructures (34%) and governance and compliance requirements (31%) as the biggest hurdles. A lack of expertise plays a lesser role at 25%. The success rate is correspondingly low: Only 9% of companies report that at least three quarters of their AI projects are successful.

"The transition from AI experimentation to measurable business outcomes is no longer a vision of the future - it's already a reality," says Woodson Martin, CEO of OutSystems. "The results of the State of AI Development Report show a fundamental shift: developing software and developing AI systems are now practically one and the same."

Another problem is the fragmented system landscape. Worldwide, 38% of companies combine self-developed and prefabricated AI agents, which makes standardization and validation difficult. Only 12% have a central platform for managing these systems.

The report is based on a survey of around 1900 IT managers worldwide, which was conducted between December 2025 and January 2026.