Boosting the efficiency of software development with the help of AI

The implementation of AI-based functions in our software solutions and the use of AI-based tools for their further development are key pillars of PROSTEP's AI strategy. Our objective is not to pare down staff as the result of AI-based co-programming but to make more efficient use of what are traditionally scarce resources in software development. We hope that by using GitHub Copilot we will be able to complete future development tasks much faster, reduce time-to-merge, improve code quality, and generate better and more complete software documentation more easily.

Microsoft's GitHub Copilot is an AI-based assistant, which like ChatGPT uses generative AI methods to automatically generate code, detect and eliminate errors in existing code, generate tests, and document the generated code. We compared the tool with various other assistants and decided to test it in a productive environment because we believe it is currently the most advanced. One of the reasons for this is the fact that GitHub manages the largest amount of code in the cloud and therefore has the largest amount of potential training data.

However, when using Copilot in our software development, we make sure that our intellectual property and that of our customers remains fully protected. Because we use business licenses, Microsoft provides a contractual guarantee that the code will not be stored externally or be used for training purposes. As before, we use static code analyses to check that the generated code is protected against hacker attacks.

PROSTEP believes that the use of AI-based assistants is a key step towards boosting the competitiveness of software development in Germany compared to offshore software development. Our customers benefit from the fact that new functions or greater functionality per release are made available to them more quickly. Better documentation also makes it easier for them to integrate the software in their existing IT landscape.

As mentioned earlier, we expect the use of AI to significantly increase efficiency in software development. This does not, however, mean that developers will be replaced as the need for programming on the whole will continue to increase. If anything, developers who do not make use of AI tools run the risk of being replaced because they are inevitably less efficient.