Dionysios Satikidis, Head of Developer Platforms at Mercedes-Benz, shares why embracing AI is more important than perfecting it, how flagship projects help shape not just technology but culture, and why boldness sometimes matters more than methodology.
Mr. Satikidis, where are you currently focusing your attention when it comes to AI—and what topics in AI development are top of mind for you right now?
Right now, my primary focus is on human acceptance of AI. Even the best technology is useless if people don’t embrace it. We need to design AI in a user-centered way to foster exactly that kind of acceptance. The same applies to the methods and processes we use in development. That’s why my main goal at the moment is to identify potential barriers early—and address them in a targeted, proactive way.
What are the key challenges you’re currently focusing on in this context?
In development, we’re currently grappling with challenges like hallucinations and the often opaque decision-making processes of AI systems. These are some of the biggest hurdles when it comes to using AI to support software development. The more we move toward autonomous systems, the harder it becomes to understand their behavior and the rationale behind their decisions. Even with manually written code, it’s often difficult to fully reconstruct every step after the fact and identify the source of errors. When a larger share of the code is generated by AI, that lack of transparency increases, making debugging even more complex.
This challenge is especially apparent with highly user-friendly tools that intentionally abstract away the technical implementation. Here, a gap emerges between ease of use and technical traceability. Other systems prioritize transparency but require more in-depth technical knowledge. In the end, we need both: intuitive usability and reliable traceability. The key lies in strategically selecting and combining tools that are well-suited to each specific use case within the company—like how we use GitHub Copilot at Mercedes-Benz. We’re seeing the emergence of new toolchains, much like in traditional software engineering—but this time powered by agents and AI technologies.
What role does openness toward external tools and systems play—and when is an in-house solution the better choice?
Openness plays a key role, especially for developers and the overall development experience. Companies rely on external tools and systems to avoid constantly reinventing the wheel. But of course, that also brings complexity—and cost.
That’s why we’re building a centralized platform for our software development teams, internally known as Stargate. It’s not about giving every developer access to every tool. The goal is to foster collaboration and visibility at the project and innovation level: Who’s working on what, and with which tools? We’re creating transparency around both costs and value—particularly when it comes to new AI ideas and initiatives. This allows us to strategically align investments with our broader company goals. It also helps us determine which tools to adopt, and where in-house solutions are necessary to create real, differentiated value.
You can read the full interview with Dionysios Satikidis in our whitepaper, "Road to Agentic AI: The Fascination of Automation."