Interview with Saskia Kohlhaas, IAV
“Platforms are the backbone of AI-supported engineering”
Saskia Kohlhaas is Chief Digital Officer and Executive Vice President IT at IAV. A computer science graduate of Furtwangen University, she combines more than 20 years of experience in IT, digitalisation and business strategy with a strong focus on scaling AI-driven engineering in automotive development.
© 2023 IAV GmbH / Foto: Sonja Hornung
AI will only deliver real productivity gains in automotive software engineering when it is scaled across processes, toolchains and governance. Saskia Kohlhaas of IAV explains how AI-infused engineering can be implemented effectively.
Artificial intelligence promises
significant productivity gains in automotive software development, from
requirements engineering through architecture and coding to testing and
validation. The key question, however, is whether AI can move beyond isolated
pilot use cases and become reliably embedded in existing engineering processes.
Saskia Kohlhaas, CDO and Executive Vice President IT at IAV, is driving this
transformation by positioning IT as a strategic enabler of innovation and
efficiency.
At the upcoming Automotive Software
Strategies Conference in Munich, she will present “From IT to AI – At
Scale: An Approach for AI-Infused Engineering”, outlining how AI-supported engineering can be scaled across
organisations, platforms and development workflows. In this interview,
she explains where the main challenges lie and why success depends less on
individual models than on integration, governance and trust.
ADT: Looking ahead three to five years, what will be the
biggest challenge in scaling AI-supported software engineering in automotive
development?
Kohlhaas: The biggest challenge over the next few years will
be moving from individual AI use cases to scalable and reliable integration
across the engineering lifecycle. In automotive development, we operate in a
highly regulated and complex environment. This means AI
cannot remain an isolated capability. It has to work within existing
toolchains, processes and collaboration structures, and it has to meet the same
expectations in terms of quality, safety, traceability, IP protection and cyber
security as any other part of the system. Today, we still see a great deal of
experimentation at the edges. The real change happens when AI becomes part of
the way engineering work is actually done end to end, across different
projects, locations and partners. Scaling AI is ultimately less about model
performance and more about integration, governance and the ability to run these
systems in a stable and trusted way at scale.
Which technological capability will most strongly shape
the future of AI-driven engineering processes?
A key capability will be AI systems that can work in context
across different steps of the engineering lifecycle. In practice, we still
mostly see AI in relatively isolated roles, for example as coding assistants or
for specific tasks. The real shift happens when AI understands the
relationships between requirements, architecture, implementation and
validation, and can operate within that context. What matters most is not a
single technique, but the ability to combine AI with structured engineering
knowledge and make it reliably usable within complex development environments
as we already begin to see in advanced engineering setups.
How can AI help connect digital platforms and engineering
processes across SDV value chains?
AI can play an important role as a
connecting element across fragmented tools and data in the engineering
landscape. In many organisations, information remains distributed across
different systems and domains. This makes consistency and traceability
difficult. AI helps reduce this gap by working with that data in a more
integrated way and supporting alignment across requirements, design, software
and testing. What becomes particularly relevant in the context of
software-defined vehicles is continuity. Engineering is no longer a sequence of
isolated steps, but a continuous flow in which information is maintained and
updated across the lifecycle.
Where do you see the greatest productivity gains from AI
in automotive software development today?
We already see productivity gains today, especially when AI
is applied across the full flow from requirements through to implementation and
testing. AI can help transform initial requirements
into higher-quality, structured specifications and then support the derivation
of architectures, software components and test cases. This also includes
the use of coding agents, for example to work more effectively with existing
legacy code and handle tasks that traditionally required significant manual
effort. The key point here is not individual optimisation in single steps, but
the effect when the whole chain becomes more connected and consistent. This is
where we already see measurable improvements in delivery speed and overall
productivity.
How must engineering organisations evolve to successfully
integrate AI into their development workflows?
Organisations need to move from isolated experimentation
with AI tools towards more systematic integration into engineering workflows.
This requires clarity on where AI creates value in core processes, but also
consistent enablement so that engineers can actually work with these tools in
their daily work. At the same time, there needs to be ownership of the
platforms and capabilities that enable this, so that they can evolve in a
structured way over time and scale across projects and organisations. Overall,
the key shift is from experimenting with AI tools to
systematically integrating AI into how engineering work is performed and
improved.
What role do digital platforms play in enabling
intelligent engineering value chains?
Digital platforms are essential because they make
engineering data accessible, usable and processable at scale. They create the
foundation for connecting data across tools and domains, turning fragmented
information into a consistent and usable knowledge base. This is what
ultimately enables AI to deliver meaningful results in complex environments. In
this sense, platforms are not just infrastructure, they are the backbone of
data-driven and AI-supported engineering.
Finally, what do you personally hope to take away from
the Automotive Software Strategies Conference in Munich?
I am particularly interested in exchanging
perspectives on how companies are practically advancing AI in engineering, what
works, what does not, and where real progress is being made. For me, the value
of the conference lies in learning from different approaches, discussing
concrete challenges and building connections that help move the topic forward
collaboratively and translate insights into practical impact.
