INSIDE

Eckard: Vasi, you recently started with Siemens. In your opinion, what is Siemens’ unique value proposition when it comes to AI?

Vasi: Eckard, I think physical AI is the next frontier for generative AI, and very few companies can actually execute on this vision. At AWS, where I spent the last 8 years of my career, we were incredibly successful at building AI services, but you couldn't say that for industrial applications. You need deep domain expertise that takes decades to develop - understanding how manufacturing works, how buildings operate, how industrial processes connect. The unique opportunity at Siemens is that we can bring together the physical and digital worlds in ways that AWS, Google, or OpenAI simply cannot. They can build great AI models, but they don't understand the nuances of a production line or the complexity of building operations. Siemens has that industrial DNA combined with the scale to validate AI solutions internally before taking them to customers. That intersection is incredibly rare and powerful.

Eckard: In your new role, you have a clear strategy for where Siemens' AI portfolio will be in the next 3 – 5 years. As we are unlocking data access and building core infrastructure for Siemens, how do you see this foundation enabling that future vision? What aspects of our work being carried out at Siemens internally are most critical for the AI breakthroughs you're envisioning?

Vasi: The foundational work that GBS in collaboration with other Siemens units and functions is doing is absolutely critical because it's creating what I call the "nervous system" for our AI vision. When I look at building operations today, for instance, we are working with a diverse set of systems - Desigo for controls, Brightly for managing assets, Surveillance ensuring security. What we need is a unified intelligence layer that becomes the "brain" integrating all these signals. 

The data infrastructure being currently built enables us to move from connected buildings to truly autonomous buildings that continuously learn and optimize across energy, comfort, and security. This same principle applies to manufacturing - we need that unified data foundation to support the flexible, AI-driven production lines I envision. Without this foundational work, we'd just be building isolated AI applications instead of creating the interconnected intelligence that truly differentiates Siemens.

Eckard: Through initiatives like Data Access, we are working at GBS on establishing a Siemens-wide data ecosystem. How does this democratized access to industrial data streams impact your team's ability to develop AI models faster? What customer opportunities can you now pursue that weren't feasible before?

Vasi: The democratized data access is a game-changer for developing what I call "industrial foundation models." Unlike general AI models that help you create text or images, we're building models that can design manufacturable machine parts, optimize production processes, and understand the deep connections across industrial value chains.

With access to Siemens' vast industrial data streams, we can train models that have true world knowledge about how things work in factories, buildings, and infrastructure. This enables us to pursue opportunities like creating foundation models that can generate not just 3D designs, but also tell you how to manufacture the product, what processes to use, and what paths CNC machines should take.

We can now tackle the holy grail of manufacturing - moving from rigid, pre-programmed production lines to flexible systems that use vision-language-action models. These systems can adapt to changes, recover from mistakes, and handle variations without requiring exact programming. This opens up customer opportunities in automated manufacturing that simply weren't feasible with traditional approaches.

Eckard: Given your experience building AI businesses at AWS, how do you approach prioritization when there are hundreds of potential AI use cases across Siemens? What's your framework for deciding where to focus first?

Vasi: The key is thinking bigger than individual use cases and focusing on foundational capabilities that can scale across multiple applications.

My framework has three criteria: First, does this leverage Siemens' unique industrial domain expertise? Second, can we as Siemens act as customer zero before taking it to customers? Third, does it create a scalable platform rather than just solving one specific problem?

That's why I focus on areas like autonomous buildings, flexible manufacturing, and industrial foundation models. These aren't just use cases - they're platforms that can transform entire industries. 

I also rely heavily on internal teams like GBS who understand different parts of Siemens' business to help identify where the highest value pain points are. The intersection of high business value and unique Siemens capability is where we'll create the most impact.

Eckard: You are quite familiar with the type of business GBS is doing. Our ProcessAI initiatives like TenderGPT and exception handling in accounts payable with agentic AI are generating significant internal productivity gains. What do you see as the biggest advantages of these solutions?

Vasi: These internal AI successes demonstrate something profound about the future of business operations. What we're seeing with agentic AI is the ability to completely automate complex, multi-step processes that previously required thousands of people in various countries doing manual work. But more importantly, we're enabling what I call "low-cost scaling of high-touch engagements."

Think about hyper-personalization: with reasoning models, we can now create customized products and services on the fly based on individual customer behavior and needs. Whether it's a bank creating products for an Uber driver who only works weekends, or Siemens creating tailored industrial solutions, we can deliver that high-touch, personalized experience at massive scale without the traditional cost barriers.