Everyone is talking about AI. Most are experimenting with it. Only a few are truly building their businesses around it. These few organizations are called Frontier companies. Frontier companies do not use AI as a tool, feature, or innovation project – they use it as a foundation. AI is not just on top there; it is built into processes, decisions, products, and business models. Without AI, the company simply would not perform as well.
In this interview, Frederik Jelden, Strategic Account Technology Strategist at Microsoft, discusses how to recognize AI-native companies, which structural and mental barriers are holding many organizations back, and what companies need to do now to generate real business impact with AI, rather than just going through the motions. Frederik combines deep technical expertise with strategic consulting in the areas of cloud, data, and artificial intelligence. He works on global transformation and innovation initiatives, lives in Berlin, and is just as passionate about discussing corporate transformation as he is about good food. But now, on to the interview about frontier companies.
For me, a company is AI-native when AI is not merely an add-on but a constitutive element. So it i s not «We have AI projects,» but rather: Without AI, this business model would not work the way it does. AI is not an end in itself; existing processes are redefined through AI, and new business models are developed through and with AI. A major difference for me is that AI is not limited to flagship projects (value cases, use cases, etc.) but is embedded in every core process. AI is not a limited program for the company; it is part of its new DNA – essentially a new muscle for the company. This also means that AI is not purely an IT matter; rather, control clearly originates from the business and functional departments and is linked to strategic business goals.
In a country of engineers, we always want to understand everything 120%, so non-deterministic outputs are initially viewed with extreme skepticism. And while that can often make perfect sense, unfortunately, a pragmatic approach to identifying the areas of real value is sometimes lacking. Furthermore, I see an extreme focus on tool and model selection. In many cases, models are almost a commodity here. I believe the issue is more about organizational hurdles: AI means change, and this change must be managed. Organizational design and human resources are called upon to implement the right programs here. And the biggest structural barrier is still data quality and fragmented data landscapes. Without trust in the data, AI can hardly deliver any added value.
Do not build grandiose, pie-in-the-sky use cases. Instead, empower every department and every team to make their processes more efficient with AI. Create the right incentives for this and make the impact of AI measurable across the entire company. We live in a world of media discontinuities today. Employees are increasingly overwhelmed by this and tend to be manual data integration machines rather than productive workers. This is exactly where you can start. Start small but at scale. On the technical side, it is important to think about AI archetypes. Far too often, I see redundant integrations in systems of record and completely different implementation approaches. Both significantly increase operating costs and can, in some cases, even negate the added value of AI.
Start where there is a high volume of business transactions. That is where I see the greatest leverage. Focus on areas where traditional automation has consistently failed, for example, due to unstructured data. But as mentioned above, manual, repetitive processes exist in virtually every department and area, just perhaps not with the same frequency. Classic examples include: offer review, needs assessment and supplier communication, report review, HR case management, IT ticket management, project control and reporting... – I could go on and on. more hands-on. Start fast, fail fast, and keep going until AI is used in all core processes where this technology truly makes sense and helps.
Yes, the moment I realized that AI is not just a tool for efficiency, but actually has a positive impact on us during times of high workload and complexity. I would like to quote from a Microsoft study here: «It’s about time: The Copilot Usage Report (2025): The data suggests that we are not just using AI to do our work faster; we are using it to navigate the complexities of being human, one prompt at a time».
Less academic, more hands-on. Start quickly, fail quickly, and keep going until AI is used in all core processes where this technology truly makes sense and helps. Too often, we get bogged down in technocratic debates over terminology. We talk far too little about the challenges we want to solve.
Thanks to Frederik Jelden for his candid insights, straightforward advice, and perspective drawn from real-world transformation experience.
Did you missed it? No worries. We have recorded them all.