Generative AI Product Management: What Does LLM Product Management Look Like in Day-to-Day Practice? – with Till Scholich (Product @ Ficus Health, ex Stanford Research Assistant)

Generative AI is on everyone’s lips – but how many product people really engage in depth with what it means to build software that is fundamentally based on large language models? Between LinkedIn posts about prompt engineering and conference talks about the future of AI, there’s a gap in the discourse: operational reality. Anyone developing AI-first products today has to deal with topics such as observability, evals, and prompt management – all concepts that don’t exist in classic deterministic software design. It’s high time we talked about them more!

That’s exactly why I’m especially pleased about my conversation with Till Scholich, because Till brings the ideal perspective: on the one hand, he has done academic research on AI, both at Stanford and at the University of Michigan. On the other hand, he currently works as a Senior Product Manager at Ficus Health, where he faces the practical challenges of AI-first product management every day and, because he handles health data, has to meet the highest data privacy standards. His appearances as a speaker – at Berlin’s re:publica conference or the Merantix Developer Day, for example – show that Till is actively shaping the discourse. In short: Till knows what he’s talking about, in theory and in practice.

In the twelfth episode of the PRODUKTKRAFT Podcast, we look together at what is really new in AI product management. We talk about the concepts that are essential today, discuss which data structures are needed, and how AI-native companies differ fundamentally from traditional organizations. But we don’t just talk about tools and techniques: Which traps should you avoid? And above all: How do we remain the pilots, and AI the copilot, of AI software? Have a listen to find out more!