Everyone’s talking about AI in healthcare. But how much of it is real, and how much is still a slide deck?
That was the question at the heart of one of HLTH Europe 2025’s most-watched main stage sessions. Prof. Dr. Heyo K. Kroemer, CEO of Charité – University Medicine Berlin, one of Europe’s largest and most influential hospital systems sat down with Dr. Karen DeSalvo, Chief Health Officer at Google, to cut through the noise.
The conversation was part of a wider programme at HLTH Europe 2025 in Amsterdam, where over 5,000 health innovators, clinicians, investors, and policymakers gathered to explore what’s actually transforming healthcare, not just what’s promised to.
Kroemer was direct: the challenge isn’t finding AI tools that work in controlled conditions. It’s scaling them across complex, resource-constrained hospital environments. At Charité, a 3,000-bed academic medical centre running more than a million patient cases a year, every implementation decision has real consequences.
His view? AI needs to earn its place in clinical workflows. Promising results in a lab setting mean little if they can’t be embedded into the day-to-day reality of a busy ward or emergency department. The technology has to fit the system, not the other way around.
DeSalvo brought a different vantage point. From Google’s position, working at the intersection of data infrastructure, consumer health, and enterprise partnerships, she sees both the scale of what’s possible and the friction points that slow adoption. Data interoperability remains one of the biggest blockers. Health systems are sitting on vast amounts of potentially life-saving information that simply can’t be used effectively because it’s siloed, inconsistently structured, or inaccessible across institutions.
The session wasn’t pessimistic. Far from it. Both speakers pointed to areas where AI is already delivering meaningful impact: predictive analytics flagging deteriorating patients before a crisis hits, AI-assisted diagnostics improving the speed and accuracy of imaging reads, and workflow tools reducing the administrative burden that’s burning out clinical staff across the continent.
DeSalvo highlighted the importance of building AI that clinicians actually trust and that means being transparent about how models make decisions, not just what decisions they make. “Explainability” has moved from a technical nice-to-have to a clinical requirement.
Kroemer spoke about the opportunity ahead with AgentAI – systems that don’t just assist with a single task but can coordinate across multiple clinical workflows autonomously. It’s early, but the direction is clear. The hospital of the future won’t just use AI as a tool. It will be built around it.
The European context matters here. Health systems across the continent face a shared set of pressures: ageing populations, workforce shortages, rising chronic disease burdens, and tightening public budgets. AI isn’t a silver bullet. But deployed thoughtfully, it can help clinicians do more with less and do it better.
The European Health Data Space (EHDS), now moving from policy to implementation, could be a turning point. If it delivers on its promise of cross-border data access and interoperability, it creates the foundation for AI tools to work at a scale that individual health systems simply can’t achieve alone.
In less than 5 weeks, HLTH Europe 2026 returns to Amsterdam, 15-18 June, with AI in healthcare once again a central thread running through the programme. If the conversation at HLTH Europe 2025 was about what’s possible, 2026 will be about what’s next.
Watch the full session here
Make sure you get your ticket here
P.S. Looking at attending HLTH USA 2026? Prices increase on 15th May. Register here before then to lock in the current rate.