Beyond the Hype: Building Patient-Centric AI That Works
Co-authored by By Claudio Silvestrin
February 3rd, 2026
Walk into any clinic or open your laptop, and you’ll find people using AI in healthcare, but not in the way you might think. Millions are asking ChatGPT about their symptoms, comparing answers, and seeking advice. On the surface, it feels like AI is already embedded in daily healthcare. In reality, most hospitals and pharma companies are still struggling to integrate AI into even basic workflows. The tension between perception and reality defines where we are today.
Claudio Silvestrin framed it well: “There’s a bit of a discrepancy between the general perception of the power of AI, people talking about artificial general intelligence, versus how even big companies with lots of funding struggle with using AI effectively, even on smaller problems.”
The problem is not a lack of technology. Generative AI, natural language processing, predictive analytics, the tools are here. The barrier lies in how these systems are deployed. AI in practice is less about algorithms and more about integration: product design, usability, and above all, collaboration with end users. As Claudio emphasised: “Deploying AI effectively is also a software development and deployment problem related to making products usable. It involves a lot of handholding, and that takes time and people.”
This matters because healthcare’s challenges are not abstract; they are lived every day by patients. Take the explosion of new weight-loss drugs. Sales are skyrocketing, yet real-world use has raised concerns: muscle loss, reduced bone density, uncertain long-term outcomes. Patients often stop treatment prematurely, leading to rebound weight gain. Regulators are scrambling to catch up. In this space, AI could be transformative, not by predicting the next molecule in the lab, but by monitoring patients in the real world.
Claudio pointed to this opportunity: “For a certain condition with certain drugs, you could have an intelligent agent or partner for the patient that helps them through their journey. GPT is probably already being used this way, but it would be much better to have a proper healthcare solution that is certified and built for this task.”
Imagine post-market surveillance powered by AI: systems that track adherence, side effects, lifestyle changes, and long-term outcomes across millions of users. Imagine decentralised clinical trials where wearable sensors and digital tools provide a continuous stream of data, analysed in real time. Such systems could offer insights not just to pharmaceutical companies but to payers, regulators, and, most importantly — patients themselves.
But for startups and entrepreneurs eyeing this space, Claudio’s advice is pointed: “Don’t focus so much on having the best algorithm, but think about how you’ll work most effectively with end users. Where are the low-hanging fruits? Where can you create the most value? And how do you make sure the product actually works and can be maintained over time?”
AI will not transform healthcare overnight. The path forward is slower, more incremental, and more collaborative than the headlines suggest. Yet that is no reason for cynicism. With each step, an error is caught, a treatment plan improved, a patient better supported, we move closer to a future where AI genuinely improves lives. That future will be built not on hype, but on patient-centered solutions that work in practice. And as Claudio reminded us, “There’s reason for optimism. We shouldn’t forget about the huge potential AI has, because I’ve seen what it can do.”
