Enterprise AI architecture considerations

Dr. John Brownstein has been on the leading edge of digital health since medical school, where he was drawn to epidemiology because of the opportunity to use data sets for large-scale impact beyond the exam room. He’s been on a hunt for data ever since, shaping a career that has spanned public health, technology, startups, and healthcare innovation.

Dr. Brownstein helped develop Google Flu Trends, served as a healthcare advisor to Uber, and launched one of the first data analytics companies in public health, where he applied AI long before it became today’s dominant topic. In 2015, he became Chief Innovation Officer at Boston Children’s Hospital, where he has helped turn it into a proving ground for emerging technologies.

With its manageable size, nimble culture, and an administration willing to invest in innovation, Dr. Brownstein has helped turn Boston Children’s into an ideal setting for testing emerging technology. This environment, which put Boston Children’s on the front edge of healthcare IT for more than a decade, has positioned it to help lead the current AI wave.

In this episode of Healthcare is Hard, Dr. Brownstein joined Keith Figlioli to discuss AI adoption, how it’s changing health systems, and what those changes mean for the companies selling to them. They discussed topics including:

• The fast pace from pilots to production. Only half joking, Dr. Brownstein mentioned that his conversation with Keith would be outdated by the time the podcast aired. The reality is, everything related to AI is moving so quick – even in healthcare, which is bucking its reputation as a slow mover on tech adoption when it comes to AI. Health systems are no longer just dabbling in isolated pilots; they’re taking significant steps toward broad deployment, using AI for intelligent automation, workflow support, and clinical tools in ways that are beginning to produce measurable impact.
• The build versus buy equation. After launching Boston Children’s innovation program, Dr. Brownstein recognized that having a large engineering team wasn’t sustainable and began turning to startups that could innovate and deploy quickly. But that tide is shifting again, with the ability to use AI for internal development. He hasn’t closed the door on startups, but says the bar is much higher. Companies selling into providers will need to show a stronger moat – whether through proprietary data, regulatory expertise, deep workflow integration, or some other advantage that cannot be easily recreated in-house.
• Designing AI architecture. AI procurement has become much more strategic than investing in point solutions. Dr. Brownstein discussed how health systems need to think about foundation models, hyperscalers, core systems like Epic, and point solutions, and how they all fit together to benefit users including clinicians, researchers, patients and families. For startups and incumbents alike, success will depend not only on product performance, but on how well a solution fits into the broader technology stack health systems are now building around AI.
• Why culture is just as important as capability. Even with strong interest in AI across healthcare, adoption is not purely a technical issue. Dr, Brownstein notes that while demand for AI tools is high, there are also signs of fatigue and anxiety as AI becomes an unavoidable topic in boardrooms, leadership meetings, and day-to-day work. With real questions about trust and job impact, organizations will need to pair technical progress with a thoughtful approach to culture, communication, and workforce readiness.

To hear Keith Figlioli and Dr. John Brownstein discuss these topics and more, listen to this episode of Healthcare is Hard: A Podcast for Insiders.