Understanding Why AI Innovations Struggle to Scale in Healthcare with Adam Brickman

One of the biggest challenges in tech transfer isn't generating innovation — it's helping promising technologies move from early success into sustained, real-world use. That pattern shows up across industries, but today we're going to explore it through one fast-moving example: AI in healthcare. My guest is Adam Brickman, a healthcare innovation leader and part of the team behind Vega Health, a company focused on helping organizations identify, implement, and scale validated AI solutions.

Adam brings a practitioner's perspective to a problem that's becoming harder to ignore. Technologies that show real promise, sometimes even strong clinical results, can still end up stuck at their site of origin, never reaching the patients and health systems that need them most. Vega Health was built to change that by creating a new commercialization pathway that connects proven AI models from leading academic medical centers and health systems with the community hospitals that make up the vast majority of healthcare in this country.

We discuss why AI that works at one institution doesn't automatically translate somewhere new, and what it actually takes to bridge that gap. We talk about workflow discovery, the importance of testing models against local patient data before full deployment, and why user experience and staff buy-in are just as critical as the technology itself. Adam also shares what Vega Health looks for when evaluating whether an AI solution is ready to scale and has some pointed thoughts for tech transfer offices on licensing strategy in an increasingly crowded market.

In This Episode:

[02:29] Adam describes why many AI innovations remain trapped at their site of origin, even after demonstrating strong clinical or operational results.

[03:10] The conversation breaks down four traditional commercialization paths and introduces Vega Health’s role as a fifth, scale-focused alternative.

[04:05] A common assumption is challenged: the belief that only large academic medical centers can access or afford high-quality AI solutions.

[04:48] Adam explains why success in one health system rarely translates directly, emphasizing that implementation context and workflow differences are critical.

[05:32] Vega Health’s approach is outlined, including retrospective data testing to determine which models perform best in a specific patient population.

[06:40] The typical AI purchasing process is critiqued, highlighting the risks of committing to full deployment before validating real-world performance.

[07:31] The shift from “technology that works” to “technology that is used daily” is framed as a human and organizational challenge, not just a technical one.

[08:12] Adam stresses that technology must adapt to clinicians and staff workflows rather than expecting already-burdened users to change behavior.

[09:05] Validation is defined through live clinical deployment combined with peer-reviewed evidence, reducing the risks of first-time real-world testing.

[10:18] Transparency gaps in AI documentation are addressed, with Vega Health advocating standardized reporting on training data, origins, and performance.

[12:02] Adam reflects on the disconnect between innovation teams solving local problems and vendors pursuing only the most prestigious institutions.

[13:15] The imbalance in vendor strategy is highlighted, noting that most AI companies target a small percentage of elite hospitals while community systems remain underserved.

[14:10] Non-technical barriers take center stage, including alert fatigue, workflow friction, and the outsized importance of thoughtful UI and UX design.

[18:18] A story of initial resistance illustrates how skepticism can soften when end users feel heard through collaborative workflow discovery.

[20:31] Evaluation expands beyond model accuracy to include adoption metrics, clinical outcomes, administrative impact, and measurable return on investment.

[22:23] Adam offers strategic guidance to tech transfer offices: determine whether an innovation stands alone as a company or functions better as a feature.

[24:40] The risks of mandatory exclusivity are discussed, especially in a rapidly crowding AI market likely to experience consolidation.

[26:05] The episode closes with a reflection on why scaling innovation is difficult, resource-intensive, and still deeply worth pursuing.

Resources:

AUTM

Adam Brickman - LinkedIn

Vega Health