The episode features Steve Worrell, CEO of Riverrain, and centers on what it takes to build and run a profitable AI company in radiology. The conversation starts with Steve’s unusual path: he came from a technical and machine-learning background (including military research) and was doing “AI before AI was a thing.” He joined a breast imaging startup in the late 1990s (Quality Computing, later CADx, later acquired by iCAD), and that early CAD era shaped how he thinks about real-world adoption - not just model performance.
Steve explains that his move to Riverrain wasn’t primarily about “lung” at the time, but about staying in healthcare and joining a locally committed group rebuilding R&D after leadership churn elsewhere. He also emphasizes the value of a tenured core team - including several people he worked with previously at iCAD - because early-stage companies can’t afford constant hiring gambles. From his perspective, hiring is one of the hardest parts of scaling, and non-rigorous interview processes (especially in technical roles) lead to expensive mistakes later.
A major theme is what actually gets adopted in radiology AI: it’s never just accuracy - it’s accuracy plus workflow. Steve argues a product can be technically superior, but if it’s clumsy, adds clicks, or doesn’t integrate seamlessly, adoption will stall. He uses Riverrain’s experience with bone suppression and vessel suppression as examples of tools that resonated because they’re visual, intuitive, and fit naturally into a radiologist’s reading process (as “just another series/image”), while also reducing the “black box” feeling by making it clearer why the system is flagging something.
On product strategy, Steve describes relying heavily on clinical advisors to validate ideas early, then being willing to kill projects when the evidence doesn’t justify continuing. He gives an example of developing a pulmonary embolism detector through a clinical trial, but shelving it because performance wasn’t compelling enough and the market was already mature - a case where pushing forward would have been “throwing good money after bad.” He also critiques a common industry mistake: going too broad too fast, underestimating how messy and non-standardized healthcare data is across systems, modalities, and sites.
The conversation then turns to why Riverrain has stayed profitable while many AI vendors struggle. Steve says a great product is necessary but not sufficient - you also have to respect long enterprise sales cycles, run lean, and hire at a pace the organization can actually onboard and support. Riverrain is intentionally small (around the mid-30s headcount) with strong revenue per employee and sustained growth, and Steve prefers the risk of being conservative over burning tens of millions annually hoping to “change the sales cycle.” Looking ahead, Riverrain plans to stay focused on the chest (lung and heart), expanding breadth across use cases (for example nodules, calcium scoring, emphysema, aortic and skeletal health) while going deeper where it matters (visualization to detection to characterization and diagnosis), leaning on partnerships rather than trying to be everything. Steve’s closing advice to founders is consistent: focus, live within your means, respect the market’s inertia, don’t scale ahead of reality, and build products people will actually buy - not products that merely impress investors.