.png)
Turning deep science into a globally-scaled health technology is not about hype — it’s about disciplined execution.
This interview distills the top lessons from Volpara founder Ralph Highnam in conversation with **Angus Blair** (GP, Outset Ventures) focused on what deep-tech founders must get right: data standardisation, early clinical validation, go-to-market strategy, regulatory sequencing, capital strategy, and sustainable engineering culture.
These insights apply well beyond breast imaging; they are a practical blueprint for anyone building in medtech, diagnostics, AI-in-healthcare, or research-driven startups.
:angus-2: Angus:
Where did the original idea come from, and what made you commit to actually building something around it?
:wfw-case-study-icons--7-: Ralph:
“I was very good at maths and computing and stuff like that and I wanted to do a PhD because I didn't know what else that I wanted to do. So I thought, I'll do a PhD. But I didn't really even know what research was back then. I was poking around. There was all kinds of boring crap. Then I met this guy, Professor Sir Mike Brady, and he said his mother-in-law had just been diagnosed with late-stage breast cancer despite being screened, and he suggested we could improve screening using artificial intelligence. That kind of conversation sparked me. It was like, okay, this uses all my maths and computing, and I get to learn all about the breast — how the breast works, physiology, breast cancer. We tried to commercialise it in 1999 and it was just too early, but we returned to it in 2009 when we felt we had a duty to bring it to market due to clinical evidence building up. We choose the start-up route because you need complete belief to take it forward.”
Takeaway:
The technical spark only matters if someone decides it must exist in the world — and is willing to carry it.
:angus-2: Angus:
How did you get from the idea to something validated enough for FDA and early customers?
:wfw-case-study-icons--7-: Ralph:
“We needed to come up an AI algorithm for breast density measurement from X rays. We needed to start testing it and because of our backgrounds we decided early on to get the algorithm out for independent validation by world-leading researchers. There’s risk involved in that, early algorithms often have issues, but with trusted but independent partners you can iterate the algorithm quickly to get to something FDA-ready. You have to really curtail that perfectionist angle.”
Takeaway:
External partners will break your assumptions faster than internal experiments. That’s the point.
:angus-2: Angus:
What shifted once you started seeing real clinic usage?
:wfw-case-study-icons--7-: Ralph:
“We installed it into several beta sites. Some of them complete disasters (for various reasons). Most of them were fantastic. We learned a lot. We learned customer service. They said, ‘Don’t take it out. We love it.’ They said, ‘With your scorecards, we can talk to the women while they’re still at the clinic.’”
Takeaway:
Real workflow is the fastest teacher. It shows you what actually drives clinical and economic value.
:angus-2: Angus:
You’ve said distribution mistakes are common. What’s the core rule?
:wfw-case-study-icons--7-: Ralph:
“Never, ever go exclusive. Always have the ability to sell direct. They wanted exclusivity. We said no way. Distributors aren’t going to be able to sell early deep-tech, they like an easy-life. You have to get out there and do it yourself.”
Takeaway:
Education-heavy, category-creating products cannot be handed off to a channel. You need your own voice & ears in the market.