Intuitive's former digital chief on what product developers need to know about AI and customer research
Brian Miller spent years as executive vice president and chief digital officer at Intuitive, the surgical robotics company. Now CEO of Sovato, a telesurgery platform, he offers product developers practical advice on customer observation, AI implementation, and building digital expertise.
Watch customers, not what they say
Device developers often ask customers what they need and get a list of nine things. Miller's method: watch what they actually do.
"I have spent my career in operating rooms for hundreds and hundreds of procedures, watching what happens, where people are struggling," he said. "The only way a product or a solution is successful is by deeply understanding those needs."
Miller records procedures with site approval to catch mannerisms and friction points. He also pulls data directly from devices - pauses in procedures, robot motions, and interaction patterns reveal where workflows break down.
The key is scale. Asking a few people doesn't work. You have to be there repeatedly and live the problem.
Align on value before asking for data
Device makers often struggle to access clinical data from hospital partners. Miller's approach: start by agreeing on what question you're trying to answer.
"Be clear about what you're trying to prove," he said. "If you can align with collaborators and say, 'We believe that if we can answer this question, there's something we can do that's of value,' once you align those elements between the med device companies and the customers, it gets much, much easier."
Walking in and asking for all available data creates resistance. Walking in with a specific hypothesis opens doors.
AI in surgery isn't new, but the tools are changing
Medical device makers have used machine learning for years - analyzing data, enabling remote servicing, studying surgical video. That work continues.
What's different now is the convergence of three things: robotics for interventions, AI (especially multimodal and generative models), and high-speed secure networks. That combination is what enables clinical decision support and eventually some level of automated motion.
"We are definitely at a point where there is a critical convergence" of these three elements, Miller said. "That's the triad of things that are going to be important."
Memory shortages affecting data centers haven't slowed medical device development. Developers can prototype on the cloud, then optimize for edge computing when moving to real-time applications.
Use AI coding assistants as partners, not oracles
Tools like Claude can improve code quality and catch errors, but only if used correctly. Miller treats them as colleagues you sometimes listen to and sometimes don't.
"If you're using it as that, it can be wildly powerful in your ability to go in, test things much more thoroughly, run through different scenarios," he said. "Be very diligent about how you use it. Don't just assume that it's right."
Used as a true assistant, these tools can deliver higher quality code and let developers run scenarios faster. The risk comes from blind trust.
For developers looking to build these skills, AI Coding Courses offer structured training in using AI assistants responsibly in development workflows.
Consider deskilling when AI does the work
As AI takes over certain tasks, professionals risk losing skills they might need if the technology fails. Miller points to airline pilots, who are required to manually land periodically so they don't lose the ability.
This matters less when AI aids human precision - a surgeon using AI-guided tools is still engaged. It matters more when AI does something "because it's more efficient or easy," and the person never practices the underlying skill.
"People should keep in mind that second category and understand the potential dynamics around that," Miller said. Device makers should think through these scenarios during product design and clinical trials.
Build data infrastructure first
Medical device companies adding digital capabilities should start with a strong foundation in data governance and access rights.
"Make sure you have a strong handle on data use rights and how you are going to leverage that in the right way," Miller said. "Spend the time on the platform and the foundation to make sure it's right."
Once that foundation is solid, developers have an easier path to build applications and analyze how products perform in the field. Data access shows product value and informs new development - but only if handled responsibly.
For teams building these capabilities, AI for Product Development courses cover data strategy and digital infrastructure planning.
Engineering mindset transfers to leadership
Miller's background in engineering prepared him for the CEO role. The critical thinking and scenario analysis required in product design applies directly to running a business.
"There's different inputs and different dynamics, but problem solving, critical thinking and making sure you're looking at all scenarios is the best way to make sure you're prepared," he said.
At technical companies, leaders don't need to be deep in every technology. But understanding the technical fundamentals helps you assess risk when making strategic moves.
Your membership also unlocks:
