Omnicell
Designing for safety-critical
hardware-software
experiences
When a misread screen isn't a UX failure — it's a safety incident, every design decision carries weight. This is four years of that work.
removal workflow
care per nurse per shift
one design system
every decision lives long
The device on every hospital floor.
The software nobody had redesigned in a decade.
Omnicell's Automated Dispensing Cabinets are the locked medication storage units found in every nursing unit of every hospital floor. A large hospital might have 40 of them. They operate like vending machines for controlled substances — every dispense, return, and restock is tracked. The software running these cabinets had been built for a specific device in a specific era, and for a long time that was enough.
When I joined as Product Design Manager for the Point of Care line, Omnicell was betting on the next decade: new hardware, new interaction models, a platform that could scale across dispensing cabinets, kiosks, tablets, and web simultaneously. I led that redesign from field research through shipping — and stayed close enough to the hardest problems to guide quality personally.
The work below is organized around three problems. Each one started as a design problem. Each one turned out to be a safety problem we solved through design.
What made this hard
Most of the difficulty in this project wasn't visible in any single screen. It lived in the organizational and environmental gaps that nobody had named yet — the seam between hardware and software teams, the mismatch between how the system was designed to work and how people actually used it, the assumption that color alone could carry safety information in a clinical context.
I conducted on-site contextual inquiry in live hospital pharmacies: gowned up, standing next to clinical staff, watching how they actually used this hardware. Those observations drove every design decision that followed. We didn't come back from those site visits with a report. We came back with a system.
Three problems. Three safety incidents waiting to happen.
Each story below is a design problem that turned out to be a safety problem. Select one to go deeper into the research, decisions, and outcomes.
Four teams. One device. Nobody owned the seam.
Hardware decisions were being made that would permanently constrain what the interface could do — and nobody had asked what any of it meant for the software experience. I became the person asking those questions, built a service blueprint nobody requested, and extended an existing design system to survive an environment it was never built for.
The system and the user were moving in opposite directions.
The system said: find the medication, put it here. The pharmacy tech had the medication already in hand and needed the system to tell them where to go. A wrong restock isn't a UX failure — it's a medication error. We went back to hospitals with clipboards and stopwatches before we touched a single screen.
Red and green. In a hospital. Nobody asked if that was safe.
Accessibility wasn't a business requirement — there was no ticket, no spec line. Design brought colorblindness expertise to the table because nobody else was going to. What started as a contrast audit opened a much larger conversation about how safety information was being communicated at the most critical moment in the workflow.
We didn't ship feelings.
We shipped time studies.
The efficiency came from eliminating friction in the parts of the workflow that weren't doing safety work — the extra clicks, the redundant screens, the interactions that existed because the system was designed around its own logic instead of the nurse's.
Faster and safer are not in tension here. When you design around how people actually work, in the environment they actually work in, you get both.