A Rosie Place for Children is a 10-bed respite care facility in Indiana. The children who stay here are medically fragile. Many are on life-saving equipment. The east side of the building runs on a backup generator because the power cannot go out on those kids. Bryan and I drove down for a three-hour discovery session with Donna, the Director of Nursing, and Kathy, the Nurse Coordinator. The goal was to understand every workflow well enough to replace paper charting with a proprietary EMR.
But before we sat down with any paperwork, Teal, the founder, walked us through the building.
The rooms have names
Desert Dreams. Sunny Meadow. Barefoot Beach. The Waterfall. The Pond. Every room has a name because children learn language before numbers. A child picks their own room. They have sleepovers with flashlights. Nurses follow the kids around, not the other way around. The whole place is designed to feel like staying at an auntie's house.
I kept looking for the hospital. I never found it. The medical infrastructure is there, obviously -- backup generators, crash carts, medication storage, nursing stations. But it's all folded into a building that feels like a home. That dissonance stuck with me through the rest of the day. The care here is serious. The environment refuses to act like it.
The paper trail
Then we sat down, and the dissonance flipped. This warm, intentional place runs its medical documentation on a system that would frustrate anyone.
I walked through each checkpoint the way I approach any system design: whose pain is this? At every step, I wanted to know who carries the burden. The answer was almost always Kathy or Donna.
A single child's chart takes Kathy three to six hours to build. The same data gets transcribed four or five times across Google Forms, handwritten notes, Excel medication orders, printed MARs, and faxed doctor approvals. Donna reviews every line after Kathy builds it. Then it's faxed to the physician. Then reviewed again. Then printed. Then assembled into the physical chart. If a parent mentions a medication change on admission day, the entire addendum cycle restarts from the beginning.
Donna said something I haven't been able to shake: "As long as we have paper charting, we are subject to human error... We spend more time with our heads in the chart than what we should be spending time with the kids."
That sentence is the project brief. Everything else is details.
What Teal actually asked for
When I asked Teal what she wanted from the new system, her specific request was voice dictation per field. A nurse standing at a child's bedside should be able to speak her notes instead of writing them. That's a real feature, and we'll build it.
But the deeper ask was something she didn't have to say out loud. Give my nurses back to the children. Every hour Kathy spends building a chart is an hour she's not on the floor. Every time Donna reviews handwritten medication orders with a magnifying glass -- and I'm not exaggerating, she literally uses a magnifying glass -- that's attention pulled away from the kids who need it.
The system we're building won't be a SaaS product. Rosie's Place will own it outright. The data stays on-site -- they already run Raspberry Pi servers for other infrastructure. This is a bespoke EMR for a one-of-a-kind facility. Ten beds. Named rooms. Children who can't advocate for themselves. The architecture has to match the intimacy of the care.
What I'm carrying out of that building
I've built automation systems, landing pages, audit tools, brand collateral generators. This is different. The gap between the current system and the one we're going to build is measured in hours that nurses get back. Hours that go directly to children who are, in many cases, spending their only nights away from a hospital in a place that was built specifically so they wouldn't feel like they were in one.
The first screen we design will be the admission form. One entry point. Data flows forward. No re-transcription. Physician approval routed digitally. MARs generated automatically. Chart assembly eliminated because the chart is the system.
Kathy's three-to-six-hour chart build should take under thirty minutes. That's the target. Everything we design gets measured against it.