Giocare
Redesign of the nutrologists' system: from a flat table to the protocol that organizes the appointment.
Paula La Rosa, Product & Service Designer and Design Engineer
- Sector
- HealthTech · B2B SaaS
- Role
- Product Design · Design Engineering
- Year
- 2026
- Stack
- Fraunces + Schibsted Grotesk + Spline Sans Mono, Tokens próprios, GSAP, Next.js · Supabase (produto)
- Live
- giocare.app
Context
Giocare is the system for the nutrology doctors of Nutrology Academy: it transcribes the appointment, builds the clinical record with the ABCDEFS protocol and also handles schedule, exams and finances. I redesigned the visual direction and the screens from the real product. The direction started as a preview and now lives in the product itself, on the landing page, pricing and the Founding Members program.
Problem
The product had a strong core, AI transcription and structured records, stuck in a tired shell. Everything became a giant flat table with no filter or sorting, more than 450 appointments in a single list. The finance month filter did not filter and dark mode existed only in part of the dashboard. And since the AI suggests clinical conduct, the interface needed trust, not more crowded screens.
Solution
The first direction, in lavender and plum, was read by the Academy as too feminine. I rebuilt it from the Academy's own brand: its red with the navy of the crest, on a warm paper, and the red to coral gradient only where it matters. Internal screens swap the flat table for lists that prioritize and segment. And the AI flow gets trust rules: consented recording, a generated record as a draft to review, and the conduct step freezes the anamnesis so nothing leaks between phases.
What the research showed
How it was done. Observation of appointments with nutrology doctors using the system, interviews about their scheduling and billing routine and a UX laws study measured on the phone. The first visual direction was also reviewed by the Academy, and the rejection became research data.
Findings
- Doctors do not read tables, they look for who needs attention: In the daily routine, the question is who has an overdue follow-up or a pending exam. A flat list with hundreds of rows hid exactly that.
- Clinical AI is only accepted as a draft: Doctors accept help with transcription, but want to review before signing. A record that looked too finished bred distrust, not time savings.
- The house brand builds trust: A new palette, however beautiful, looked like another company's product. The Academy's color made the system feel like home.
Personas
Dr. Renata, 44, Nutrologist with her own practice. Sees patients all day and fills in records between one patient and the next. Needs: Finish the appointment with the record ready to review, without typing everything again. Gets in the way: A huge unfiltered list and a record the AI fills in as if it were final.
Dr. André, 36, Nutrologist trained by the Academy. Is building his patient base and handles scheduling and billing on his own. Needs: Know who needs a follow-up and how much he will receive this month. Gets in the way: A month filter that does not filter and screens that break in dark mode.
Personas composed from the research. Names and ages are illustrative.
What was left out
I discarded the first direction, in lavender and plum, after the Academy's feedback. I also discarded the idea of the AI filling in the treatment plan on its own: the plan stays with the doctor.
How to know it worked
- How long it takes from the end of the appointment to a reviewed, signed record.
- How many AI-generated records are edited before signing.
- How many overdue follow-ups show up in the list without a manual search.
Design decisions
The brand came from the Academy
After the first palette was rejected, I stopped inventing colors and went to the source: the red of Nutrology Academy and the navy of its crest. The system now feels like the house the doctor already knows.
Gradient in moderation
Red to coral shows up on calls to action, states and highlights, never as a background. I measured how many times coral appears per screen and brought it down to one, so it means something again.
From a flat table to a list that prioritizes
The app's biggest debt was giant lists with no filter, sorting or segmentation. I reorganized patients and appointments around who needs attention now, instead of dumping everything at once.
AI with consent and a way back
Transcription is explicit and authorized. The ABCDEFS record the AI builds arrives as a draft to review and edit, never as ready truth. And when conduct opens, the anamnesis freezes to keep the record intact.
Results
- From preview to product: The direction left the showcase and lives on giocare.app, with landing page, pricing and Founding Members sign-up.
- A structure that handles volume: Lists that prioritize and segment instead of the flat table. What choked at 450 rows makes sense at any size.
- Touch that fits the finger: In a UX laws study measured on mobile, targets smaller than 44 px fell from 60% to 5% of screens.