0→1 Product Design · Wix & Base44 · 2025
Reducing clinician documentation time through real-time, AI-powered capture
I led the product direction, UX strategy, and technical implementation of an AI-powered clinical documentation system — taking it from research to a fully functional prototype within a 4-week sprint.
This project won the AI Product Sprint and continued development with Wix and Base44 in NYC.
Product Overview
My Role & Leadership
I emerged as the de facto team lead, owning the core product decisions and execution across design and development.
While supported by a strong team, I was ultimately responsible for turning our ideas into a viable, testable product and ensuring we delivered at a high level.
Product Direction
Defined the core solution and interaction model, aligning the team around a focused approach
UX Strategy
Designed the primary workflow for capturing and structuring patient data in real-world conditions
Technical Execution
Built the full working prototype in Base44, including LLM integration, transcription, and data handling
Process Design
Established how design and development worked together, bridging Figma exploration with live implementation
Key Research Insights
We found that clinicians routinely rely on ‘shadow practices’—informal handwritten notes and delayed data entry from memory—to manage the weight of documentation.
34–55%
of clinician time is spent on documentation, taking precious time away from patient care.
“During busy shifts, I'll jot notes on scraps of paper or even on my glove, just so I don't lose the details before I can update Epic.”
— PICU Nurse
Goal:
Reduce doctors' and nurses' workload by creating an app to capture and summarize medical information automatically using AI.
Defining the Core Interaction
The core challenge was flexibility: clinicians need to capture information quickly, in any format, without breaking their workflow.
I designed a system that allows users to:
- Select a patient first, then capture input
- Or capture input first, and assign it afterward
Result
At the center of this is a persistent input dock — giving immediate access to audio, photo, and text. This reduced friction while maintaining structured, assignable data.
Design & Vibe Coding
I translated the product into a working system by building the prototype directly in Base44.
Within one day of receiving access to the tools, I developed a functional prototype — allowing the team to focus on refining UX rather than building from scratch.
- Integrated LLM-based summarization
- Implemented audio transcription and multimodal inputs
- Built a simulated medical record system for testing
Result
This enabled us to validate the experience through a fully interactive product.
Explore the full Ascribe app
Audio input
Photo input
Text input
Usability Testing
To evaluate Ascribe's performance, we conducted a comprehensive usability test with a registered nurse from the NY Presbyterian Hospital ICU.
- Camera capture was fully accurate (5/5) — significantly faster than manual entry.
- AI summaries rated 10/10 — output was fully accurate and used appropriate medical jargon and abbreviations to match typical manual documentation.
- Overall satisfaction 8/10 — Ascribe felt ready for real-world use with minor UI refinements.
- Recommended improvements: larger display of Patient I.D. Number, and a clearer confirmation system after sending notes to the medical record.
Usability test session
Outcome & Reflection
We won the AI Product Sprint, with judges highlighting the strength of the problem-solution alignment and overall execution.
This project demonstrates my ability to take an ambiguous problem and turn it into a working product — by connecting research, UX strategy, and technical implementation into a cohesive system.
At the Wix Offices in NYC with my team. That's me in the middle :)