End-to-end visual system for a real-time speech-to-text nursing documentation platform, crafted for two user groups across web and mobile.
Product
Autochart (AI-powered nursing documentation)
My Role
UX, Visual Design
Platform
Web & Mobile (Responsive)
Deliverable
Visual Design, Light & dark themes, Interactive Figma prototype
Overview
Nursing documentation is one of healthcare’s most persistent inefficiencies. Nurses spend up to 35–40% of their shift is spent on manual charting, time taken directly from patient care. The cognitive load of remembering and re-entering observations after the fact leads to errors, omissions, and burnout.
Autochart is built to solve this with a voice-first AI approach: nurses speak naturally during or after a patient interaction, and the platform uses LLM-powered transcription and structuring to generate accurate clinical notes automatically.
My role was to design the complete visual system, light and dark mode, for both the nurse-facing app and the admin interface used to manage nurses, patients, and assignments.
The Impact
- 1–4 hours saved per nurse daily
- ↓ 60–75% documentation time · ↓ 40% handover time
- Improved clinician productivity and reduced burnout
The Problem
Nurses spend 1–4 hours daily on manual documentation, leading to:
- Increased cognitive load
- Burnout and reduced job satisfaction
- Risk of incomplete or inaccurate records
- Less time for direct patient care
Existing EHR systems are:
- Rigid and form-heavy
- Not optimized for real-time bedside workflows
- Poorly integrated with natural human interaction (speech)
Opportunity
What if documentation didn’t require typing at all?
What if nurses could simply speak naturally, and the system understood, structured, and documented it?
This led to a key design question:
How might we transform unstructured human conversation into structured clinical data without
increasing cognitive load or compromising trust?
Solution
Autochart introduces a voice-first documentation workflow powered by AI:
- Converts speech into structured medical data
- Classifies information into 321 nursing fields
- Automatically populates EHR systems
- Shifts effort from: data entry → intelligent review
Instead of filling forms, nurses now: Speak → Review → Submit
Design Challenges
Conversational AI in Healthcare
The hardest design challenge wasn’t the screens; it was the invisible conversation.
How do you show a user that the AI is listening, processing, understanding, and confident?
Every state needed a visual answer.
The design had to make AI feel accurate, reviewable, and within the nurse’s control at every step.
Designing for two different user groups
Nurses — Primary Users
Works across ICU, wards, and emergency. Uses the product primarily on a tablet or phone during or after patient interactions. Needs speed, clarity, and minimal interruption to their care workflow.
Design focus
- Minimal interaction required during recording
- Clear, scannable note review interface
- Quick error correction and inline validation
- Reduced cognitive load through smart grouping and defaults
- Large touch targets for one-handed bedside use
- Dark mode as the primary experience
Admin — Manage Nurses & Patients
Desk-based. Uses web interface during business hours. Needs data visibility, management controls, and the ability to assign nurses to patients and oversee documentation completion at a glance.
Design focus
- Nurse and patient management at a glance
- Configurable assessment fields
- Information-dense but scannable dashboard layout
- Status-coded visibility across all active documentation
- Light mode as the primary experience
Final Experience for Nurses
The nurse-facing experience was designed around the voice to note workflow, moving from idle state to active recording, AI processing, review, and final submission.
Visual hierarchy and color are used to clearly communicate system states, guiding nurses through each step without relying on instructions, enabling fast and intuitive interaction in real-time care environments.
Final Experience for Admin
The admin portal covers the full operational surface; dashboard overview, patient records, nurse management, and configurable settings. Every screen was designed responsively across desktop breakpoints.
Nurse Management
Patient Management
What Was Delivered
- Admin portal — visual design across all screens, desktop and responsive
- Nurse-facing screens — end-to-end voice-to-note UX and visual design
- Dual-mode design system — light and dark, component-level
- Interactive Figma prototype — nurse and admin flows across both themes
- Production-ready designs for development handoff
The design had to make AI feel accurate, reviewable, and within the nurse’s control at every step.
The Impact
1–4 hours saved per nurse daily
↓ 60–75% documentation time
↓ 40% handover timeImproved documentation accuracy and completeness
Designing the layer between AI and human trust
In healthcare AI, the challenge is not just building accurate systems, but designing how those systems are experienced and trusted.
The critical layer lies between what the AI generates and what the clinician chooses to rely on. If that layer lacks clarity or control, the product risks being ignored regardless of its underlying capability.
The design focused on making AI outputs clear, verifiable, and easy to act on, allowing nurses to move through the workflow with confidence. The goal was not to highlight the intelligence of the system, but to make it feel predictable, supportive, and unobtrusive in real-world use.