Redesigning
Redesigning
Redesigning
an AI Dental
an AI Dental
an AI Dental
Charting Tool
Charting Tool
Charting Tool
for Clinical Reality
for Clinical Reality
for Clinical Reality
Mapped real clinical workflows to ensure the AI actually fit how dental teams operate day-to-day.
Mapped real clinical workflows to ensure the AI actually fit how dental teams operate day-to-day.
Problem
Problem
Problem
ChartAI could generate a note in seconds but had no patient record, no clinical history, and no workflow for the team passing that note down the line.
ChartAI could generate a note in seconds but had no patient record, no clinical history, and no workflow for the team passing that note down the line.
ChartAI could generate a note in seconds but had no patient record, no clinical history, and no workflow for the team passing that note down the line.
Details
Details
Details
Role: Product/UX Designer (solo)
Duration: 12 weeks
Scope: End-to-End Process, Interaction Design, Visual System, React Prototype
Status: Live
Role: Product/UX Designer (solo)
Duration: 12 weeks
Scope: End-to-End Process, Interaction Design, Visual System, React Prototype
Status: Live
Role: Product/UX Designer (solo)
Duration:12 weeks
Scope: End-to-End Process, Interaction Design, Visual System, React Prototype
Status: Live
Tools
Tools
Tools
Highlights
Highlights
Highlights
ChartAI uses ambient voice AI to capture clinical conversations and generate dental notes automatically. The founding team had proven the core technology. What they needed was a product that fit how dental clinics actually work.
ChartAI uses ambient voice AI to capture clinical conversations and generate dental notes automatically. The founding team had proven the core technology. What they needed was a product that fit how dental clinics actually work.
What they needed was a product that fit how dental clinics actually work.
Chapter 1
Chapter 1
Framing
Framing
Overview
Design Brief
Workflow
PROJECT STRUCTURE
End to End Progress
Research findings fed directly into structural decisions without a formal pause. The two core gaps were identified by week three and held without significant change through delivery.
Overview
Design Brief
Workflow
PROJECT STRUCTURE
End to End Progress
Research findings fed directly into structural decisions without a formal pause. The two core gaps were identified by week three and held without significant change through delivery.
Chapter 2
Chapter 2
Chapter 1
Phase 1 - Discover
Phase 1 - Discover
Framing
Interviews
Competitive Analysis
Problem Statement
USER RESEARCH
Semi-structured interview
Four dentists and a dental assistant, each at a different point with the tool. I sat in as part of the team, not a stranger with a clipboard. I went in hunting for interface problems, but the same structural one kept surfacing: the product had no memory of a patient, and no sense of who else was charting.
Interviews
Competitive Analysis
Problem Statement
USER RESEARCH
Semi-structured interview
Four dentists and a dental assistant, each at a different point with the tool. I sat in as part of the team, not a stranger with a clipboard. I went in hunting for interface problems, but the same structural one kept surfacing: the product had no memory of a patient, and no sense of who else was charting.
Overview
Design Brief
Workflow
PROJECT STRUCTURE
End to End Progress
Research findings fed directly into structural decisions without a formal pause. The two core gaps were identified by week three and held without significant change through delivery.
Chapter 2
Phase 1 - Discover
Interviews
Competitive Analysis
Core Problem
USER RESEARCH
Semi-structured interview
Four dentists and a dental assistant, each at a different point with the tool. I sat in as part of the team, not a stranger with a clipboard. I went in hunting for interface problems, but the same structural one kept surfacing: the product had no memory of a patient, and no sense of who else was charting.
Chapter 3
Chapter 3
Phase 2 - Define
Phase 2 - Define
Phase 2 - Define
Interactive Wireframe
Priority Table
System Flow
TESTING THE LOGIC
Lock one piece at a time
I built the structure as a clickable low-fi prototype. Stakeholders had a habit of reacting to everything at once, so I generated a few versions of each screen and locked them one piece at a time: patient identity first, then the note detail, then the referral handoff. By the time color went on, the hard structural calls were already settled.
Interactive Wireframe
Priority Table
System Flow
TESTING THE LOGIC
Lock one piece at a time
I built the structure as a clickable low-fi prototype. Stakeholders had a habit of reacting to everything at once, so I generated a few versions of each screen and locked them one piece at a time: patient identity first, then the note detail, then the referral handoff. By the time color went on, the hard structural calls were already settled.
Interactive Wireframe
Priority Table
System Flow
TESTING THE LOGIC
Lock one piece at a time
I built the structure as a clickable low-fi prototype. Stakeholders had a habit of reacting to everything at once, so I generated a few versions of each screen and locked them one piece at a time: patient identity first, then the note detail, then the referral handoff. By the time color went on, the hard structural calls were already settled.
Chapter 4
Chapter 4
Phase 3 - Develop
Phase 3 - Develop
Phase 3 - Develop
Brand Expression
Moodboard
Component Library
Brand Expression
Moodboard
Component Library
Brand Expression
Moodboard
Component Library
Chapter 5
Chapter 5
Validate
Validate
Validate
RESEARCH
User Testing
Across five dentists, the design was validated as natural and close to their real workflow ("they like how everything is designed, for the most part.") Here are the main takeaways:
Referral generation worked, but the handoff was missing;
Greyed-out appointments were misread as "completed" rather than another provider's;
Dentists wanted the template as a live checklist during the appointment
RESEARCH
User Testing
Referral generation worked, but the handoff was missing;
Greyed-out appointments were misread as "completed" rather than another provider's;
Dentists wanted the template as a live checklist during the appointment
RESEARCH
User Testing
Referral generation worked, but the handoff was missing;
Greyed-out appointments were misread as "completed" rather than another provider's;
Dentists wanted the template as a live checklist during the appointment
Chapter 7
Chapter 7
Before vs. After
Before vs. After
Before vs. After
New Appointment
Recording
Note Detail
New Appointment
Recording
Note Detail
New Appointment
Recording
Note Detail
Chapter 8
Chapter 8
New Features
New Features
New Features
Multiple Operators
Template Library
Referral Letters
New Appointment
Recording
Note Detail
Multiple Operators
Template Library
Referral Letters
(What I learned)
This was my first time designing inside a live AI product, not around one. Three things stuck.
AI doesn't remove design problems, it moves them. The voice model already worked. The hard part was everything around the output: how a dentist reviews it, catches what the AI dropped, and trusts it enough to sign. The intelligence was solved. The workflow around it wasn't. That's where the design lived.
I learned that working with a founder is about being the person who can say "here's a better way" and back it up.
Talking to engineers changed what I designed. Asking engineer "is this hard to build?" before committing saved me from designing things that couldn't ship in nine weeks. Constraints weren't the enemy of the design, they were what made it real.
This was my first time designing inside a live AI product, not around one. Three things stuck.
AI doesn't remove design problems, it moves them. The voice model already worked. The hard part was everything around the output: how a dentist reviews it, catches what the AI dropped, and trusts it enough to sign. The intelligence was solved. The workflow around it wasn't. That's where the design lived.
I learned that working with a founder is about being the person who can say "here's a better way" and back it up.
Talking to engineers changed what I designed. Asking engineer "is this hard to build?" before committing saved me from designing things that couldn't ship in nine weeks. Constraints weren't the enemy of the design, they were what made it real.
This was my first time designing inside a live AI product, not around one. Three things stuck.
AI doesn't remove design problems, it moves them. The voice model already worked. The hard part was everything around the output: how a dentist reviews it, catches what the AI dropped, and trusts it enough to sign. The intelligence was solved. The workflow around it wasn't. That's where the design lived.
I learned that working with a founder is about being the person who can say "here's a better way" and back it up.
Talking to engineers changed what I designed. Asking engineer "is this hard to build?" before committing saved me from designing things that couldn't ship in nine weeks. Constraints weren't the enemy of the design, they were what made it real.
(Project)
Next Project
Reducing peak-time wait from 15+ min through real-time availability maps.
02
/ 05
Redefining Campus E-Bike Charging System
Duration
8 weeks
Role
Product Designer
Services
Rapid Prototyping
User Research
Usability Testing
App Design
Reducing peak-time wait from 15+ min through real-time availability maps.
02
/ 05
Redefining Campus E-Bike Charging System
Duration
8 weeks
Role
Product Designer
Services
Rapid Prototyping
User Research
Usability Testing
App Design
Reducing peak-time wait from 15+ min through real-time availability maps.
02
/ 05
Redefining Campus E-Bike Charging System
Duration
8 weeks
Role
Product Designer
Services
Rapid Prototyping
User Research
Usability Testing
App Design