Fusion Dental
A dental application inside an EHR that had never supported dentistry. It replaced paper charting and helped retain a $10M client.
The Stakes
Clinicians were documenting dental care in software so limited that they had gone back to paper.
It couldn't capture proper odontogram drawings, notes on individual teeth, or periodontal charting, and treatments had no way to tie to a condition, a date, or completion. With high staff turnover, nobody could follow who had documented what, or why. A key client's contract was going through a competitive RFP, with $10M in annual revenue on the line.
What I Owned
Product direction and design: End to end, as Lead Product Designer.
PM work: Alongside our one PM: requirements, user stories, acceptance criteria, and feature prioritization.
The phased roadmap: What shipped first and what waited.
Constraints
No in-house dental expertise: With no dental SME on staff, I gathered market research, notes, and conversations with outside stakeholders who had dental experience, and compiled it all in Claude to fact-check, study how competitors handled complex workflows, and build requirements. I drafted workflows with Claude before design work started.
No direct observation: Correctional restrictions ruled it out, so I met with stakeholders weekly to surface limits I couldn't see, from missing equipment to workflows unique to correctional settings.
High turnover: The interface had to make sense to someone on their first day.
Decision 01
Charting at the Center
The odontogram became the center of the application rather than one view among many, because charting is the task clinicians return to most. Tying conditions, dates, and completion status directly to a charted tooth closed the gap that had been sending staff back to paper. We considered putting all of a patient's dental data on one page, but that meant information overload and constant scrolling. Tabs let clinicians switch context quickly, while charting stays the anchor.
Before: the software clinicians had
After: charting at the center
Decision 02
Built for Someone's First Day
With constant turnover, the interface couldn't depend on long-tenured users who know where things hide. Common conditions and treatments live in a quick actions toolbar on the surface, not buried in menus.
The toolbar is configurable. An organization chooses which conditions and treatments appear for everyone, and individuals can personalize theirs or opt back into the organization's setup.
Every state was designed before handoff. Drag and drop shows a focus border on grab, a line where the tile will land, and the settled order. The full spec also covers an arrow-button alternative, filters, empty and full states, and save and error messages.
Decision 03
Ship the Core, Phase the Rest
The first release focused on charting. Three things deliberately waited:
Toolbar customization: It needed a settings page, so it came later in two steps: organization-level customization in phase 2, then user-level in phase 3, letting each clinician choose the conditions and treatments relevant to their practice.
Image viewing: Most clients already viewed X-rays, photos, and documents in other systems. In phase 3 we added a file viewer that connects to their X-ray viewers and lets staff upload documents and photos to individual charting entries.
Exporting: Exports would contain sensitive PHI, which meant building new permissions first. We didn't want every user to be able to export.
.png)
Designed now, shipped later. Image viewing was designed alongside the core and held for phase 3: a Files tab where clinicians open, rename, and archive images and attachments without leaving the chart.
Outcome
A net-new dental application inside an existing healthcare ecosystem, and the design work that retained a key client through a competitive RFP, preserving $10M in annual revenue.
Fusion Dental was also the first complete application built on the Fusion Design System.
What I took from this
Sometimes there is little reference for the product you are building. When precedent doesn't exist, the people who will use it become the reference, and extensive research with them is the only way in.
Optimization matters most in the workflows people repeat. Frequently used paths deserve disproportionate attention, because small friction compounds every time.
What you leave out of a first release is a product decision, not a delay. Each cut needs a reason, and a plan for when it comes back.
Next project
↑ Back to top