AI-Powered Medical Documentation
Luma is a HIPAA-compliant AI platform that generates medical necessity documentation in seconds, helping healthcare providers secure prior authorizations faster while protecting patient data.
The prior authorization process is broken. Physicians spend 12-16 hours weekly fighting paperwork instead of treating patients. Meanwhile, patients wait weeks for approvals due to incomplete documentation, and practices lose millions to audit clawbacks from insufficient records.
Healthcare providers know they need AI to stay competitive, but the fear of HIPAA violations creates paralysis. The challenge: How do you design an AI tool that feels safe enough to actually use with patient information?
Key themes from physician and healthcare staff interviews
Instead of asking users to trust that the system is HIPAA-compliant, I designed the entire experience around the Safe Harbor de-identification method. The core insight: if you never collect PHI in the first place, there's nothing to breach.
If you never collect PHI, there's nothing to breach
Healthcare forms are notoriously overwhelming. Instead of presenting all fields at once, I designed a 4-step progressive flow where each section reveals only after the previous is complete. This transforms a daunting form into a manageable conversation.
4-step form reduces cognitive load by revealing complexity gradually
The biggest UX challenge wasn't the AI, it was trust. Healthcare providers needed to feel confident that using Luma wouldn't put their practice at risk. I designed a multi-layered approach to compliance communication.
Every touchpoint reinforces compliance confidence
Do not paste SSN, dates of birth, full addresses, phone numbers, or other HIPAA identifiers.
View full list of 18 HIPAA identifiers →Patient name, age (not DOB), state, diagnosis codes, lab values, treatment history, and clinical notes.
I confirm that I have reviewed the information above and have not included any of the 18 HIPAA identifiers in my clinical notes submission.
Agreement logged: Jan 23, 2026 at 2:34 PM • User ID: usr_7x92kSafe Harbor requires removing all 18 identifiers for de-identification
The dashboard needed to do more than list cases. It needed to communicate value. The "Revenue Protected" metric transforms abstract documentation work into tangible financial impact, reinforcing the product's value with every login.
Case management with real-time revenue tracking
Quick edits without losing context
Modify patient and claim information
Healthcare software often feels cold and clinical. Luma's design balances professional credibility with approachable warmth: sage greens, soft gradients, and serif typography that feels trustworthy without being sterile.
Tokens, components, and patterns built for healthcare
Frosted glass effect with backdrop blur for layered UI depth.
Dribbble
.glass-card
Subtle grid pattern for clinical precision aesthetic.
.medical-grid
.animate-fade-in-up
Entry animation
.animate-slide-in-left
Lateral entry
.animate-scale-in
Modal open
.animate-glow-pulse
CTA emphasis
As the founding designer and full-stack developer, I took Luma from initial concept through to a production-ready SaaS platform. The end-to-end ownership meant every decision, from UX strategy to database schema, was aligned toward the same goal: making compliance effortless.
"I used to spend my lunch breaks writing prior auth letters. Now I paste my notes and have a complete, compliant document in under a minute. The fact that I don't have to worry about HIPAA violations is what sold me."
Dr. Sarah Mitchell, Wound Care Specialist, Texas
"From an audit perspective, Luma's documentation is exactly what we look for. The LCD criteria references, the treatment history, the clinical justification, it's all there. Providers using this tool are submitting cleaner claims than I've seen in years."
Michael Torres, Healthcare Compliance Auditor, Florida
Luma has grown to $10K MRR and continues to expand steadily. By building on Vercel's edge infrastructure and Supabase's scalable PostgreSQL backend, the platform is architected to grow alongside its users without requiring a complete rebuild. What started as a solo project now serves practices across multiple states, with the technical foundation ready to support the next phase of growth.
The reason Luma found product-market fit so quickly comes down to relationships. After 10 years working in healthcare, I had built genuine connections with physicians, their clinical teams, and administrative staff. When I started exploring this idea, I didn't have to guess at pain points. I picked up the phone, talked to people I'd worked with for years, and asked them directly: what's breaking your workflow? The answer was unanimous: prior authorization documentation was eating their time and putting their revenue at risk.
A real example of generated prior authorization documentation
Biologics Prior Authorization • Aetna
This letter is written in support of prior authorization for biological therapy for Margaret Thompson, a 67-year-old female presenting with a chronic venous leg ulcer, classified as Wagner Grade 2.
Per Aetna LCD L33831, biological therapy is medically necessary when standard wound care has failed. Patient meets all inclusion criteria.