Longeviti digitizes a physician's expertise into automated marketing, sales, and patient management, letting a single clinic launch a concierge practice without months of manual work. I designed the entire product across five surfaces and two completely different audiences, through continuous strategy conversations with a physician-founder who was still clarifying the vision as we went.
The first design decision was structural: all five surfaces had to share a single design system, or I'd never finish. That constraint shaped every component choice: if it couldn't serve both a physician admin view and a patient onboarding step, I had to rethink it.
Marketing Site
Physician-facing. Sells the concierge conversion.
Physician Onboarding
Guided setup including AI-avatar recording step.
Physician Portal
Dashboard, Patients, Financial, Admin, Services, Messaging.
Patient Onboarding
Physician video first. Then account, docs, payment.
Patient Portal
Dashboard, Services, Members, Financial, Messaging.
Discovery
Personas
User Journeys
Wireframes
Hi-Fi UI
Prototypes + Handoff
Founder strategy conversations running through all phases; scope evolved as we designed
Patient sees the physician first, not the brand
I made the patient onboarding open with the AI-generated welcome video of the physician: full-bleed, voice on, the physician's face. The Longeviti brand is present but quiet. That framing treats the product as a continuation of an existing doctor-patient relationship, not a new service to evaluate.
Pricing guardrails as a soft door, not a hard wall
Recommended pricing ranges use interactive sliders with the recommended value pre-selected. Pushing past the cap triggers a his will require review path, not a validation error. Revenue projections update live so physicians can see the adoption trade-off in real time rather than finding out after the fact.
Family privacy inside a shared account
The primary payer can see the fact of a family member's subscription and billing, but not the content of their care. I built permission-aware drawer patterns on both the Members and Financial pages that enforce this privacy logic without requiring the user to think about it. A 16-year-old's physician messages stay private from the parent paying the bill.
AI as default with a first-class manual override
Every AI-generated artifact (marketing copy, patient messages, the welcome video) has a visible manual override with full UI parity. Physicians who trust the AI use it. Physicians who want to author specific content have a first-class path. Nobody is locked into automation or made to feel the manual path is a workaround.
HIPAA-aware architecture, not HIPAA-flavored UI
No HIPAA-compliant! badges. No clinical-grey color palette. The compliance lives in the architecture (auth, data scoping, audit, message retention), not in the visual language. I made a deliberate decision to keep the UI warm and trust-building precisely because the compliance work is doing its job underneath, invisibly.
Designing the AI avatar step with no prior art
There's no established UX pattern for ecord a video of yourself that will be AI-personalized and delivered to patients as a medical welcome. I had to invent the onboarding flow from scratch, figuring out how to set expectations about what the AI would do, how to guide a non-technical physician through recording, and how to preview the output before committing. The hardest part was making something genuinely novel feel reassuringly familiar.
Scope evolving while I was designing it
The founders were clarifying their product vision in parallel with my design work, which meant new surfaces appeared mid-engagement. The design system had to absorb each addition without breaking what was already built. I dealt with this by keeping every component more modular than strictly necessary and establishing naming conventions that made extension predictable rather than ad hoc.
Medical trust requires earning, not claiming
Healthcare products can't rely on patterns borrowed from consumer apps. A physician won't trust a product that feels like a SaaS dashboard if they're putting patient relationships through it. Every interaction design decision had to ask: does this feel worthy of the clinical context? That's a higher bar than following best practices, and it slowed down certain decisions productively.
Complete delivery
Every flow in the project blueprint (all five surfaces, all persona states, all physician portal sections) has corresponding wireframes, hi-fi designs, and interactive prototypes ready for handoff.
Founders' product clarity
The founders ended the engagement with a sharper articulation of their product than they started with. Designing forces concrete decisions that blueprints and investor decks leave abstract. The design work served as a product strategy tool as much as an execution tool.
System built to grow
When Longeviti adds new verticals, the design system is already there. One system, five surfaces. The engineering team has a coherent reference, not a stack of disconnected screens.