Case 01
Allie v3 — Codex Health · July 2025 – ongoing

Setting the vision. Driving it to delivery.

Leading a complete reframe of a patient app — from content tabs to a product organized around two questions every patient actually asks.

Role
Design Director
Team
Design, engineering, clinical, data
Timeline
July 2025 – ongoing
Status
Internal — full case on request

The previous version was organized around the org. Not the patient.

Allie v2 split content into Tasks, Charts, and Library tabs. Patients had to navigate to find what they needed. The product knew nothing about their day, their priorities, or where they were in their care. The team had built around content types because that's how the org thought — not how patients lived.

The redesign needed to do more than refresh the UI. It needed to change the architecture of the app — to organize everything around what the patient was actually trying to do.

Six user needs, drawn from foundational research.

Earlier in my time at Codex, I led foundational user research with patients managing chronic conditions. Six needs surfaced consistently across every conversation. They became the lens for every product decision the team made — not just for v3, but for everything after.

01
A sense of control.
Patients live with a constant fear of the unknown — and a feeling that their bodies are out of their control. The product has to help them move from reactive to proactive.
02
Consistency in healthy habits.
Patients know what a good day looks like. They struggle to do it consistently. The gap between knowledge and daily action is where we have to live.
03
Education they can actually use.
Information matters when it's directly applied to a patient's life — like learning that certain foods spike them. We have to be the trusted source that turns information into decisions.
04
Reliable monitoring and feedback.
Patients trust their own intuition. They don't want constant testing. Feedback should confirm what they already feel, or quietly alert them to something they didn't notice.
05
A connection to their why.
The most powerful motivator isn't health data — it's their reason for staying healthy. Family. Travel. Showing up. The product has to connect the work to the why.
06
Support without dependence.
Patients rely on a small, trusted circle. The product should make it easier to ask for and receive help — not replace the people they already have.

Two questions, every day.

I believed the entire app should answer two questions for the patient, every time they opened it. Everything else — what's permanent, what's temporary, when Allie speaks up, when she stays quiet — would flow from those two questions.

How am I doing?
What do I do next?
— the two questions a patient app should answer, every day.

Five design principles for how the team would work.

Vision is the destination. Principles are how the team gets there together. I wrote five design principles so that anyone on the team — designer, engineer, PM, clinician — could make decisions consistent with the work, even when I wasn't in the room.

01
Patient-first, always.
Every decision starts with the patient. Not the org, not the content type, not what's easiest to build.
02
Be opinionated about their care.
Neutrality is the easy answer. We do the harder one — we tell patients what we believe will help them manage their condition, grounded in clinical guidance.
03
Bias toward simplicity.
Make every interaction feel obvious. The "stereotypical grandma" should be able to use it without instruction.
04
Respect their time.
Patients have full lives outside our product. Don't expect them to spend a lot of time here, and don't burden them with unnecessary tasks.
05
Reward the effort.
When a patient takes the time to use the app, they should get something back. Investment always results in a return.

PRDs that engineering could actually build from.

I authored detailed product requirements for every major piece of the redesign. Behavior, edge cases, content rules, interaction patterns — at a level engineering could pick up and build without needing me in every meeting.

This is the work that fills the gap between vision and execution. It's also the part of design leadership that looks like product management, and I do it because the work needs it. Vision without specification doesn't ship.

Scrubbed PRD page — Allie v3 delivery plan covering streaks, quick iterations to the Glass UI, and notes on iOS Nav decisions
Document structure
Redacted PRD page — Allie v3 'My Health Tab' specification with section structure visible (Overview, Product Requirements) and body content intentionally blurred
Requirement depth
Confidential — body content redacted.
Redacted PRD page — Allie v3 'Story Cards' specification with section structure visible and body content intentionally blurred
Breadth across surfaces
Confidential — body content redacted.
Full PRDs available on request.

Sequencing the work so design could stay ahead of build.

I broke the redesign into phases — what shipped first, what depended on what, what could be parallelized — and mapped the sequence with engineering on a Gantt that became our shared source of truth. The plan wasn't just a phase sequence. It was broken into weekly sprints against engineering capacity, so we had a week-by-week picture of what the team could absorb and when design needed to be ahead of build.

Jan
Feb
Mar
Apr – ongoing
Foundation
Navigation · design system
Design → Build

Hero widgets
Permanent data surfaces
Design → Build

Story cards
Temporary content moments
Phase 1 → Phase 2

Allie's Take + agent
Daily summary · deeper AI
Design → Build → Iterate

Migration & polish
Move v2 · retire old patterns
Ongoing
Abstracted delivery timeline · January 2026 – ongoing · full plan available on request

Building the structure, then designing inside it.

I defined the design principles. I built the user journeys. I created the structure my team and our cross-functional partners could work inside.

Then I contributed to the designs themselves. Sometimes by designing directly. Sometimes by giving guidance and encouragement. I knew what fundamentally needed to change — the architecture of the app — and I identified the key moves that would get us there. The final designs aren't mine alone. But my fingerprints are all over the vision, the structure, and the work.

Redacted overview of the Allie v3 Figma design board — four section titles visible (Story Cards PRD, Story Card Anatomy, Story Card Navigation, Phase 1 Story Cards) with the frame content within each section intentionally blurred
Confidential — full design board available on request.

A redesign in delivery.

Allie v3 is rolling out in phases. Each one is tested. Each one brings the patient experience closer to what those two questions demand.

The full delivery story isn't public yet — the screens, the metrics, the structural detail. I'd be glad to walk through the rest of it in person.

Want the full case?
Get in touch →
Next case
Ask Allie introducing the SMART goals framework
02
Teaching an AI health coach to think like a clinician.
Ask Allie · SMART Goals — built an AI conversation that does what static goal-setting forms never could.