Elliot RylandsWorking at the intersection of Web, Brand, Product, and AI for healthcare royalty.
Senior UX Designer at Jane App, a healthcare practice-management platform used by clinics around the world. Working across web, AI, design systems, and product. Led the new marketing website on a lightweight system shipped in a month, built the Figma plugin that translates tokens into engineering-friendly outputs, and leads exploratory AI initiatives shaping Jane's product direction.
Jane needed someone who could work in the gaps between brand, product, systems, and AI. The marketing site hadn't meaningfully evolved in years, the structure had drifted, accessibility consistency was limited, and there wasn't a flexible system underneath that marketing and engineering could confidently build from.
Build enough structure underneath the work that the site could evolve properly afterwards. A lightweight marketing design system shipped in a month, refined while delivering. A Figma plugin built alongside, bridging tokens into engineering. AI prototyping in parallel, quiet AI, not performative AI.
Seventeen feature pages live on the new system, with the homepage imminent. The Figma plugin picked up across design and engineering. AI prototypes influencing wider roadmap conversations. The role itself stopped being about individual screens and became about the structure that lets teams keep moving.
Designing across web, AI, and the seams between teams
Jane App is a healthcare practice-management platform used by clinics around the world. Practitioners rely on it to run their businesses day-to-day, and patients rely on it to access care without friction.
I joined as a Senior UX Designer in April 2025, working across web, AI, brand, systems, and product. The role naturally sat between disciplines. Part product design, part systems thinking, part design engineering, and increasingly AI workflow exploration as the company accelerated its investment in tooling and experimentation.
Three people the work has to land for.
Jane spans practitioner, admin, and patient surfaces, each with very different tolerances for friction. Every system decision had to read as right for all three.
Sarah Chen
- Setup
- Solo clinic
- Books
- Online + in-person
- Tech
- Confident
Sarah runs a solo massage-therapy practice. Jane is her schedule, her chart, and her invoicing. She switches between intake, treatment notes, and rebooking inside a single 60-minute window.
- Minimal admin between back-to-back appointments
- Charting that gets out of her way
- Patients self-rebooking without phone calls
- Tools that demand context-switching mid-session
- Forms that ask the same thing twice
- Anything that makes patients call the desk
Marcus Reed
- Locations
- 3 clinics
- Team
- 12 practitioners
- Reporting
- Weekly
Marcus owns three physiotherapy clinics. He needs an honest view across all three without logging in and out, and the team needs autonomy without him losing the plot.
- One-glance health of every location
- Revenue and utilisation that match the books
- Permissioning he can change without an email to support
- Data in three places that mostly agrees
- Reporting that requires a spreadsheet to finish the job
- Per-location config drift
Tom Garcia
- Reason
- Lower-back rehab
- Device
- Phone-first
- Habit
- Books at 11pm
Tom books physio between work meetings, usually on his phone, often outside business hours. He'd rather not phone anyone.
- Real availability he can act on immediately
- Reschedule and cancel without a phone call
- A reminder that doesn't feel like spam
- Forms that won't autofill on mobile
- Receiving a confirmation but no calendar invite
- Being asked to call the clinic to change a time
How might we…
The questions that anchored the marketing system, the plugin, and the AI prototyping. Each one had to be answerable for both clinicians using Jane and the teams building it.
Ship a marketing design system in a month without it feeling rushed, fragile, or single-use?
Translate an evolving brand into product surfaces without breaking the trust practitioners place in the tool every day?
Get tokens out of Figma and into engineering in one step: CSS, JSON, and human-readable docs?
Introduce AI to clinicians in ways that feel genuinely useful, not performative or risky?
Hold seventeen feature pages together with shared structure without making any of them feel templated?
Marketing system objectives.
The system landed inside a moment when brand, product, and engineering were all moving. The matrix below is how we mapped what the system had to absorb up-front, so each page didn't re-negotiate it.
Token to interface, one path.
Before the Figma plugin, design and engineering re-negotiated tokens on every project. The new pipeline collapses that into a single linear flow with two safe branches (a docs export and a sandbox preview).
How the work came together
The work sat between systems, brand, and product
The role quickly became about the gaps between teams. Brand evolving in one direction, product in another, marketing timelines moving fast, AI becoming a serious conversation across the company. The website project landed inside that environment. The site hadn't meaningfully evolved in years and there wasn't a flexible system underneath the experience that marketing and engineering could confidently build from. The challenge wasn't just redesigning pages, it was building enough structure underneath the work that the site could evolve properly afterwards.
A month of building enough structure to move
The first month was spent building a lightweight marketing design system quickly enough that the wider project wouldn't stall behind it. Not a giant enterprise system. Not a six-month redesign exercise. Just enough structure to support the homepage, seventeen feature pages, marketing experiments, and accessibility improvements without every page becoming bespoke. The interesting part wasn't the speed. It was the adoption. The system was picked up quickly by both design and engineering without the usual friction mature companies create around new process. Alongside it, a Figma plugin that translated primitive and semantic tokens directly into engineering-friendly outputs, CSS, JSON, and developer documentation.
Three moving problems at the same time
The website project became difficult because three problems were moving at once. Brand translation, as Jane's visual identity evolved while the site was actively being shipped. System flexibility, with seventeen feature pages needing variation without turning into seventeen disconnected layouts. Alignment across product, marketing, and brand, all three teams with different valid priorities. The only realistic way through was systems thinking. Instead of solving the same tension repeatedly at page level, the design system absorbed the negotiation upfront through reusable patterns, layout rules, and shared interaction behaviour.
Building quieter AI experiences
Alongside the website work, I lead exploratory AI initiatives across multiple teams at Jane. A large part of the work focuses on prototyping how AI should actually appear inside healthcare software. Not AI for the sake of visibility. Useful AI. Quiet AI. The kind that supports practitioners without becoming distracting or performative. Interaction models, workflow integration, heuristic testing, prototyping systems, and internal tooling designed to help teams experiment faster without losing quality. Several prototypes have already started influencing wider roadmap conversations.
Most large design problems start underneath the interface long before they appear on the interface itself.
Marketing system, plugin, first pages.
The first six months at Jane. Foundations were laid quickly so the wider work, pages, plugin, AI, could land in parallel rather than queue behind the system.
Four cycles, one stack.
Each cycle began with a tight discovery against the same primitives, then a build that fed the next cycle. The point wasn't speed in isolation. It was that each cycle compounded on the one before it.
Foundations
- Primitive token audit
- Semantic token layer
- First-page shakedown
Token stack and layout grammar in place. First feature page shipped on the new system.
Plugin & pipeline
- Plugin architecture & API
- Token → CSS / JSON export
- Internal docs surface
Figma plugin shipping CSS, JSON, and developer docs in one pass. Design-to-engineering loop closed.
Brand & accessibility
- Brand evolution absorbed into tokens
- Accessibility patterns baked into primitives
- Pages 6–11 on the system
Brand evolution absorbed without page-level rework. Accessibility patterns live in the foundation, not retrofitted.
Quiet AI surfaces
- AI interaction model exploration
- Prototype surfaces on the same primitives
- Roadmap conversations with product
Multiple AI prototypes built on the marketing primitives, already shaping wider product roadmap.
From negotiated to absorbed.
Before the system, every feature page repeated the same negotiation between brand, product, and engineering. After the system, that negotiation lived in the foundation, and page-level cost dropped sharply.
Each page absorbed the cross-team negotiation independently. Cost stayed roughly flat as more pages shipped, because every page was its own project.
With the marketing system absorbing brand / product / eng decisions upfront, page-level cost dropped rapidly. By page seventeen, a new page was a layout choice, not a design problem.
How the work shifted month by month
The role started systems-heavy and shifted toward web and AI surfaces as the platform foundation stabilised. The chart below is a rough representation of where attention sat across the months rather than precise tracking.
A system underneath, the work on top
Most of the artefacts from this work live in the layer underneath the marketing site rather than on it. Tokens, plugins, layout systems, AI prototyping surfaces. The pages are the visible part. The system is the part that lets the pages keep evolving.
Building systems that help teams move
The biggest surprise after joining Jane was the pace. Five hundred-plus people, but still operating with a level of momentum and trust that feels much closer to a startup environment. The work rewards people who can build structure quickly, make decisions without over-ceremonising them, and stay hands-on while complexity increases around the project.
A lot of the role has become less about individual screens and more about helping teams move clearly through ambiguity. Design systems. AI tooling. Prototyping workflows. Accessibility. Engineering collaboration. Product thinking. Brand translation.
The interesting part is rarely the interface itself. It's the structure underneath that allows the interface to keep evolving without the whole thing slowing down.
Select highlights
The lightweight marketing design system that seventeen feature pages now sit on.
Feature pages designed from shared layout, interaction, and accessibility foundations.
The Figma plugin translating primitive and semantic tokens into CSS, JSON, and developer documentation.
Picked up quickly by both design and engineering without the usual friction around new process.
Prototypes shaping how AI lands inside healthcare software, useful, trustworthy, and appropriately restrained.
A mature company still operating with the momentum and trust of a much smaller one.