_edited.png)
SERVICE DESIGN I AI AUGMENTED DESIGN I CURSOR AI
AI-Assisted Service Design for Health Innovation
PROJECT OVERVIEW
Mosaic Design Labs is a biotech innovation studio that bridges cutting-edge research and deployable diagnostics. I was tasked with transforming raw technical material spanning research papers, field studies, interviews, and engineering decks into interactive, web-ready case studies. To accelerate this process, I integrated AI tools like ChatGPT, Claude and Cursor AI throughout the entire design and publication pipeline.
IMPACT
-
Reduced content production time by ~60% by building an AI-assisted workflow using Cursor and ChatGPT
-
Delivered case studies for 6+ clients, supporting investor pitches, grant applications, and partner acquisitions
-
Built a reusable component library that cut page build time from days to hours for each new project
DURATION
Ongoing
TEAM
4
MY ROLE
Service Designer at Mosaic Design Labs
CONTEXT
Mosaic Design Labs is a biotech innovation studio. One of my roles is to transform raw research and transcripts into structured, web-ready case studies and ship them using Cursor, an AI-powered coding + writing IDE.
CHALLENGE
No repeatable system to communicate complex biomedical IP to broader audiences
APPROACH
Service design process + AI collaboration at every stage from research to deployment
OUTCOME
Live case studies, a scalable content framework, and a 60% faster workflow
DISCOVER
Understanding the stakeholders and the science
Mosaic Design Labs develops diagnostics, microfluidics, and health tech tools in partnership with leading research institutions. The challenge: their work is dense, multidisciplinary, and buried in transcripts and technical PDFs. Audiences from funders to potential collaborators, need to understand the impact quickly and clearly.
I audit the current content ecosystem, conduct stakeholder interviews, and map out how different audiences interact with the material. What do grant reviewers need? What does a potential hire want to feel? This shapes every design decision downstream.
AI in Use
I use ChatGPT to synthesize long stakeholder transcripts and research PDFs uploading raw documents and prompting for key themes, story-worthy quotes, and narrative tension. This compresses hours of insight gathering into focused design inputs.
Define
Framing the system architecture
Without a defined structure, every case study was being recreated from scratch. I design a modular content blueprint consistent narrative arcs (Need → Innovation → Impact), reusable section types, and a visual language that scales across projects like Lucira, CellScope, and future clients.
This stage also involves building prompt templates a library of AI inputs that reliably transform academic language into human-centered copy, from intro paragraphs to pull quotes.
AI in Use
I co-create narrative frameworks with ChatGPT feeding in technical excerpts and prompting for restructured story arcs. Prompt templates I build here become reusable assets for future projects, turning AI collaboration into a design system of its own.
Ideate
Exploring narrative structures and visual directions
I sketch multiple approaches to how each story unfolds long scrolls, sticky sidebars, interactive diagrams. The system must support both skimmable summaries for busy funders and deep-dive content for researchers. I rapidly prototype visual directions using Figma and AI-assisted ideation.
A key insight from this phase: readers don't read linearly. Designing for entry points the stat that makes them stop, the quote that makes them lean in shapes the entire layout logic.
AI in Use
I use Notion for visual moodboarding and Cursor AI to test layout variants rapidly in code skipping the Figma-to-handoff friction. ChatGPT and Claude helps brainstorm section headers, alternative framings, and CTAs at speed.
Prototype
Building directly in Cursor AI
I develop the case study system in Cursor an AI-native IDE where I write, prompt, and refine code side by side. This collapses the traditional Figma → handoff → dev cycle. I build sticky navigation, scroll-activated transitions, responsive image grids, and dynamic quote callouts all directly from design intent to working page.
The output is not just individual pages, but a component library that makes future case studies faster and more consistent to build.
AI in Use
Cursor helps me generate Tailwind + React components through natural language prompts. I iterate on layout logic, sticky sidebars, image proportions, and animation timing by describing intent then refining the output. Every prompt is a design decision.
Existing Lucira Case-study
New Lucira Case-study designed on Cursor AI x Figma
Test & Iterate
Collaborating with stakeholders and refining
Weekly design reviews with the Mosaic founder and scientific collaborators ensure every case study is technically precise and narratively clear. I test how real users move through the content, what they skip, and where they engage then refine accordingly.
The most meaningful iteration often happens at the intersection of design and science: finding the image that replaces three paragraphs, or the single quote that carries an entire section.
AI in Use
ChatGPT assists with tone adjustments, accessibility-friendly alt text, and simplified explanations for technical diagrams. I prompt for multiple phrasings and curate the best fit, AI generates the options, I make the call.
.png)
Impact
What the system enables
60%
Faster content-to-page workflow using AI tools
6+
Clients supported with investor-ready narratives
∞
Scalable system for future case studies without rebuilding
The system goes beyond individual pages it creates a repeatable, AI-augmented pipeline that any future team member can build on. Complex science gets communicated clearly. Stories that used to take weeks to write now take days.
"This project proved to me that design is no longer just visual. It's strategic, it's narrative and now, it's deeply collaborative with AI. Every prompt is a design decision."