The Context: When Families Need Care in a Crisis
Searching for home care is rarely a calm, planned shopping experience. In the majority of cases, family members seek domiciliary care following an acute event: a parent has fallen at home, has been unexpectedly discharged from hospital, or has experienced accelerated cognitive decline from dementia.
The family members reaching out are emotionally overwhelmed, sleep-deprived, and navigating unfamiliar medical jargon. Yet most care agency websites make this harder: text walls, obscured pricing tiers, clinical intimidation, and long, multi-page enquiry forms.
How can we redesign a domiciliary care website so that an anxious daughter or son can understand care options, verify quality and safety, and request an assessment in less than 90 seconds?
My Process: Frontline Empathy → PRD → AI Prompt Engineering
I did not build this website by spending weeks in Figma drawing static wireframe boxes. Instead, I operated as an AI-native Product Builder:
Frontline User Journey Scoping
Drawing directly from my 6+ years managing shifts and talking to families in senior care, I mapped out the three distinct mental states of visitors: Immediate Emergency Reablement, Long-Term Dementia Support, and Carers Looking for Fair Work.
Writing the PRD & Structured Prompts
As I wrote in my Medium essay, "prompting is learning how to explain what you want properly". I created structured product requirement prompts outlining exact layout hierarchies, high-contrast color palettes, CQC compliance trust signals, and mobile-friendly tap targets.
AI Generation, Curation & Refinement
Directed AI code generation to produce clean, accessible semantic HTML and responsive CSS. Reviewed and iteratively refined the output for WCAG AAA contrast, mobile performance, and calm, non-stigmatizing medical terminology.
Design Showcase & Interface Breakdown
Key views from the completed redesign. Each interface is engineered to balance high emotional reassurance with immediate, actionable clarity.
Key Takeaways & Current Status
This project proved to me that the future of product building isn't about choosing between human empathy and AI speed—it is about fusing them:
- Domain knowledge is the moat: AI cannot know what an exhausted carer feels like during a handover, or how an anxious family reads a care home brochure. That human insight dictates the requirements.
- Prompt engineering is requirements engineering: When you know how to structure the problem clearly, AI becomes a powerful execution engine that translates intent into clean code in days rather than months.
The full redesign has been structured, built, and delivered. The provider is currently preparing production domain and hosting migration for live release.