Inspiration
Healthcare letters are often difficult for patients to understand. Many people receive appointment letters, referral updates, waiting-list notifications, and prescription-related communications without fully understanding what action they need to take next.
Research showed that a significant proportion of adults struggle with health literacy, while millions of NHS appointments are missed every year due to confusion around healthcare communications. CareClarity was inspired by a simple question:
"What if every patient could instantly understand what their healthcare letter is asking them to do?"
Rather than creating another medical chatbot, we focused on a safer and more practical problem: helping patients understand healthcare administration and navigate the next steps with confidence. This aligns directly with the Health Impact challenge of making NHS administration easier for patients.
What it does
CareClarity is an AI-powered healthcare administration companion that converts confusing NHS-style healthcare letters into a structured patient dashboard.
The platform can:
Extract appointment dates, times, locations and departments Generate plain-English summaries Create actionable patient checklists Highlight missing or unclear information Produce appointment preparation guidance Generate safe questions patients may wish to ask clinicians Translate content into multiple languages Generate summaries for carers and family members Explain confusing healthcare administration terminology
CareClarity maintains a strict safety boundary and provides administrative guidance only. It never diagnoses conditions, recommends treatments or provides medical advice.
How we built it
CareClarity was built as a full-stack web application using:
React TypeScript Vite Tailwind CSS Vercel deployment Z.ai GLM API
Z.ai acts as the core intelligence engine throughout the workflow.
Healthcare letters pass through multiple AI-powered stages:
Information extraction Plain-English translation Action checklist generation Appointment preparation guidance Clinician question generation Safety validation
Every response is validated using structured schemas and an additional safety review layer before being shown to users. If validation fails, CareClarity returns a safe fallback response rather than generating potentially unsafe information.
We also used:
Manus for research, personas, competitor analysis and workflow design Fotor for marketing materials and visual assets Cursor to accelerate development and interface refinement Challenges we ran into
One of the biggest challenges was designing a healthcare-focused AI product without crossing into clinical advice.
We had to ensure that:
The system never diagnoses conditions The system never recommends medications The system never suggests treatments Outputs remain administrative rather than medical
Another challenge was handling the huge variation in healthcare letters. Different hospitals, departments and services use different wording and structures, making reliable extraction difficult.
We also needed to create multilingual support while preserving critical information such as appointment dates, locations and contact details.
Finally, building a strong safety validation workflow within the limited hackathon timeframe required careful prompt engineering, schema validation and fallback design.
Accomplishments that we're proud of
We are proud that CareClarity is much more than a simple AI summarisation tool.
Key achievements include:
A complete healthcare administration workflow powered by Z.ai Structured extraction of appointment and referral information Plain-English explanations of complex healthcare communications AI-generated action checklists Appointment preparation guidance Multilingual support Carer and family summaries Missing-information detection A dedicated AI safety validation layer A mobile-friendly and accessible interface
We are particularly proud of maintaining a strict admin-only boundary while still providing meaningful support to patients. This makes CareClarity both useful and safe.
What we learned
Throughout the hackathon we learned that:
Healthcare communication is often a bigger problem than healthcare information itself. Many patients do not need medical advice; they need help understanding what to do next. Safety is just as important as intelligence when building AI for healthcare. User-centred design significantly improves product decisions. AI can create real value when integrated into complete workflows rather than isolated features.
We also learned how important accessibility, multilingual support and trust are when designing products intended for diverse patient populations.
What's next for CareClarity
Our vision is to develop CareClarity into a scalable patient-support platform that complements existing NHS digital services.
Future plans include:
PDF and image upload support OCR for scanned healthcare letters Expanded multilingual support Trusted family and caregiver sharing tools Integration with NHS digital communication platforms NHS Trust pilot programmes Public API access for healthcare software providers Accessibility enhancements Advanced appointment readiness packs
Long-term, we believe CareClarity can help reduce patient confusion, improve appointment attendance, reduce pressure on NHS administrative teams and make healthcare communication more accessible for everyone. The business model supports growth through patient subscriptions, NHS Trust licensing and a developer API, providing a realistic path beyond the hackathon.
CareClarity Devpost Submission | VibeHack London 2026 Team Eleven | Track 1: Health Impact | Challenge 2: Making NHS Admin Easier for Patients
Project Title CareClarity Team Name Team Eleven Team Members Md Abdullah Al Mamun (UoB) | Imran Al Munyeem (NTU) | Rudrasinh Parmar / Zoro (UoB) Team Members Info (Website) https://almamun.tech/ | https://munyeem.netlify.app/ | https://zoro.website/
Main Track Track 1: Health Impact Challenge Challenge 2: Making NHS Admin Easier for Patients Live Demo (Website) https://careclarity-eleven.vercel.app
Live Demo (Video) Product Video Link
GitHub Repo https://github.com/imranalmunyeem/CareClarity
SHORT PROJECT SUMMARY
CareClarity is a mobile-friendly healthcare administration companion that turns confusing NHS-style appointment letters, referral updates, prescription paperwork and admin instructions into clear, plain-English next steps. It produces a structured patient dashboard complete with an action checklist, appointment readiness pack, missing-detail flags and a downloadable summary for trusted family members or carers — while enforcing a strict boundary: administrative understanding only, never medical advice.
WHAT WE BUILT
CareClarity is a full-stack web application (React + TypeScript + Vite) deployed on Vercel, with server-side API routes that call the Z.AI GLM API. The API key is never exposed to the browser.
Feature Description Letter Understanding Extracts letter type, department, appointment date, time, location, contact info, clinician/team and action required Plain-English Explanation Rewrites formal NHS admin wording into clear everyday language Action Checklist Safe admin-only next steps: confirm appointment, check missing details, etc. Missing Detail Flags Highlights no date, no time, no location, unclear contact, conflicting instructions or deadline-free actions Readiness Pack Before-you-go steps, documents to bring, items to confirm at the clinic What Changed? Compares two letters side-by-side to identify changed appointment or admin details Carer Summary Downloadable TXT/PDF summary with details, checklist, contacts and safety notice Prescription Helper Extracts collection/admin wording and reference numbers without any medicine advice NHS App Guide Admin-only guidance on finding appointments, messages, referrals or prescriptions in the NHS App Explain a Sentence Paste one confusing sentence and receive an admin-only plain-English explanation Multilingual Translation Translates letters into 13 languages including Bengali, Urdu, Arabic, Polish, Punjabi and Ukrainian Language Selector Reloads the full CareClarity interface in the patient's preferred language Product Chatbox Answers how-to-use questions in the patient's language; refuses medical advice requests Accessibility Mode Large text, high contrast, dyslexia-friendly layout and browser read-aloud support Safety Banner Always-visible banner reminding patients that CareClarity explains admin information only Safe Fallback If Z.AI is unavailable or output is invalid, the app returns a safe fallback rather than guessing
WHO IT IS FOR
Patients in the UK who receive healthcare admin correspondence and find it confusing, stressful or hard to act on — particularly: • Older patients unfamiliar with clinic or NHS admin language • Patients whose first language is not English • People with low digital confidence or literacy difficulties • Carers managing paperwork on behalf of a family member • Anyone who has received a letter with missing details, unclear instructions or no obvious next step
WHAT PROBLEM IT SOLVES
Healthcare paperwork often assumes patients understand referral wording, clinic procedures, appointment rules and NHS admin processes. In practice, many struggle to answer basic questions: What is this letter asking me to do? Where do I need to go? Is there a date and time? What should I ask the clinic?
One missed date, unclear location or forgotten instruction can mean a missed appointment, extra anxiety and unnecessary calls to already-stretched NHS admin teams.
Standard tools do not solve this: PDF readers only show the document. Generic translators miss healthcare admin meaning. Search engines can be unsafe or irrelevant. The NHS App shows confirmed appointments but does not explain confusing correspondence.
CareClarity solves a narrower, specific problem: helping patients understand and act on healthcare admin paperwork safely, quickly and in their own language.
HOW AI WAS USED
Z.AI GLM API (Core Engine) All AI calls are made server-side through Vercel API routes. Z.AI powers: • Letter understanding and structured admin detail extraction • Plain-English explanation generation and action checklist production • Appointment preparation guidance and clinician question generation • Missing and unclear detail detection • Safety validation — a dedicated Z.AI call checks every output for unsafe medical content before it reaches the patient • Letter comparison analysis • Multilingual translation into 13 patient languages • Sentence-level explanation • Product-support chatbox responses • Safe fallback decision support when the primary call fails
Safety pipeline: every output passes through (1) admin-only system prompts, (2) Zod schema validation, and (3) a Z.AI safety validation call. If UNSAFE or unavailable, CareClarity displays a safe fallback — never an unvalidated response.
Other AI Tools Used During the Build Manus Patient admin pain-point research, user personas, competitor comparison, demo script preparation, workflow documentation Cursor UI layout improvements, TypeScript error fixes, interface polish during the hackathon build Fotor Product poster, before/after patient confusion graphic, Z.AI workflow infographic, multilingual accessibility graphic, Devpost banner and demo slide imagery
DEMO & REPOSITORY LINKS
Live Product https://careclarity-eleven.vercel.app
Marketing Video Product Video Link
GitHub Repo https://github.com/imranalmunyeem/CareClarity
SPECIAL AWARDS APPLIED FOR
Award What We Used It For Evidence Fotor Vibe Marketing Award Product poster, before/after graphic, Z.AI workflow infographic, Devpost banner and demo imagery created with Fotor Example Screenshots Attached Manus Real-World Use Case Award Research, user personas, competitor comparison, demo script and workflow documentation prepared with Manus Example Screenshots Attached Z.ai x Orbit Builder Workflow Z.AI GLM API powers the full analysis, safety validation, translation and chatbox pipeline; Ortie Package generated Evidence GitHub Code
SAFETY NOTES FOR JUDGES
CareClarity is not a medical chatbot. It does not diagnose conditions, recommend medicines, suggest treatment plans or claim NHS endorsement. Safety is enforced at three levels: (1) admin-only system prompts, (2) Zod schema validation on all AI responses, and (3) a dedicated Z.AI safety validation call before any output is shown to the patient.
For urgent medical help in the UK: NHS 111. For life-threatening emergencies: 999.
Built by Team Eleven at VibeHack London 2026 | UCL, 6–7 June 2026
Built With
- api
- glm
- react
- typescript
- vercel
- vite
- z.ai

Log in or sign up for Devpost to join the conversation.