Vibe Coding with Kiro IDE

Business users build AI apps by describing what they want — run on your local machine

Kiro IDE Run Local Healthcare AI

What is Vibe Coding?

Describe what you want in plain language → Kiro IDE builds the app → Run it on your machine instantly

1Open Kiro
2Describe
3Answer Qs
4Kiro builds
5Run locally
6Use it
Requirements: Kiro IDE (free download) + Node.js or Python (Kiro will guide installation) — no AWS account needed for this workshop

Use Case 1: OPD Wait Time Dashboard

Track OPD queue wait times per department — real-time status on screen

What you get

For: Department Managers, Quality, Nursing Directors

Web app running locally — enter patient queue data → dashboard shows: queue count, avg wait time, which department is busiest, with charts

Local Server Python / Node.js Web UI

Steps in Kiro IDE

Step 1: Open Kiro IDE → File → New Project → "opd-wait-dashboard"

Step 2: Type in Kiro chat:

Build a OPD queue wait time dashboard for a hospital 6 departments, bar chart, status table (normal/busy/critical) Thai UI, Python Flask, run locally, include sample data
Build a web app to track OPD queue wait times for a hospital, run on local machine Requirements: - Input form: department name, number of patients in queue, avg time per patient (minutes) - Dashboard page showing: - Summary cards: total waiting, average wait time, busiest department - Bar chart comparing queue count per department - Table: department name, queue count, avg wait, status (normal/busy/critical) - Status: normal (< 30 min), busy (30-60 min), critical (> 60 min) - Include sample data for 6 departments - All UI in Thai - Use Python (Flask or FastAPI) + HTML/CSS/JS - Single command to run locally

Step 3: Kiro may ask → answer:

Step 4: Kiro generates code → check Explorer panel

Step 5: Run locally:

pip install flask python app.py

Step 6: Open browser → http://localhost:5000 → see dashboard with sample data

Use Case 2: Patient Feedback Analyzer

Upload CSV of patient reviews → analyze sentiment + themes → dashboard on your machine

What you get

For: Quality Team, Branch Managers, Executives

Web app running locally — upload review CSV → analyze with keyword matching → dashboard: pie chart sentiment, bar chart themes, branch comparison table

Local Server Python Web Dashboard AI (optional)

Steps in Kiro IDE

Step 1: Open Kiro IDE → New Project → "feedback-analyzer"

Step 2: Type in Kiro chat:

Build a patient feedback analyzer dashboard Upload CSV reviews, sentiment analysis, theme detection, branch comparison Thai UI, Python Flask + Chart.js, run locally, include 20 sample reviews
Build a web app to analyze patient feedback, run on local machine Requirements: - Upload CSV page (columns: date, rating, text, branch) - Analyze each review: - sentiment: positive (rating 4-5), neutral (3), negative (1-2) - theme detection from keywords: "wait/queue/long" = wait_time "staff/doctor/nurse/polite/rude" = staff "clean/dirty/facility" = facility "expensive/cost/price" = billing "parking" = parking - Dashboard: - Pie chart: positive/neutral/negative ratio - Bar chart: review count per theme - Table: branch comparison (name, review count, % negative, top complaint theme) - Filterable table of recent negative reviews - All UI in Thai - Use Python + Flask + Chart.js - Include 20 sample reviews - Single command to run locally

Step 3: Kiro may ask:

Step 4-5: Kiro generates → run:

pip install flask python app.py

Step 6: Open http://localhost:5000 → try uploading this sample CSV:

date,rating,text,branch 2026-07-01,5,หมอใจดี อธิบายละเอียด นัดได้เร็ว,สาทร 2026-07-02,2,รอนานมาก 3 ชั่วโมง ที่จอดรถเต็ม,พระราม9 2026-07-03,4,สะอาด บริการดี แต่ค่ายาแพง,สาทร 2026-07-04,1,พนักงานไม่สุภาพ พูดจาไม่ดี,เชียงใหม่ 2026-07-05,5,ประทับใจมาก ทุกอย่างเรียบร้อย,พระราม9 2026-07-06,2,รอคิวนานเกือบ 2 ชม. แอร์ไม่เย็น,ภูเก็ต 2026-07-07,3,หมอดี แต่ระบบนัดยุ่งยาก,เชียงใหม่ 2026-07-08,1,ค่ารักษาแพงมาก ไม่คุ้ม,ภูเก็ต

Use Case 3: Clinical Guideline Q&A Bot

Chatbot answers questions from hospital CPG/SOP PDFs — with page citations

What you get

For: Nurses, Residents, Pharmacists, Quality

Web app running locally — place PDF guidelines in a folder → system chunks the text → ask questions in Thai → get answers with citations (filename + page)

Local Server Python Web Chat UI AI (RAG local)

Steps in Kiro IDE

Step 1: Open Kiro IDE → New Project → "guideline-qa-bot"

Step 2: Type in Kiro chat:

Build a Q&A chatbot that answers from hospital CPG/SOP PDF documents Read PDFs from docs/ folder, search by keyword, show answer with page citation Thai UI, Python Flask, support OpenAI or Bedrock API key, run locally
Build a chatbot that answers questions from hospital CPG/SOP documents, run on local machine Requirements: - A "docs/" folder for PDF guideline files (CPG, SOP, policies) - On startup, read all PDFs, chunk text into segments, store in memory - Web Chat UI: - Question input (Thai language) - AI answer with citation: "Reference: [filename.pdf] page XX" - Show source text used for the answer (expandable) - Search: use keyword matching or TF-IDF (no vector database needed) - AI: support OpenAI API key or Amazon Bedrock API key (entered in settings page) - If no API key: show relevant chunks only without AI summary - All UI in Thai - Use Python (Flask) + PyPDF2 for reading PDFs - Include 1 sample PDF (mock CPG: "Diabetes Care Guidelines" 3 pages) - Single command to run locally

Step 3: Kiro may ask:

Step 4-5: Kiro generates → run:

pip install flask pypdf2 python app.py

Step 6: Open http://localhost:5000 → ask "What is the HbA1c target for diabetic patients?" → get answer + page citation

Tip: Use with real documents immediately — just put your CPG/SOP PDFs in the docs/ folder and restart. The system re-indexes automatically.

Use Case 4: Claim Rejection Analyzer

Analyze claim rejection patterns — group root causes, calculate financial impact, suggest fixes

What you get

For: Revenue Cycle, Finance, CFO, Medical Records Managers

Web app running locally — upload rejected claims CSV → group by root cause → Dashboard: impact ranking, branch comparison, fix recommendations per pattern

Local Server Python Web Dashboard AI (optional)

Steps in Kiro IDE

Step 1: Open Kiro IDE → New Project → "claim-analyzer"

Step 2: Type in Kiro chat:

Build a Claim Rejection Analyzer dashboard Upload CSV of rejected claims, classify root causes, rank by financial impact Show branch comparison + actionable fix per cause Thai UI, Python Flask + Chart.js, run locally, include sample data
Build a Claim Rejection Analyzer web app, run on local machine Requirements: - Upload CSV page (columns: claim_id, date, branch, payer, amount, rejection_reason, department) - Classify each rejection by root cause: "incorrect code/ICD/coding" = coding_error "missing document/incomplete" = missing_docs "not covered/exclusion/eligibility" = eligibility "exceeds limit" = limit_exceeded "late submission/overdue" = late_submission other = other - Dashboard: - Summary cards: total rejected claims, total amount lost, #1 root cause - Bar chart: amount by root cause (ranked by financial impact) - Branch comparison table - Recommendations table: each cause → actionable fix - Filter by payer - All UI in Thai - Use Python (Flask) + Chart.js - Include 15 sample records - Single command to run locally

Step 3: Kiro may ask:

Step 4-5: Kiro generates → run:

pip install flask python app.py

Step 6: Open http://localhost:5000 → upload sample CSV:

claim_id,date,branch,payer,amount,rejection_reason,department CLM001,2026-06-01,Sathorn,Muang Thai Life,15000,Incorrect ICD code,Internal Medicine CLM002,2026-06-02,Rama 9,AIA,8500,Missing referral document,Surgery CLM003,2026-06-03,Chiang Mai,Bangkok Insurance,45000,Exceeds coverage limit,Orthopedics CLM004,2026-06-04,Sathorn,Social Security,3200,Late submission over 30 days,Dentistry CLM005,2026-06-05,Phuket,AIA,22000,Not covered exclusion,Ophthalmology CLM006,2026-06-07,Rama 9,Muang Thai Life,12000,Procedure coding error,OB-GYN CLM007,2026-06-08,Chiang Mai,Bangkok Insurance,5500,Missing lab results,Internal Medicine CLM008,2026-06-10,Phuket,Social Security,18000,Exclusion not covered,Surgery

Workshop Tips

Things to know before you start

Before the Workshop

  • Download Kiro IDE from kiro.dev (free, no AWS account needed)
  • Install Python 3.10+ (check: python3 --version)
  • Install Node.js 20+ if choosing a Node-based use case
  • Internet connection for Kiro IDE (AI runs in cloud, app runs on your machine)

Troubleshooting

  • Code error? → Copy error into Kiro chat: "Fix this: [error]"
  • Want to change look? → "Change color to blue" or "Add a column"
  • Want more features? → "Add export CSV button" — Kiro updates instantly
  • Port 5000 in use? → "Change port to 8080"

Next Steps (After Workshop)

  • Use real data (anonymized) instead of mock data
  • Get an AWS account → deploy to cloud for team-wide access
  • Add AI (Amazon Bedrock) for smarter analysis
  • Connect a real database (RDS, DynamoDB) instead of JSON files