WASH3000

WASH3000 — AI Mobility OS on VETC (3M vehicles). Predicts, automates & pays for car services. Vision: Self-caring cars

Comment

WASH3000 — Devpost Submission

Inspiration

3 million VETC cars every day — but car owners still have to REMEMBER when to wash manually.

When participating in LotusHacks 2026, we realized Tasco is sitting on a "treasure trove" of data: a 99.98% accurate RFID infrastructure, automated traffic payment accounts, and the movement history of 3 million cars. But it is all only used for... road toll collection.

What would happen if AI could "read" this data and predict what a car needs BEFORE the owner even realizes it?

That's how WASH3000 was born — not just a car wash app, but the first step in building a Mobility OS in Vietnam.

What it does

🎯 Core Functionality

Car passes VETC station → AI predicts → Timely suggestion → One-tap booking → Auto payment

Feature Description
VETC Integration Automatic RFID recognition — no QR scanning, no registration required
AI Prediction Engine Analyzes mileage, weather, and history → predicts washing/maintenance needs
WashBot AI Assistant Proactive chatbot that automatically reminds, advises, and compares prices
Auto Payment Automatic deduction via VETC traffic account
Owner Dashboard Analytics for station owners to manage slots, pricing, and promotions

🚀 The Magic Moment

"Car 51F-123.45 just drove 150km through 3 toll stations → AI predicts 75% dirtiness → WashBot sends a notification: 'There's a car wash 5 mins away, a slot is open at 14:00, 20% off?' → One-tap book → Traffic account auto-deducts"

Total time: 30 seconds. Friction: 0.

How we built it

🏗️ Architecture

┌─────────────────────────────────────────────────────────────┐
│                     WASH3000 PLATFORM                       │
├─────────────────────────────────────────────────────────────┤
│  Frontend (React + Vite)     Backend (Vercel Serverless)    │
│  ├── Dashboard               ├── api/v1/ai-diagnose         │
│  ├── InsurancePage           ├── api/v1/tasco-insurance     │
│  ├── BookingFlow             ├── api/v1/payment             │
│  └── Chatbot Widget          └── api/v1/washbot/chat        │
└─────────────────────────────────────────────────────────────┘

🛠️ Tech Stack

Layer Technology
Frontend React, Vite, TailwindCSS, TypeScript, Framer Motion
Backend Vercel Serverless Functions, Node.js, TypeScript
AI/ML Qwen DashScope (Primary), Interfaze.ai AI (Fallback)
Database Supabase (PostgreSQL) with Row Level Security
APIs TASCO Insurance, VETC (mock), Open-Meteo Weather API
Chatbot WashBot — Proactive AI, Native Fetch + SSE, Tool Calling

📁 Key Files

  • api/v1/washbot/chat.ts — Core WashBot logic with Multi-AI Fallback
  • api/v1/tasco-insurance.ts — TASCO Insurance API endpoints
  • supabase/migrations/ — Database Schema, RLS & Seed Data
  • src/pages/InsurancePage.tsx — UI integration with TASCO branding

Challenges we ran into

1. Integrating real VETC API

  • Challenge: No access to the real VETC API.
  • Solution: Built a fully functional mock VETC API, ready to be replaced with the real API once access is granted.

2. AI Rate Limits & Fallback

  • Challenge: Qwen DashScope API quota is easily exhausted, causing 429/401 errors.
  • Solution: Built a Multi-AI Fallback System to Interfaze.ai automatically without downtime. Written 100% using Native Fetch JSON API instead of bulky SDKs.

3. Vercel Serverless Migration

  • Challenge: Traditional Node.js servers are hard to scale and maintain for a hackathon demo.
  • Solution: Fully migrated the backend (ai-diagnose, chat, payment, tasco-insurance) to Vercel Serverless Functions (api/v1/) for rapid deployment.

4. Framer Motion + React

  • Challenge: Complex component animation in a web environment.
  • Solution: Optimized components, focused on smooth layouts and natural UX.

5. Time Constraint

  • Challenge: Developing Frontend, Backend, AI & Database in a short time.
  • Solution: Adopted Backend-as-a-Service (Supabase) + Vercel, effectively dividing into phases.

Accomplishments that we're proud of

🏆 1. Leveraging VETC Infrastructure (Moat)

Element Savings
99.98% RFID recognition $2M
Traffic account $500K
3M existing users $5M
Total $7.5M+

Result: No need to reinvent the wheel — we leverage existing infrastructure.

🏆 2. TASCO Insurance & Supabase Integration

  • Built a professional DB with Supabase PostgreSQL and Row Level Security (RLS).
  • Seeded real TASCO insurance product data (Civil Liability, TascoSure Material, Two-Way).
  • Applied TASCO Branding standards (19001562 Hotline, 24/7 rescue, ONE STOP SHOP).

🏆 3. WashBot — Multi-AI & Tools

  • Proactive AI: Proactively calls tools like diagnose_vehicle and updates Dynamic Pricing.
  • Fail-safe Backend: Integrated Interfaze AI as a fallback when the primary AI runs out of quota, using Native Fetch JSON API with Auto-Streaming SSE.
  • Provided Insurance knowledge directly to the bot via the get_insurance_info tool.

🏆 4. Serverless End-to-End Flow & Auto-Payment

  • 100% deployment using Vercel Serverless Functions instead of a manual Express Backend.
  • VETC trigger → AI prediction → Notification → Booking → Auto-Payment → Success — Zero friction.

🏆 5. Professional Documentation

  • CEO Plan (SCOPE EXPANSION)
  • Engineering Review (architecture, error handling, tests)
  • Design Review (UI/UX, interaction states)
  • Elevator Pitch (HTML + Markdown)
  • TODOS.md with 20+ items

What we learned

💡 Technical Learnings

  1. VETC Data is "gold" — 3M cars + real movement behavior = massive AI training data.
  2. Rule-based AI can be good enough — No complex ML needed for MVP, a 75% accuracy is already impressive.
  3. Graceful degradation is important — Weather API fails? Fallback to cached data + low confidence.

💡 Product Learnings

  1. "Frictionless" is key — Each added step = 20% user churn.
  2. Proactive > Reactive — Waiting for customers to search (Uber) vs. anticipating needs (WASH3000).
  3. Platform > App — A Mobility OS vision attracts more funding than a simple car wash app.

💡 Process Learnings

  1. CEO Review before coding — Scope EXPANSION helps visualize the 10x vision.
  2. Task management with priorities — P1/P2/P3 helps focus when time is tight.
  3. Parallel documentation — Writing docs while coding, not leaving it for the end.

What's next for WASH3000

🚀 Phase 2: Multi-Service AI (3 months)

  • [ ] Maintenance — Predict oil changes and air filter replacements.
  • [ ] Tire Replacement — Monitor tire pressure and wear.
  • [ ] OBD-II Integration — Real-time car health monitoring.

🚀 Phase 3: Mobility OS (6 months)

  • [ ] Insurance — Automatic compare & renew.
  • [ ] Car Rental — Car-sharing platform.
  • [ ] 3rd Party APIs — Allow other automotive services to integrate.

🚀 Phase 4: Predictive Maintenance (12 months)

  • [ ] AI Failure Prediction — Predicts problems before the car breaks down.
  • [ ] Preventive booking — Book services before they are strictly needed.
  • [ ] Full autonomous care — Cars that are 100% self-caring.

📈 Business Goals

Milestone Target
Month 6 10,000 active users
Month 12 100,000 users, 500 connected stations
Year 2 500,000 users, full Mobility OS
Year 3 IPO or Acquisition by Tasco

🤝 Partnership Opportunities

  • Tasco/VETC: Real API access, marketing to 3M cars.
  • Car Manufacturers: Integration into head units (Toyota, Honda, VinFast).
  • Insurance Companies: Data-driven risk assessment.
  • Gas Stations: Cross-sell services.

🏆 Conclusion

WASH3000 is not just a car wash app. It is the first step in building a Mobility OS — where AI cares for the car on behalf of the owner, leveraging existing VETC infrastructure, and creating a $7.5M+ moat that no one can copy.

Tagline: "Cars take care of themselves, users just drive."

Submitted to LotusHacks 2026 Team: [VA Team]

Built With

  • ai
  • framer
  • interfaze.ai
  • motion
  • native-fetch-+-sse
  • node.js
  • qwen-dashscope-(primary)
  • react
  • supabase
  • tailwindcss
  • typescript
  • vercel-serverless-functions
  • vite
  • washbot-?-proactive-ai
Share this project:

Updates