-
-
AutomobileWash3000
-
VETC Automotive Fintech Ecosystem
Inspiration
VETC is currently the leading electronic toll collection (ETC) application in Vietnam, with over 4 million active users. However, it is primarily used on a monthly basis (MAU) for occasional tasks such as toll payments, parking, or traffic fines. Meanwhile, Vietnam’s used car market is growing rapidly, projected to reach $13.2 billion by 2026 with a CAGR of approximately 14%. Car owners often face price pressure when selling vehicles due to a lack of transparent information about the car’s real condition (actual mileage, accident history, maintenance records). At the same time, Tasco aims to aggressively expand its car wash network from 100 to 3000 locations. This project is inspired by the idea of transforming VETC from a simple utility app into a daily smart car super app, where AI helps car owners:
- understand the true value and health of their vehicles
- easily buy, sell, and maintain cars
- receive real-world incentives Goal: Convert monthly users into daily users, while turning every car wash into an opportunity to increase vehicle value and drive revenue across Tasco’s ecosystem.
What it does
VETC 2.0 – Automobile Fintech Ecosystem enables car owners to interact with the app daily through four core MVP features:
- Car valuation and Car Health Score: where users take 6 to 8 photos of their car including exterior, interior, odometer, and maintenance records, and the Qwen3.6-Plus AI analyzes them instantly to provide an accurate market value, a health score, accident history, and a suggested buyback price from VETC, with results displayed clearly such as “Your car is worth 680 million VND today, up 15 million compared to last month.”
- Smart marketplace: that allows users to list used cars at AI-determined prices, supports trade-in to new cars within 48 hours with examples like “Your car is currently worth 500 million, upgrade to Geely with only an additional 100 million,” and connects buyers and sellers quickly and transparently while reducing time and risk
- Car wash subscription combined with check-in: where users pay around 159,000 VND per month for unlimited washes along with Premium tier benefits such as free parking for one year and 24/7 roadside assistance, and when checking in, the app automatically performs a vehicle inspection, updates the Car Health Score, and notifies users that their car value has increased by 12 to 20 million after the wash and inspection
- ATM loan: that allows fast loan estimation based on the current car value from Car Health score, with future integration planned with banks As a result, users open the app daily to check their car value, find car wash deals, and participate in the marketplace, turning VETC into a daily habit instead of a monthly tool
How I built it
I am AI Engineer and built this entire MVP alone with zero capital, using AI Copilot and the sponsored Qwen model within four days:
- Frontend: built with React Native by reusing as much as possible from a VETC-like interface
- Backend: Firebase including Firestore for real-time data, Storage, and Authentication, temporarily replacing Node.js at the prototype stage and enabling extremely fast development without managing servers
- AI core: call Qwen API. I leveraged the Qwen AI model, particularly Qwen3.6-Plus, as the core intelligence behind the system for: Multi-image understanding: analyze multiple car images, extract key attributes such as odometer reading, exterior scratches, condition, potential damage, maintenance book and overall quality, and convert them into structured outputs like valuation and health score. Structured output: I crafted precise prompts so Qwen3.6-Plus returns clean JSON containing:
- OCR license plate, Kilometerage on speedometer
- Maker, Model, Color, Car type recognition
- estimated market value in VND/USD
- car health score (0–100)
- detected damages
recommendation for “VETC Certified” status Production readiness: Fast response time and reasonable token cost, making it suitable for real-time use inside the mobile app. This allows the system to deliver transparent and explainable results to users in real time.
Database is designed simply with eight collections such as users, vehicles, valuations, marketplaceListings, and washSubscriptions
Integrate Firestore real-time listeners, simulated QR for car wash check-ins, and subscription logic with automatic reward points that increase the health score after each wash The entire MVP can be demonstrated live on a real mobile device, from capturing images to AI valuation, storing Car Health Score, listing a car for sale, or checking in for a car wash and seeing the car value increase.
Challenges I ran into
Faced challenges in handling multi-image analysis with Qwen AI model within a short time , which required optimized prompting efficiently for the prototype
Integrating real-time updates between valuation, Car Health Score, and the marketplace without causing performance issues was also a challenge
Another major challenge was scope limitation, as we had to decide to cut several secondary features such as tourism, Tasco real estate, and social car sharing in order to focus on the core MVP within four days despite having many initial ideas
Additionally, delivering a sufficiently impressive project with great business model as a single developer without a professional designer was not easy.
Accomplishments that I'm proud of
- Built a working MVP within four days as a single developer using AI, demonstrating significant cost efficiency for Tasco
- Integrated Qwen3.6-Plus to accurately analyze vehicles and create a unique “Car Health Score” feature that very few automotive applications currently offer
- Established a strong connection between car washing and increasing vehicle value, allowing users to clearly see financial benefits when visiting Tasco’s car wash locations, directly supporting the goal of scaling from 100 to 3000 locations
- Created a smooth user flow where users open the app in the morning to check their car value, then choose to participate in the marketplace or wash their car, receive rewards and incentives, and naturally increase daily active usage
What I learned
- Multimodal AI models like Qwen3.6-Plus can significantly replace manual effort in car valuation and condition assessment, bringing higher transparency to the used car market
- To convert monthly users into daily users, it is necessary to create clear daily value such as dynamic car valuation and personalized incentives rather than just adding utility features
- Loss-leader strategies such as initial car wash promotions are necessary to acquire users, but they must be tightly connected with subscription models and marketplace commissions to quickly reach break-even
- Rapid prototyping using Firebase and AI is highly effective in validating ideas
- How to manage time, spend time on work, build everything by myself
What's next for
In the short term over the next three to six months, I plan to scale the MVP to production using real data from VETC, deploy a car wash franchise program and subscription incentives to reach 300,000 to 600,000 active users, and integrate with banks for loan services as well as Tasco showrooms for trade-in functionality in next 15 months. For future work, I will develop smart travel tours with optimized routes and integration with Tasco resorts and real estate, build a “Show your car” community with verified Car Health Score, and expand deeper integration with Tasco insurance and real estate services. The long-term goal is to transform VETC into the most comprehensive car super app in Vietnam, helping Tasco reach 5 million users by 2027, scale its car wash network aggressively, and dominate the transparent used car market.
