CarMatch

Inspiration

🚘 Cars are high-ticket purchases in Vietnam, but the buying process is slow, fragmented, and low-trust. Today, many sales consultants still lack:

Deep product knowledge
Up-to-date pricing and policy clarity
And the ability to provide consistent, objective advice

=> As a result, the quality of consultation varies widely ⚠️

👦 At the same time, a new generation of buyers is emerging. They are:

Digital-first
Prefer to research on their own
And want to explore options freely before engaging with sales

But current tools don’t support this behavior => This creates a broken experience: low trust, low confidence, and delayed decisions ‼️

What it does

CarMatch is an AI co-pilot that helps you choose the right car by letting you interact naturally - just like talking to a knowledgeable, professional sales consultant - with clear recommendations, transparent pricing, and no sales pressure. 👩‍💼

How we built it

LLM-powered recommendation engine
Structured automotive knowledge base
Voice-enabled UX
Rapid prototyping with Qwen coder

Challenges we ran into

As a fast-moving prototype, we didn’t have enough time to fully refine all user flows, resulting in some rough edges in the experience.

Accomplishments that we're proud of

Shipped a working AI co-pilot in a short time.

What we learned

We learned how to integrate and adapt modern LLM platforms like Qwen into a real product, from prompt design to system orchestration, turning raw model capability into a usable experience.

What's next for CarMatch

Evolve CarMatch from a reactive assistant into one that can deeply understand user needs and proactively guide them with more personalized, context-aware recommendations.