Inspiration
Buying a used car is a minefield of scams, hidden accidents, and overpriced listings. We wanted to level the playing field for everyday buyers by giving them the same "insider data" that professional dealers use, powered by AI.
What it does
CarDealScout takes any car listing URL and instantly generates a "Deal Score." It extracts the VIN to pull official factory specs and safety recalls from the NHTSA, searches for public accident history, and even writes a custom negotiation script to help you lower the price.
How we built it
Frontend:# React 18, Tailwind CSS, and Framer Motion for a "premium tool" feel.
AI:##Gemini 2.5 Flash for deep URL analysis and VIN extraction.
Backend:## Node.js/Express with MongoDB to persist and track deal history.
Integrations:## NHTSA API for vehicle decoding and Google Search for surfacing public damage records.
Challenges we ran into
Managing API rate limits while performing deep URL analysis was tough. We had to implement automatic retry logic and optimize our AI prompts to extract structured JSON data reliably from messy, unstructured Craigslist-style text.
Accomplishments that we're proud of
We successfully integrated real-world safety data (NHTSA) with AI reasoning. Seeing the app automatically flag a "Salvage Title" or a "Safety Recall" from just a raw URL feels like magic.
What we learned
We learned how to orchestrate multiple asynchronous data sources AI, government databases, and search results into a single, cohesive user experience that provides immediate value.
What's next for CarDealScout
We plan to add Visual Damage Detection using Gemini Vision to spot mismatched paint or frame damage in photos, and a Total Cost of Ownership calculator to forecast maintenance and insurance costs.
Log in or sign up for Devpost to join the conversation.