Discovery Street

Discovery Street is an AI-powered platform that helps users search, compare, finance, and locate Toyota vehicles in one place using natural language search and an intuitive, modern interface.

Comment

Inspiration

We were inspired by modern car-buying tools that blend AI search, comparison features, financial planning, and interactive maps.

What it does

Discovery Street is an AI-powered car discovery and financing assistant. Users can:

  • Search Toyota vehicles using natural language
  • Compare models and trims side by side
  • Calculate financing and lease payments
  • Get pre-qualified based on financial inputs and print a prequalification to take to a dealer
  • Find nearby dealerships on an interactive map

How we built it

  • Frontend: React, TypeScript, Vite, Tailwind CSS, Zustand, React Router
  • Backend: Flask (Python), Flask-CORS, APScheduler
  • AI Search: Google Gemini AI and embeddings APIs for semantic ranking and summaries
  • Maps: Mapbox GL JS with react-map-gl for dealer visualization
  • UI: Dark theme with Toyota-red accents and a custom spotlight border effect using CSS gradients and styling
  • Finance: Loan amortization, lease calculations, and a pre-qualification evaluator with PDF export

Challenges we ran into

  • Mapbox tokens not loading due to missing VITE_ prefixes and whitespace; added .trim() and fallbacks
  • Inconsistent image rendering until we created a unified ModelImage component and standardized slugs
  • Spotlight (GlowCard) effect conflicting with comparison tables; adjusted sizing and overflow
  • Financial calculation discrepancies; corrected amortization and lease formulas and unified rounding
  • Storing and caching states

Accomplishments that we're proud of

  • A working semantic search system powered by Gemini embeddings
  • A polished, responsive UI with advanced CSS effects and strong branding
  • Accurate loan/lease estimates and pre-qualification with PDF generation
  • Integrated maps, geolocation, dealer data, and distance calculations into a cohesive feature set

What we learned

  • Implementing semantic search with embeddings, cosine similarity, and threshold tuning
  • Building interactive UI with CSS masks, gradients, and custom properties
  • Financial modeling for loans and leases and clear user-facing calculators
  • Designing APIs, handling environment variables, and integrating external services

What's next for Discovery Street

  • Expand beyond Toyota and integrate real dealership inventory
  • Add user accounts, saved profiles, and personalization
  • Introduce trade-in estimators, credit simulations, and deeper AI-driven recommendations

What makes us startup ready

  • B2B business model - relying on car brands for profit rather than consumers
  • GTM Strategy: Begin with Toyota, expand wider to car brands by bringing them customers and serving as a marketing platform
  • First centralized car brand marketplace offering easy comparisons and results to car manufacturer's customers rather than resellers/dealerships.
  • Can be recolorized to fit branding
  • Future Feature: manufacturer portal for analytics and dashboarding

Built With

Share this project:

Updates