Project Description
Audora Music is an e-commerce platform for professional musical instruments and audio gear with over 10,000 products across 186 brands. The technical differentiator is the AI layer: a conversational assistant that recommends products based on the user’s level and budget, and a semantic search engine that understands natural language queries instead of relying on exact keyword matching.
Key Features
- AI Assistant: Conversational chat that recommends instruments and gear considering user preferences, experience level, and budget
- Semantic search: Embeddings-powered search engine that understands the intent behind a query, not just keywords
- 10,000+ product catalog: 303 categories and 186 brands with dynamic filters, sorting, and pagination
- Curated selection: Weekly featured product recommendation algorithm
- Quick search: Command palette with ⌘K for instant catalog access
Technologies Used
- Next.js 15 + App Router: SSR, API Routes, and Server Components for optimal performance
- Vercel AI SDK: Response streaming from the assistant with multi-model support
- OpenAI GPT: LLM powering the conversational assistant with product context
- Embeddings (text-embedding-ada): Catalog vectorization for similarity-based semantic search
- TypeScript: Strict typing across the entire stack
AI Architecture
The technical core of the project consists of two complementary systems:
- Chat with product context: The assistant receives relevant catalog items as context for each query, generating responses grounded in real inventory rather than generic knowledge
- Embeddings-based search: Every product in the catalog is vectorized. When a user searches for “something to record podcasts at home on a tight budget,” the system finds semantically similar products even if their descriptions don’t contain those exact words