Wizi AI Code Search is a narrowly-scoped open-source tool for React and TypeScript codebases: instead of keyword search, it lets developers describe what they're looking for in plain language and uses semantic search to find the relevant code. It's built as a Next.js application, deployed through Vercel, using OpenAI for the natural-language processing and Pinecone for storing the embeddings that make semantic matching possible.
It's MIT licensed, but the project explicitly labels itself a prototype rather than something ready for production use, meaning teams adopting it should expect to treat it as a starting template to adapt rather than a polished, supported product. Self-hosting requires setting up accounts and configuration across GitHub, OpenAI, and Pinecone, which is a meaningfully higher setup cost than a hosted SaaS tool.
It's a useful reference for front-end teams curious about building their own semantic code search rather than a tool most teams would adopt as-is for daily use.