Download
Search with Lepton is an open source demonstration project that shows how to build a conversational search engine using the Lepton AI framework. It combines traditional web search with large language models to provide natural language answers to user queries. It retrieves information from supported search engines and uses that context to generate responses through a retrieval-augmented generation approach. The implementation is intentionally minimal, containing fewer than 500 lines of code while still providing a complete working example of an AI-powered search system. It includes both a backend service written in Python and a web interface that allows users to interact with the search engine in a conversational format. Developers can configure different search providers and language models through environment variables, making it flexible for experimentation and prototyping.
Features
- Conversational search interface powered by large language models
- Integration with multiple search backends such as Bing and Google APIs
- Retrieval-augmented generation workflow for contextual answers
- Shareable and cached search results to improve performance
- Customizable web user interface for interacting with queries
- Generates suggested related questions to continue exploration
Project Samples
