TurboSeek is an open-source AI search engine designed to provide users with AI-generated answers to their queries, inspired by Perplexity. It targets users seeking an alternative to traditional search engines, offering a conversational and context-rich search experience.

How It Works

The engine takes a user's question, queries the Bing Search API for the top six results, and scrapes text from these links to create context. This context, along with the original question, is then sent to Llama 3.1 70B for a streamed answer. A separate request to Llama 3.1 8B generates three related follow-up questions.

Quick Start & Requirements

• Install dependencies and run locally with npm install and npm run dev.
• Requires API keys for Together AI (LLM inference), Bing Search API (via SERP API or Azure), and Helicone (observability).
• A .env file is needed, with .example.env provided as a template.

Highlighted Details

• Utilizes Next.js App Router and Tailwind CSS for the frontend.
• Leverages Llama 3.1 8B and 70B models via Together AI.
• Integrates Bing Search API (via Serper or Azure) and Helicone for observability.
• Future tasks include improved streaming, token management, citation accuracy, caching with Redis, and advanced RAG techniques.

Maintenance & Community

The project is maintained by Nutlope. Further community engagement details (Discord/Slack, roadmap) are not explicitly provided in the README.

Licensing & Compatibility

The project's license is not specified in the README, which may impact commercial use or integration into closed-source projects.

Limitations & Caveats

The project is in active development with several future tasks outlined, indicating potential instability or missing features. Key functionalities like accurate source citation and caching are yet to be implemented.