ItzCrazyKns/Vane

Vane is an AI-powered answering engine.

What it solves

Vane is a privacy-focused AI answering engine that allows users to search the internet and get accurate, cited answers without relying on centralized search engines. It solves the problem of privacy loss in traditional search and the lack of transparency in AI-generated answers by running on the user's own hardware and providing clear source citations.

How it works

It functions as a search interface that integrates with SearxNG for private web search and connects to various AI providers. Users can choose between local LLMs (via Ollama) or cloud-based models (such as OpenAI, Claude, and Gemini). The system processes web results and documents uploaded by the user to generate responses, offering different search modes (Speed, Balanced, and Quality) to tune the depth of research.

Who it’s for

It is designed for privacy-conscious users, researchers, and developers who want a self-hosted alternative to AI search engines, as well as those who prefer using local LLMs for their queries.

Highlights

  • Multi-Provider Support: Compatible with local LLMs via Ollama and major cloud providers like OpenAI, Anthropic, and Google.
  • Private Web Search: Powered by SearxNG to keep user identities private during searches.
  • Flexible Search Modes: Offers Speed, Balanced, and Quality modes for different research needs.
  • Document Analysis: Supports uploading PDFs, text files, and images to ask questions about specific documents.
  • Customizable Sources: Allows users to search the web, academic papers, or specific domains.
  • Integrated Widgets: Displays relevant UI cards for weather, stock prices, and calculations.

Related

  • Project
    Fosowl/agenticSeek

    A private, local AI assistant that autonomously browses the web, writes code, and plans complex tasks while keeping all data on the user's device.

  • Project
    xynehq/xyne

    Xyne is an open-source AI-first search and answer engine for work that indexes fragmented SaaS data to provide a centralized, context-aware search experience.

  • Project
    codota/tabnine-vscode

    An AI code assistant providing context-aware completions and chat to accelerate software development while ensuring data privacy and security.

  • Project
    felladrin/MiniSearch

    A self-hosted, private search engine with an AI assistant that can run entirely in the browser, providing cited answers based on aggregated web results.

  • Project
    StarTrail-org/LEANN

    A lightweight vector database that uses graph-based selective recomputation to reduce storage by 97%, enabling local RAG on millions of personal documents.