Xyne is an open-source, self-hostable AI-first search and answer engine designed to unify fragmented work information across various SaaS applications. It targets individuals and organizations seeking a private, secure, and model-agnostic alternative to commercial solutions like Glean, Gemini, and Microsoft Copilot, enabling users to find information and get answers with cited sources.

How It Works

Xyne securely indexes data from connected applications, building a relationship graph to provide a unified search experience. It emphasizes providing context to AI models securely and privately, ensuring data is not used for training and respecting existing user permissions. Its model-agnostic design allows integration with any LLM, whether cloud-hosted or local.

Quick Start & Requirements

• Install/Run: docker-compose up (after configuration).
• Prerequisites: Google Workspace Service Account integration for initial functionality. Docker and docker-compose are required.
• Resources: Local deployment is possible; cloud deployment details are available in documentation.
• Docs: https://docs.xyne.tech/

Highlighted Details

• AI-first search and answer engine for work.
• Self-hostable and deployable anywhere (laptop, on-prem, cloud).
• Model-agnostic, supporting various LLMs and cloud providers.
• Permissions-aware, enforcing existing access controls.

Maintenance & Community

The project is actively developed with a focus on building an extensible Work AI Platform. Contribution guides are available for interested parties.

Licensing & Compatibility

The README does not explicitly state the license. Further clarification is needed regarding commercial use and closed-source linking.

Limitations & Caveats

Initial integrations focus on Google Workspace, with Slack and Atlassian planned. The project appears to be in active development, and comprehensive documentation for all features and deployment scenarios may still be evolving.