About this project

Olla Nest is a self-hosted company AI workspace designed to provide a secure, admin-controlled layer over Ollama and various cloud AI providers. Its core feature is an Auto Router that automatically directs user requests to the most appropriate model based on capability, speed, quality, and privacy requirements. It includes a privacy gate that blocks sensitive data (like SSNs or API keys) from being sent to external cloud providers. Key capabilities include: - **Enterprise Management**: Role-Based Access Control (RBAC), team hierarchies, per-user token limits, and SOC 2 compliant audit trails. - **Data Integration**: RAG (Retrieval-Augmented Generation) support with 20+ connectors for services like GitHub, Slack, Notion, Jira, and Google Drive, featuring hourly background synchronization. - **Developer Tools**: A secure code sandbox supporting Python, JS, Ruby, Java, and Bash; an interactive terminal; and eight specialized chat modes (e.g., build, fix, debug). - **Multimodal Features**: Local speech-to-text via faster-whisper, text-to-speech via OpenAI TTS, and image generation through DALL-E 3 or Stable Diffusion. - **Advanced Research**: A 'Deep Research' pipeline that decomposes complex queries into sub-questions, performs web and RAG searches, and synthesizes a cited report. - **Authentication**: Support for Google OAuth 2.0, generic OIDC, and SAML 2.0, alongside local authentication. - **Infrastructure**: Built with Java 26 and Spring Boot, using SQLite for storage and integrating with Grafana and Loki for structured logging.