展示 HN：Omni – 基于 Postgres 的开源工作场所搜索和聊天。

展示 HN：Omni – 基于 Postgres 的开源工作场所搜索和聊天。
Show HN: Omni – Open-source workplace search and chat, built on Postgres

## Omni：人工智能驱动的工作场所助手 Omni 是一款自托管的人工智能助手和搜索平台，旨在提高工作场所的生产力。它连接到流行的应用程序，如 Google Drive、Slack、Confluence 和 Jira，提供**统一搜索**，利用全文搜索和语义搜索功能。除了搜索，Omni 还具有**AI 代理**，可以与用户聊天、访问连接的应用程序、分析文档，甚至在安全、沙盒环境中执行代码。一个关键优势是**数据隐私**——Omni 完全在您的基础设施上运行，尊重现有的权限结构。在技术上，Omni 利用 Postgres 来存储所有数据和进行搜索，避免了多数据库的复杂性。它使用 Rust、Python 和 SvelteKit 构建，并提供通过 Docker Compose 或 Terraform 进行灵活部署的选项。用户还可以**自带 LLM**（Anthropic、OpenAI、Gemini 或开源模型）。 Omni 是开源的（Apache 2.0 许可），并欢迎贡献。

## Omni：开源工作场所搜索与聊天一个名为 Omni 的新开源项目，为 Glean 等工作场所搜索和聊天平台提供了一种自托管的替代方案。Omni 完全基于 Postgres（使用 ParadeDB 和 pgvector 构建），旨在为小型到中型团队提供更经济实惠且更具扩展性的解决方案。 Omni 连接到 Google Workspace、Slack 和 Confluence 等常用应用程序，同步数据以创建传统的 BM25 索引和向量索引，用于混合搜索。它具有由用户提供的 LLM（OpenAI、Anthropic、Gemini）提供支持的聊天界面，并具有搜索索引数据的工具，超越了基本的 RAG。目前处于 Beta 阶段，开发者正在寻求反馈，特别是关于自托管体验以及对仅使用 Postgres 架构进行扩展的担忧。该项目在 GitHub 上以 Apache 2.0 许可协议提供：[https://github.com/getomnico/omni](https://github.com/getomnico/omni)。

原文

Omni is an AI Assistant and Search platform for the Workplace.

Connects to your workplace apps, helps employees find information and get work done.

Features • Architecture • Docs • Deploy • Contributing

Unified Search: Connect Google Drive/Gmail, Slack, Confluence, Jira, and more. Full-text (BM25) and semantic (pgvector) search across all of them.
AI Agent: Chat interface with tool use: searches your connected apps, reads documents, and executes Python/bash in a sandboxed container to analyze data.
Self-hosted: Runs entirely on your infrastructure. No data leaves your network.
Permission Inheritance: Respects source system permissions. Users only see data they're already authorized to access.
Bring Your Own LLM: Anthropic, OpenAI, Gemini, or open-weight models via vLLM.
Simple Deployment: Docker Compose for single-server setups, Terraform for production AWS/GCP deployments.

Omni uses Postgres (ParadeDB) for everything: BM25 full-text search, pgvector semantic search, and all application data. No Elasticsearch, no dedicated vector database. One database to tune, backup, and monitor.

Core services are written in Rust (searcher, indexer, connector-manager), Python (AI/LLM orchestration), and SvelteKit (web frontend). Each data source connector runs as its own lightweight container, allowing connectors to use different languages and dependencies without affecting each other.

The AI agent can execute code in a sandboxed container that runs on an isolated Docker network (no access to internal services or the internet), with Landlock filesystem restrictions, resource limits, and a read-only root filesystem.

See the full architecture documentation for more details.

Omni can be deployed entirely on your own infra. See our deployment guides:

Google Workspace: Drive, Gmail
Slack: Messages, files, public channels
Confluence: Pages, attachments, spaces
Jira: Issues and projects
Web: Public websites, documentation and help pages
Fireflies: Meeting transcripts
HubSpot: Contacts, companies, deals, tickets
Local Files: File system indexing

See CONTRIBUTING.md for development setup and guidelines.

Apache License 2.0. See LICENSE for details.

展示 HN：Omni – 基于 Postgres 的开源工作场所搜索和聊天。 Show HN: Omni – Open-source workplace search and chat, built on Postgres

展示 HN：Omni – 基于 Postgres 的开源工作场所搜索和聊天。
Show HN: Omni – Open-source workplace search and chat, built on Postgres