Show HN: Omni – Open-Source Workplace Search and Chat, Built on Postgres

Why It Matters

Enterprises gain a secure, cost‑effective way to embed AI search without exposing data to third‑party services, streamlining knowledge access and productivity.

Key Takeaways

• Open-source AI search runs entirely on Postgres
• Supports BM25 and pgvector semantic queries
• Self‑hosted with permission inheritance across integrated apps
• Connects to Google, Slack, Confluence, Jira, more
• Executes sandboxed code via Docker, supports any LLM

Pulse Analysis

The rise of generative AI has spurred demand for workplace assistants that can retrieve and synthesize information from disparate sources. Traditional stacks often combine Elasticsearch for keyword search, a separate vector database for embeddings, and a proprietary AI layer, creating operational complexity and security concerns. Omni tackles this by consolidating both BM25 and vector search within Postgres using ParadeDB, offering a single point of maintenance, backup, and scaling while preserving the rich query capabilities needed for modern enterprise knowledge bases.

From an engineering perspective, Omni’s architecture leverages Rust for high‑performance indexing and search services, Python for LLM orchestration, and SvelteKit for a responsive web UI. Each connector—whether to Google Drive, Slack, or Jira—runs in an isolated container, allowing language‑specific dependencies without cross‑contamination. The AI agent’s sandbox employs Docker network isolation, Landlock filesystem restrictions, and resource limits, ensuring that code execution cannot reach internal services or the internet, a critical safeguard for enterprises handling sensitive data.

Business leaders see immediate value in Omni’s self‑hosted model and permission inheritance, which respects existing access controls across integrated platforms. The ability to bring any LLM—OpenAI, Anthropic, Gemini, or open‑weight models—means organizations can align costs and compliance with their AI strategy. With straightforward deployment options ranging from a single‑server Docker Compose to Terraform‑managed multi‑region clusters, Omni lowers the barrier for companies to adopt AI‑driven knowledge work without sacrificing data sovereignty, positioning it as a compelling alternative to cloud‑only solutions.