Show HN: Polynya – Postgres를 AI 작업 공간으로 전환
hackernews
|
|
📰 뉴스
#claude
#오픈소스
원문 출처: hackernews · Genesis Park에서 요약 및 분석
요약
Polynya는 30초마다 Postgres 데이터를 Iceberg로 스트리밍하여 AI가 바로 사용 가능한 데이터 워크스페이스로 변환합니다. AI 에이전트는 생산 DB가 아닌 온디맨드로 생성되는 일시적인 ClickHouse 인스턴스를 통해 24/7 데이터 웨어하우스처럼 데이터에 접근할 수 있습니다. 세션 간에 유지되는 뷰 컬렉션을 구축하여 워크스페이스를 생성하고 관리할 수 있습니다.
본문
# Polynya > Turn your Postgres into AI-ready data workspaces. Polynya streams your Postgres to Iceberg every 30 seconds. Your AI agents get their own ephemeral ClickHouse instance on demand — not your production database. Build workspaces (collections of views) that persist across sessions. From your agent's point of view, it's a 24/7 data warehouse. ## How it works 1. Connect your Postgres: `npx polynya create --all` 2. Build workspaces: create collections of views your agents can query 3. Let your agents query: connect Polynya as an MCP server ## Quickstart ``` npx polynya create --all ``` ## MCP Connect Polynya as an MCP server so your AI agent can query your data: ``` claude mcp add --transport http polynya https://api.polynya.dev/mcp ``` ## Features - **Always fresh**: Data syncs every 30 seconds. No Kafka, no Debezium, no Airflow. - **Workspaces that compound**: Create workspaces — collections of views that persist across sessions. Analytics, fraud, experimentation. Your agent builds up an analytical layer over time. - **Your agent's own database**: An ephemeral ClickHouse spins up when your agent queries and shuts down when it's done. Your production database never sees the query. ## Use cases - **Product analytics**: Revenue breakdown, user metrics, retention analysis - **Fraud detection**: Surface suspicious patterns, save fraud rules as persistent views - **Performance monitoring**: Track latency, match times, SLOs with visualizations - **Customer data platform**: Unify user profiles, segment audiences, track lifecycle events - **Experimentation**: Analyze A/B tests, compare variants, measure uplift ## Pricing Free during early access. No credit card required. No egress fees. No per-row charges. - **Free**: 1 project, 5 tables, 1 workspace (10 views), ephemeral ClickHouse, 30s sync - **Pro $39/mo**: 3 projects, 25 tables, unlimited workspaces & views (coming soon) - **Team $99/mo**: 10 projects, unlimited tables, 10 team members (coming soon) ## Links - Website: https://polynya.dev - MCP endpoint: https://api.polynya.dev/mcp - CLI: https://www.npmjs.com/package/polynya - Built on pg2iceberg: https://pg2iceberg.dev
Genesis Park 편집팀이 AI를 활용하여 작성한 분석입니다. 원문은 출처 링크를 통해 확인할 수 있습니다.
공유