Nubase launches an open-source backend platform designed to let AI coding agents directly deploy full applications—database, auth, storage, functions, and memory—without separate infrastructure work.

What happened

Nubase released an open-source, self-hostable backend service with eight capability modules (Database, Auth, Storage, Assets, Functions, AI Gateway, Memory, and cron). Coding agents like Claude Code can connect via MCP tools and directly provision databases, deploy edge functions, publish frontends to a CDN, and schedule jobs—all in a single platform.

Why it matters

AI-generated code today typically produces demos that require weeks of backend infrastructure setup before going live. Nubase embeds that infrastructure into one service, so agents can model data, deploy logic, and publish a live app in minutes without agents or developers needing separate hosting accounts or external services.

What to watch

The platform bundles PostgreSQL, Redis, and a Studio dashboard into a single Docker image for self-hosting, and supports both a hosted version and private deployment. Memory is built in as a first-class primitive (entity extraction, hybrid retrieval, history), rather than bolted on as a separate vector-store service.