butterbase-ai/butterbase

What it solves

Butterbase is an AI-native, open-source backend-as-a-service (BaaS) designed to eliminate the "glue code" typically required to build AI-driven applications. It provides a unified set of building blocks—including database, authentication, and AI orchestration—so developers can deploy AI apps without vendor lock-in.

How it works

Butterbase combines a Postgres-backed data plane with a suite of AI-specific tools. It uses a three-plane Postgres architecture (control, runtime, and data) to isolate user data while providing automatic REST endpoints and row-level security (RLS).

Key components include:

• AI Gateway & RAG: A single endpoint for chat, embeddings, and managed document collections for semantic search.
• Agent Runtime: A Python-based executor for declarative multi-step LLM workflows (graph specs) that support human-approval pauses and function tools.
• MCP Server: A built-in Model Context Protocol server that exposes the entire backend (storage, KV, functions) as tools that AI agents can call directly.
• Compute: Serverless TypeScript functions running on Deno and stateful "Durable Objects" for long-running agents or real-time collaboration.

Who it’s for

Developers building AI agents, RAG-based applications, and AI-native SaaS products who want a scalable, self-hostable backend that integrates deeply with LLM orchestration.

Highlights

• MCP Integration: Every backend capability is exposed as an MCP tool, allowing agents to operate the backend directly.
• Durable Objects: Stateful per-key actors for managing complex agent states or chat rooms.
• AI-Native Tooling: Built-in RAG, an AI gateway, and a dedicated agent runtime for graph-based workflows.
• Complete Templates: Includes production-ready apps like an AI CRM and an AI support agent to demonstrate deep product integration.