BaseAI.dev was introduced by Langbase Inc. as the first Web AI Framework, targeting JavaScript and TypeScript developers who wanted to build serverless, autonomous AI agents with integrated memory (RAG) and external tools. It provided a local-first developer experience, letting engineers prototype AI pipelines entirely on their machine before deploying to the cloud. The framework's core abstractions were Pipes (agentic workflows that chain LLM calls with tools and memory), Tools (self-healing function integrations), and Memory (vector-based RAG storage). Developers could deploy everything serverless with a single command (`npx baseai deploy`), pushing agents to Langbase's serverless cloud platform. BaseAI was fully open-source and free to use, hosted under the LangbaseInc GitHub organization at github.com/LangbaseInc/BaseAI, with npm packages published as `@baseai/core`. The docs and learning guides were available at baseai.dev/docs and baseai.dev/learn. The project has since been archived by Langbase. The team concluded that opinionated frameworks become blockers in a fast-moving field like AI engineering, and shifted their strategy toward composable AI Primitives (memory, pipes, agents) exposed as language-agnostic APIs with TypeScript and Python SDKs. Users are now directed to use Langbase's primitives directly rather than the BaseAI framework layer. For teams that evaluated BaseAI, the successor path is Langbase's standalone AI primitives platform at langbase.com, which provides the same building blocks without the framework abstraction. Key features: - Local-first development: build and test AI agents entirely offline before deploying - Agentic Pipes: composable LLM workflows combining tools and memory - Integrated RAG Memory: vector-based memory (1536-dim embeddings) built into the framework - One-command serverless deploy to Langbase cloud (`npx baseai deploy`) - TypeScript-first API with consistent, well-documented endpoints - Open-source and free under the LangbaseInc GitHub organization - Multi-LLM support: works with various underlying language model providers - Self-healing tools: structured tool integrations with error recovery
Free and open-source (framework archived; successor Langbase platform has its own pricing)