TanStack AI Alpha: Your AI, Your Way

The TanStack team is thrilled to announce the alpha release of TanStack AI, a framework-agnostic AI toolkit designed for developers seeking ultimate control over their technology stack.

The current AI development landscape often presents a significant challenge: vendor lock-in. Developers frequently find themselves tied to specific frameworks or cloud providers, limiting their architectural freedom. TanStack AI aims to reverse this trend.

TanStack AI adopts a fundamentally different approach, striving to be the "Switzerland of AI tooling." It offers an honest, open-source collection of libraries, supporting multiple languages, designed to integrate seamlessly with your existing stack rather than demanding a complete overhaul.

What's in the Alpha

Server Support Across Multiple Languages

TanStack AI launches with comprehensive server-side support for JavaScript/TypeScript, PHP, and Python. All three languages enable full agentic flows with integrated tools. (Note: Python and PHP packages are pending release to their respective package systems.)

Adapters for Leading AI Providers

TypeScript adapters are available for OpenAI, Anthropic, Gemini, and Ollama. Additionally, the TypeScript server library includes capabilities for summarization and embeddings.

An Open, Published Protocol

TanStack AI features a fully documented, open protocol outlining server-client communication. This allows developers to use any language or transport layer (HTTP, WebSockets, etc.). As long as your backend communicates via a connection adapter adhering to the protocol, our client libraries will seamlessly integrate.

Isomorphic Tool Support

Tools can be defined once using meta definitions, then provided with distinct server and client implementations. This innovative architecture ensures robust, end-to-end type safety across your entire application.

Versatile Client Libraries

Client libraries are currently available for Vanilla JS, React, and Solid, with support for Svelte and other frameworks planned for future releases.

Practical, Deployable Examples

Beyond documentation, TanStack AI provides tangible, working code examples:

• TanStack Start with React
• TanStack Start with Solid
• PHP with Slim utilizing a Vanilla JS client
• Laravel with React (coming soon)
• Python FastAPI with a Vanilla frontend
• A multi-user group chat application built on TanStack Start using Cap'n'Web RPC and WebSockets

Granular Per-Model Type Safety

Given the diverse options and modalities (text, audio, video, tools) offered by various AI providers and models, TanStack AI delivers comprehensive type safety for providerOptions on a per-model basis. This ensures your IDE accurately understands each model's capabilities, eliminating guesswork and preventing runtime surprises.

Isomorphic Devtools

Gain unprecedented insight into your LLM workflows with a full AI devtools panel. This panel provides visibility into both server-side and client-side operations, allowing you to debug AI applications with the same familiarity and precision you apply to other development tasks. It's built on the robust foundation of TanStack Devtools.

Coming Soon

Headless Chatbot UI Components

Anticipate headless chatbot UI components, akin to Radix, but tailored for AI chat interfaces. These fully functional, unstyled components for React and Solid will empower you to customize the look and feel to perfectly match your application's design, while TanStack AI manages the underlying complexity. This approach mirrors the flexibility offered by TanStack Table and TanStack Form.

The Alpha Stage: What to Expect

As TanStack AI is currently in its alpha phase, users should anticipate bugs, rough edges, and features that may not yet function as intended. We are committed to transparency and dedicated to simplifying AI application development. Your feedback, suggestions, and ideas are invaluable as we work towards building the most robust AI toolkit possible.

Furthermore, the TanStack teams are small and entirely volunteer-driven. We warmly welcome contributions, whether it's helping to build new adapters, enhancing Python or PHP support, or assisting in any other capacity.

Our Motivation: Empowering Developers

TanStack AI was created out of a belief that developers deserve superior tools. We are not offering a service or a proprietary platform; there is no migration path to a new ecosystem, nor will there ever be vendor lock-in. Instead, we provide genuine, open-source tooling from a trusted team that has consistently delivered framework-agnostic developer tools for years. Our commitment to supporting developers remains unwavering.