One AI integration. Multiple providers.

One AI integration. Multiple providers. Two languages. Meet Xoin.

Building AI-powered applications often means dealing with multiple provider SDKs, different response formats, structured output validation and provider-specific implementations.

I wanted to make that experience simpler.

Introducing Xoin, an open-source, multi-provider LLM client available for both JavaScript/TypeScript and Python.

Instead of managing completely different integration patterns for every provider, Xoin gives developers a consistent interface to work with multiple LLM providers.

Two libraries. One developer experience.

01. Xoin for JavaScript / TypeScript

@xoin/xoin-js

Unified multi-provider LLM interface

Structured outputs with Zod validation

Text embeddings for semantic search and RAG

Reusable prompt templates

Automatic provider fallback

Multi-provider routing

02. Xoin for Python

xoin-py

Async-first architecture

Structured outputs with Pydantic

Multi-provider chat completions

Embeddings with supported providers

Automatic retries and provider fallback

Reusable prompt templates

Supported AI providers

OpenAI · Anthropic · Gemini · Mistral · DeepSeek · OpenAI-compatible APIs

Why does this matter?

Imagine building an AI application that uses OpenAI today but needs to switch to another provider tomorrow.

You shouldn't have to redesign your entire integration layer.

Xoin helps abstract provider differences, validate model outputs and make provider fallback easier to manage.

The goal is simple:

Less provider-specific glue code. More time building actual AI products.

Get started

JavaScript / TypeScript:

npm install @xoin/xoin-js zod

GitHub: https://github.com/kanha95/xoin-js

Python:

pip install xoin-py

GitHub: https://github.com/kanha95/xoin

Both projects are open source.

If you're building AI agents, RAG pipelines, LLM-powered SaaS products or backend AI services, I'd love to hear your thoughts.

One question for developers:

When building multi-LLM applications, do you prefer a unified abstraction layer or direct integration with each provider?

Let's discuss.

#OpenSource #ArtificialIntelligence #LLM #Python #JavaScript #TypeScript #AIEngineering #RAG #DeveloperTools #BackendDevelopment