Libraries

Use these (open source) libraries to build AI applications on watsonx.ai.

LangChain is a framework that enables the development of applications that are powered by language models and provides tools to integrate data sources, model configurations, and workflows. The framework is available for both Python and JavaScript, and is compatible with watsonx.ai API endpoints.

Follow these tutorials to learn how to use LangChain with watsonx:

• Create a LangChain AI Agent in Python using watsonx
• Create a LangChain RAG system in Python with watsonx

Documentation:

• Python Docs
• JS Docs

LangGraph is a framework for building stateful, multi-actor applications by combining language model outputs with structured graph data to create agent and multi-agent workflows; these workflows are then represented with graphs.

Use the LangGraph template in the agent quickstart.

Follow these tutorials to learn how to use LangGraph with watsonx:

• Build a RAG agent to answer complex questions

LlamaIndex (formerly GPT Index) is a tool that helps structure, index, and query external data sources using language models, making it easier to build and interact with custom knowledge bases.

Use the LlamaIndex template in the agent quickstart.

Follow these tutorials to learn how to use LlamaIndex with watsonx:

• Use watsonx.ai with LlamaIndex to build RAG applications
• Use watsonx and LlamaIndex for Text-to-SQL task

Documentation:

• LlamaIndex Docs

CrewAI is a multi-agent framework that enables autonomous AI agents to collaborate effectively in completing complex tasks through role-based design and coordinated workflows.

Use the CrewAI template in the agent quickstart.

Follow these tutorials to learn how to use CrewAI with watsonx:

• Multiagent collaboration for customer call analysis using watsonx.ai and CrewAI
• Leveraging CrewAI and IBM watsonx

Documentation:

• CrewAI Docs

AutoGen is an open-source framework by Microsoft for building collaborative AI agents. It enables multi-agent workflows with event-driven messaging and flexible orchestration.

Use the AutoGen template in the agent quickstart.

Follow these tutorials to learn how to use AutoGen with watsonx:

• AutoGen and watsonx.ai

Documentation:

• AutoGen Docs

BeeAI is an open-source multi-agent framework by IBM that enables collaborative AI agents, called “bees,” to communicate and coordinate tasks across different languages via a standardized Agent Communication Protocol (ACP), with support for both TypeScript and Python.

Use the BeeAI template in the agent quickstart.

Follow these tutorials to learn how to use BeeAI with watsonx:

• Set up and run the BeeAI Agent Framework

Documentation:

• BeeAI Docs

The IBM watsonx.ai Python SDK enables developers to train, deploy, and integrate AI models with their applications using various APIs. The SDK supports tasks, such as model scoring, and integrates with IBM watsonx.ai services.

Documentation:

• watsonx.ai Python SDK Docs
• watsonx.ai Python SDK GitHub

The IBM watsonx.ai Node.js SDK enables developers to create and deploy generative AI applications by providing programmatic access to the IBM watsonx.ai service with various API calls and authentication options.

Documentation:

• watsonx.ai Python Node.js Docs
• watsonx.ai Python Node.js SDK GitHub