Examples and tutorials for building AI applications with watsonx.ai Flows Engine
https://github.com/IBM/wxflowsStop reinventing the wheel every time you build an AI agent. IBM's wxflows repository gives you battle-tested examples, tools, and integrations that work across LangGraph, LangChain, OpenAI, and watsonx.ai - all in one place.
You've probably built a few AI agents by now. You know the drill: spend hours wiring up tool integrations, debugging framework quirks, and figuring out deployment patterns. wxflows cuts through that repetitive work with production-ready examples that actually work.
This isn't theoretical code - it's practical implementations from IBM's AI team, covering real-world scenarios like chat applications, text-to-SQL agents, and content processing pipelines.
Ready-to-Use Tools Five essential integrations already built and tested:
Framework Integration Examples Working implementations across the major AI frameworks:
Complete Application Examples Full-stack implementations you can deploy:
Here's what sets wxflows apart from typical example repositories - the code handles the messy details you encounter in production.
Tool Calling Architecture
# Clean abstraction that works across frameworks
from wxflows.tools import WeatherTool, WikipediaSearch
# Automatically handles API rate limits, error recovery, and response formatting
weather = WeatherTool()
wiki = WikipediaSearch()
# Works identically in LangGraph, LangChain, or direct OpenAI calls
agent.add_tools([weather, wiki])
Multi-Model Support Switch between Granite, Llama-3, GPT-4, and Mistral models without changing your agent logic. The examples show you how to handle the quirks of each model's tool calling format.
GraphQL Gateway Deploy your agents behind a unified GraphQL interface. Each flow becomes an endpoint you can call from any client application - web, mobile, or other services.
The repository includes Model Context Protocol examples that demonstrate how to expose your agents through MCP-compliant interfaces. This matters if you're building agents that need to work within MCP ecosystems or want standardized agent communication.
The examples address real deployment challenges:
git clone https://github.com/IBM/wxflows.git
cd wxflows
# Set up your environment
export WX_API_KEY="your-watsonx-key"
export WX_PROJECT_ID="your-project-id"
export OPENAI_API_KEY="your-openai-key"
# Run the chat app example
cd examples/chat-app
python run.py
The examples work immediately - no complex setup or infrastructure requirements.
You'll find wxflows valuable if you're:
The 92 GitHub stars and active development from IBM's team indicate this is production-grade code being used in real applications.
Links: